Electric Circuit Analysis PDF
Electric Circuit Analysis PDF
Electric Circuit Analysis PDF
---..-
Library of Congress Cataloging-in-Publication Data
Electric circuit analysis 1 David E. Johnson ... let al]. - 3rd ed.
p. cm.
Rev. ed. of: Electric circuit analysis 1 David E. Johnson, Johnny
R. Johnson. John L. Hilburn. 2nd ed. c1992.
Includes bibliographical references and index.
ISBN 0-13-252479-1
I. Electric circuit analysis. I. Johnson, David E. II. Johnson,
David E. Electric circuit analysis.
TK454.E57 1997
621.319'2-dc20 96-27572
CIP
\
MicroSim, PSpice, PLSyn, and Polaris are registered trademarks of
MicroSim Corporation. All other trademarks are the property of their
respective owners.
.IX
ACQUISITIONS EDITOR: Eric Svendsen
...........................................................................
PREFACE
;I
=
1997, 1992, 1989 by Prentice-Hall, Inc.
Simon & Schuster/A Viacom Company
1.5 Circuit Analysis and Design, 15
Summary, 17
- Upper Saddle River, New Jersey 07458
Problems, 17
The author and publisher of this book have used their best efforts in preparing this book.
These efforts include the development, research, and testing of the theories and programs to ........................................................................
RESISTIVE CIRCUITS 21.
determine their effectiveness. The author and publisher make no warranty of any kind, 2.1 Kirchhoff's Laws, 22
expressed or implied, with regard to these programs or the documentation contained in this 2.2 Ohm's Law, 30
book. The author and publisher shall not be liable in any event for incidental or
consequential damages in connection with, or arising out of, the furnishing, performance, or
2.3 Equivalent Subcircuits, 33
use of these programs. 2.4 Series Equivalents and Voltage Division, 37
2.5 Parallel Equivalents and Current Division, 42
All rights reserved. No part of this book may be 2.6 Thevenin and Norton Equivalents, 48
reproduced, in any form or by any means, 2.7 Practical Sources and Resistors, 54
without permission in writing from the publisher. 2.8 Ammeters, Voltmeters, and Ohmmeters, 61
2.9 Design of Resistive Circuits, 64
Printed in the United States of America Summary, 70
10 9 8 7 6 5 4 3 2
Problems, 71
ISBN 0-13-252479-1
IJ ........................................................................
DEPENOENi'sOURCES AND OP AMPS 79.
Prentice-Hall International (UK) Limited, London
Prentice-Hall of Australia Pty. Limited, Sydney 8 3.1 Definitions, 80
Prentice-Hall Canada Inc., Toronto a 3.2 Circuits with Dependent Sources, 82
Prentice-Hall Hispanoamericana, S.A., Mexico 3.3 Operational Amplifiers, 83
Prentice-Hall of India Private Limited, New Delhi 3.4 Role of Negative Feedback, 89
Prentice-Hall of Japan, Inc., Tokyo
Simon & Schuster Asia Pte. Ltd., Singapore
Editora Prentice-Hall do Brasil, Ltda., Rio de Janeiro iii
3.5 Op Amp Building Block Circuits, 92 6.5 Superposition in First-Order Circuits, 229
3.6 Interconnecting Op Amp Building Blocks, 97 6.6 Unit Step Function, 234
3.7 Practical Op Amps, 102 6.7 Step and Pulse Responses, 239
3.8 Design of Simple Op Amp Circuits, 104 6.8 SPICE and the Transient Response, 244
Summary, 110 6.9 Design of First-Order Circuits, 255
Problems, 111 Summary, 261
Problems, 262
r.
!!l!',,!
ANALYSIS METHODS 117
ill:\'!"':
4.1 Linearity and Proportionality, 119
................. . .........................................................................
SECONDORDER CIRCUITS
\ 7.1 Circuits with Two Storage Elements, 273
271.
iv Contents Contents v
!!~j[r,
.0 000000.00 IPo .. ~C;. ~!.~~oI?X-.~1~!~ .~~W~o~ ... 0 0 0 0 0 0 00 0 . 0 . ~~.1. ............... . . .. f~~~~~N~X .~~~.~~~~~ ..........................................~~.~
10.1 Average Power, 402 14.1 Frequency Response Function, 559
10.2 RMS Values, 409 14.2 The Decibel Scale, 563
10.3 Complex Power, 413 14.3 Bode Gain (Amplitude) Plots, 566
10.4 Superposition and Power, 420 14.4 Resonance, 577
10.5 Maximum Power Transfer, 425 .~ 14.5 Frequency Response of Op Amps, 584
10.6 Conservation of Power, 428 14.6 Filters, 588
10.7 Reactive Power and Power Factor, 432 14.7 Scaling, 595
10.8 SPICE and AC Steady-State Power, 435 14.8 SPICE and Frequency Response, 598
Summary, 437 Summary, 601
Problems, 438 Problems, 602
ktj411\ii, ,
.~1
qJ!'~J~51!"
THREE-PHASE CIRCUITS
445 ............... . ...........................................................................
MUTUAL INDUCTANCE AND TRANSFORMERS 609
"i\' ..........................................................................
FOURIER SERIES AND TRANSFORM 689
vi Contents
................. DESIGN OF LINEAR FIL1~~.~
.............................
,;' . .................................... .7.~.~
18.1 Passive Filters, 745
18.2 Active Filters, 752
18.3 Classical Filters, 763
18A Filter Transformation and Scaling, 776
Summary, 785
Problems, 786
Appendix
~ ~3
f. MATRIX METHODS
~.f~:~' ..... ................................. .
Al Matrix Fundamentals, 793 \
A2 Conversion to Vector Matrix Form, 794
A3 Determinants, 795
AA Cramer's Rule, 797 In 1978 the first edition of "Basic Electric Circuit Analysis" was published. It was well
A5 Matrix Inversion, 798 received by students and instructors, and after a decade of widespread adoption and
A6 Gauss Elimination, 799 four editions, a somewhat expanded version entitled "Electric Circuit Analysis" became
Appendix available in 1989. The book you hold is the third edition of "Electric Circuit Analysis."
It is somewhat unusual for an engineering text project to survive and prosper over a
1;:0 COMPLEX NUMBERS AND THE COMPLEX EXPONENTIAL
~Ji~: 80~ duration approaching two decades, and for the loyalty of those who use and have come
B.l Complex Numbers, 803 to rely on it we are grateful. We hope you will find its traditional virtues securely woven
B.2 Complex Exponential Function, 807 into the present edition, accompanied perhaps with a few enhancements in the bargain.
The continued acceptance of this book is, we believe, the consequence of two
Appendix essential ingredients: a sound initial design, and an ongoing commitment to upgrade and
~.. 1 adapt, to regularly reappraise both content and approach in light of actual experience. To
~ CIRCUIT TOPOLOGY
Vji:h;!~:: ................................... . ~~ .. do this effectively requires awareness of the needs of instructors and students, not in the
C.1 Network Graph, 811 abstract but "on the ground." The four authors listed on the spine of this book have among
C.2 Nodal Analysis, 812 us over one century of circuits teaching experience, and we remain in communication
C.3 Basic Loop Analysis and Mesh Analysis, 814 with the larger community of instructors and students beyond our campuses through
extensive regular surveys conducted by Prentice Hall. Changes both large and small can
Appendix be traced to comments made by many insightful colleagues and students, and, yes, even to
................. I~
complaints about specific examples or sections of narrative which could be written more
SPICE REFERENCE GUIDE 817.
i~ .......................................................................... clearly, and, we hope, as a consequence have been. To these individuals, too numerous
D.1 SPICE Input File, 818 to list but not to remember, we owe a great debt.
D.2 Title and Comment Statements, 819
D.3 Netlist Statements, 819
DA Solution Control Statements, 824 SCOPE AND PURPOSE
D.5 Output Control Statements, 828
This book is intended for a two semester or three quarter introductory course in linear
D.6 End Statements, 830
lumped circuit analysis for the sophomore post-secondary year. The first portion of
Appendix the book may also be used for a one semester or two quarter terminal course for non-
................. .lit... ~V~.~~~~~ .1R. ~~k~~T~I? g.l?R:~Y..~~~~P. .~~g.~H~~.........~~.1. majors. The development is calculus-based, as are the physical laws upon which the
methods of circuit analysis stand. The exposure to differential and integral calculus
routinely acquired by the engineering or physical science student in the freshman year is
adequate background. Other mathematical topics are raised in a limited and self-contained
... !~I?~~................................................................
839. manner, and are not assumed or required as prerequisite background. This includes the
algebraic manipulation of complex numbers, vector-matrix formulation and solution of
viii Contents
IX
linear systems of equations, singularity functions, and most particularly, the solution of New examples and exercises
linear time-invariant differential equations. The first two are the subjects of appendices Most examples and exercises are new. Previous editions had an unswerving com-
and numerous examples and exercises, the latter two are developed in the chapter in mitment to easy numerics, like C = 1 Farad. Recognizing their motivational value,
which they are needed. some more practical examples and exercises have been added to the mix. Sharp price
Care has been taken to fashion the selection and order of material to be of use to reduction for scientific calculators over the last decade makes them now universally
the electrical engineering baccalaureate student, but also to students of other engineering available. Their use lowers the computational pain level for exercises with answers like
and technical disciplines at the two and four year post-secondary level. The mathematical i 1 (t) = 0.544 cos(377t + 27.4) A to easily tolerable levels.
ideas and methods of electrical circuit analysis are carefully separated from physics-based
issues of electrical science, such as Maxwell's Laws, which are neither covered here nor
assumed as pre- or corequisites. Mathematical rigor has not been sacrificed where it is New problems
justified and can be built on. But in the best engineering spirit, where a conclusion is
Most end-oj-chapter problems are new. In response to requests from adopters, effort
obvious, it is declared to be just that, without further ado.
\ has been made to expand the range of difficulty, from drill to very challenging. The most
Sufficient space has been allocated for judicious repetition, reinforcement and learn-
challenging problems have been set off as a separate group. In addition, those problems
ing by example, but we have resisted expanding the length of the book to the extremes
intended to be solved with aid of a digital computer are set off as a second separate group
sometimes found in current practice. We have found that great length daunts many stu-
within the end-of-chapter problem sets containing them. Other problems not requiring
dents, and impedes the instructor crafting a one semester or two trimester course. It may
computer support but for which an electronic calculator would be of significant benefit
dictate a crazy-quilt approach, skipping sections and whole chapters in the interests of
are marked with a calculator icon. Finally, just as several new examples and exercises
time. This leaves lingering doubts whether some needed concept may have been missed,
have been added which emphasize practical values, for instance resistance in the kilohms
and certainly does not help develop a sense of continuity. The first nine chapters may
range and current in the milliampere range, many new end-of-chapter problems are of
be comfortably covered in a single semester exposure to AC and DC circuits, with more
ambitious instructors adding the AC power chapters 10 and 11, or an introduction to the this practical type as well.
use of Laplace transforms, chapters 12 and 13. It is our practice to cover all 18 chapters
in a two semester circuits course. New coverage
Circuit design. The payoff for studying circuit analysis is directing this skill towards
the design of novel and useful electric circuits. Few practitioners view circuit analysis,
COMPLETELY REWRITTEN THIRD EDITION the calculation of currents and voltages in a specified circuit, as a fully satisfying end in
The third edition incorporates considerably more changes than its predecessor. The itself. Engineers and applied scientists are characterized by an interest in building things,
principal ways in which it differs from the second edition are: a desire to blend creativity and imagination with mathematical analysis to produce things
that did not exist before, to create practical devices that satisfy design requirements and
specifications. The universal availability of the circuit simulators SPICE and MicroSim
Narrative PSpice make it cost-effective (and safe) to challenge even the first-year circuits student
The text narrative has been extensively rewritten. More than half the paragraphs with design problems. In our experience, starting from a blank sheet of paper and seeing
are.,ijew, and many others have been reworked. The goal has been to clarify the flow of one's own design emerge, ultimately demonstrating that it works, meets all design specs,
ideas and accommodate the changes in order and emphasis, while keeping the book to is one of the most gratifying and motivating elements of circuits work. To this end,
its relatively compact size. We do not believe that great length is a virtue in textbooks, sections entitled "Design of ... " have been added to five chapters, and a new chapter
either for the student or instructor. (Chapter 18) entirely devoted to design. This design content is self-contained, each
instructor may avail her- or himself of as much of this design exposure as judgement and
time availability dictate.
Organization
Some topics have been expanded and new topics added. Op amps have been pre-
Sequence and order have been moderately reworked. Chapters on Independence sented through a series of increasingly refined models, and the virtual short circuit princi-
of Equations and Network Theorems have disappeared, the first to an appendix and the ple justified in context of negative feedback circuits. Feedback, loading, and the concept
~econd scattered among several chapters in which the theorems arise more naturally of modular "building blocks," all essential to understanding modem linear op amp cir-
III context. Laplace transforms are introduced somewhat earlier, after phasors and AC cuits, are introduced. The fundamental requirement of stability, sometimes overlooked
circuits, but in time to be used to simplify development of transfer functions and input- in introductory circuits courses, and its link to frequency response and AC steady state is
output analysis without resort to the contrivance of complex frequency. Fourier series addressed. Design of active circuits using SPICE, an activity students find exciting and
and transform are treated in a unified single chapter. satisfying and which can stimulate interest in the more routine analysis tasks, is given
x Preface Preface xi
space. Bode gain plots are developed, and their use related to the engineering design the type which are needed for nodal and mesh analysis. Appendix D is a reference guide
cycle for active filters and other active circuits exploited. for the use of SPICE, and Appendix E lists answers to selected odd-numbered problems.
A SPICE command reference appendix is included. The public domain computer-
~ided circuit analysis an~ simulation software program SPICE has been covered tutorially USE OF SPICE
m the body of the book, m sections entitled "SPICE and ... " at the end of many chapters.
In addition, a concise listing of each of the SPICE commands used in this text, their rules SPICE has become the de facto standard for computer-aided circuit simulation and anal-
and formats, is included as an appendix. While tutorial discussions and step by step ysis in the classroom. Its use as a limited but helpful tool for the study of electrical
examples are useful to learn SPICE efficiently, it is convenient not to have to thumb circuits is warmly embraced in this book. There are two domains in which we have
through many pages before reminding oneself how a resistor is entered into a SPICE found SPICE particularly valuable, in gaining experience with circuits too big to justify
input file, for instance. the often catastrophic level of computational effort required to extract just a few simple
responses, and in design. In the first case, a five-mesh circuit is essentially as simple as
a one- or two-mesh circuit, and the experience with multiple resonances, parasitic oscil-
DESIGN AND ORGANIZATION OF THIS BOOK lations and other behaviors of complex circuits can be extremely interesting. In design,
SPICE allows students to try their own ideas on how real circuits solving real problems
The book contains eighteen chapters and five appendices. These logically divide into can be configured, and in our experience this often forms the high point of the course
~ime domain, ~hasor and transform domain sets of chapters. The first seven chapters from the student's viewpoint. It is not always easy to sustain motivation at the sopho-
mtroduce pass1ve (RLC) and active (source, op amp) circuit elements, and the time more level, when basic science and mathematics are center stage and students interested
domain analysis of AC and DC circuits which are formed when these elements are in building things and exercising individual creativity are beginning to wonder if they
interconnected. Chapters 8-11 cover phasor analysis of ac steady state circuits, including are in the right place. They are indeed in the right place, design is what engineering is
complex power and three-phase circuits. The final set of chapters, Chapters 12-18, all about, and SPICE is quite useful in making this point when it may be most needed.
develops the transform analysis methods of Laplace and Fourier, and applies them in Because not all instructors agree concerning the best use of SPICE within intro-
framing various input-output descriptions and design methods. ductory circuits courses, care has been taken to segregate SPICE-related material into
Chapters 10 and 11, on AC steady state power and three phase circuits, may be final sections recognizable by their titles "SPICE and ... " in 8 separate chapters. These
skipped without disservice to the following material. Likewise, the final chapter, Chapter sections form a bare-bones but self-contained tutorial on the commands and syntax of
18, on filter design may be considered optional for an introductory circuits course. The SPICE. These can all be skipped without loss of continuity if SPICE is not available, or
remaining 15 chapters are designed to be covered in order, and it is not recommended that the instructor chooses to forego computer support. End-of-chapter problems requiring the
any of these be skipped without leaving troublesome gaps in the student's preparation for use of SPICE are the final problems offered in each chapter containing SPICE material,
subsequent material. Single-semester courses typically cover Chapters 1-9, rewarding and these problems are clearly marked.
th~ student who will not continue with a second circuits course with a useful familiarity
W1th AC and DC analysis techniques.
INSTRUCTIONAL AIDS
In addition to the bound text, the following ancillary material is available to support the
CHAPTER PEDAGOGY
instructional use of the book.
Each section of each chapter has numerous step-by-step solved examples and ends with Solutions manual. A complete set of clear, concise step-by-step solutions to all of
answt!'red exercises. The end-of-chapter problems are designed to range over the topics the end-of-chapter problems.
of all the sections at all levels of difficulty. Those requiring an electronic calculator are Student problem set with solutions. This large format manual contains an additional
marked with a calculator icon. Boxed equations are particularly important and should set of approximately 500 problems with detailed step by step solutions covering all
not be passed over without clear comprehension as much depends on them. Italicized chapters and topics. This supplement may be purchased separately by the student seeking
areas in the text are highlighted for emphasis, they are the narrative equivalents of boxed further guided practice beyond the solved examples and the answered exercises and odd
equations. numbered end-of-chapter problems.
The appendices contain background material included to make the book more self- Transparency masters. A set of approximately 200 selected figures and tables from
contained without interrupting the basic flow of the narrative. The use of the material the text are available in the medium of overhead transparency masters.
from A~pendix A on linear algebra begins in Chapter 4 and continues regularly thereafter.
A~pend1x ~ on complex numbers should be reviewed before studying Chapter 8, which CONTINUED FEEDBACK IS ESSENTIAL
rehe~ he~vlly on co~plex algebra and the complex exponential time function. Appendix C
on C1rcmt topology 1S perhaps less necessary to solve simple problems, but clarifies im- We have acknowledged our debt to the community of users whose comments were in-
portant logical issues related to forming and solving simultaneous algebraic equations of strumental in the genesis of the present edition, and eagerly anticipate continued dialog
ACKNOWLEDGMENTS \
This work has been greatly enriched by the efforts of those who have contributed to earlier
editions, whose voice can clearly be heard here, as well as those active in preparing the
current edition. Of those in the latter category, special thanks to Profs. Dennis Tyner,
Indira Chatterjee, Keith A. Ross and Bruce Wollenberg for their careful reading and many Alessandro Volta
1745-1827
useful suggestions, to Sondra Chavez, Alan Apt and Eric Svendsen for their invaluable
editorial help. The last-named author shouldered the heaviest load in the current revision This endless circulation of
cycle, and he would like to save the final acknowledgment for his family, who somehow the electric fluid may ap-
understood. pear paradoxical, but it is
no less true and real, and
David E. Johnson you may feel it with your
John L. Hilburn hands.
Johnny R. Johnson Alessandro Volta
Peter D. Scott
()fphy~lcs.at .
with another Wel1-l\JlU'W JIL\al;l~tl1"ifll
at Bologna.
electricitywasanim(lZ . .....v> ..... ;.' . .... ....... . ... ..... .... ..... .......... ; ' . ' ............... .
the other hand mainfamedthat . ... . . . .. . . .. was .fJ'ietaltic elec;tricity; 1fJ,e
source of whic~ was the disSilirilarmehllPt()~esa~acijed.to;the frogs' leg~.-a0th . . .
xiv Preface 1
Chapter Contents More general circuit elements may have more than two terminals. Transistors and
operational amplifiers are common examples. Also, a number of simple elements may be
II1II 1.1 Definitions and Units III 1.5 Circuit Analysis and Design combined by interconnecting their terminals to form a single package having any number
II1II 1.2 Charge and Current II Summary of accessible terminals. We consider some multiterminal elements later, but our main
III 1.3 Voltage, Energy, and Power concern will be simple two-terminal elements. An example of an electric circuit with six
II Problems
such elements is shown in Fig. 1.2.
III 1.4 Passive and Active Elements To understand the behavior of a circuit element, we need to consider certain quan-
tities associated with it, such as current and voltage. These quantities and others, when
they arise, must be carefully defined. This can be done only if we have a standard sys-
tem of units so that when a quantity is described by measuring it, we can all agree on
what the measurement means. Fortunately, there is such a standard system of units that
d
is used today by virtually all the professional engineering societies and the authors of
FIGURE 1.2 Electric \ most modem engineering textbooks. This system, which we use throughout the book,
circuit containing six is the International System of Units (abbreviated SI), adopted in 1960 by the General
elements. Conference on Weights and Measures.
There are six basic units in the SI, and all other units are derived from them. Four
of the basic units-the meter, kilogram, second, and coulomb-are important in the study
of circuits, and we shall examine them in some detail. The other two are the kelvin and
Electric circuit analysis is the portal through which students of electric phenomena begin the candela, which are not essential to our study and thus will not be considered further.
their journey. It is the first course taken in their majors by most electrical engineering The SI units are very precisely defined in terms of permanent and reproducible
and electrical technology students. It is the primary exposure to electrical engineering, quantities. However, the complete definitions are both lengthy and esoteric. * Therefore,
sometimes the only exposure, for students in many related disciplines, such as computer, we shall simply name the basic units and then relate them to the very familiar British
mechanical, and biomedical engineering. Virtually all electrical engineering specialty System of Units, which includes inches, feet, pounds, and so on. Note, however, that the
areas, including electronics, power systems, communications, and digital design, rely SI incorporates the metric system, favored in all but a few countries in the world, rather
heavily on circuit analysis. The only study within the electrical disciplines that is arguably than the more cumbersome British System. Prefixes for powers of lOin the SI system
more fundamental than circuits is electromagnetic (EM) field theory, which forms the are shown, along with their standard abbreviations, in Table 1.1.
scientific foundation upon which circuit analysis stands. However, even the primacy of
EM field theory over circuits is incomplete, since many EM field theory problems tum
out to be best solved by resort to circuit analysis techniques. It is no exaggeration to say Prefixes in the 51
that the ideas and methods of this first circuits course are, for many students, the most
Multiple Prefix Symbol
important to be encountered in their entire undergraduate curriculum. To begin our study
of electric circuits, we need to consider what an electric circuit is and what we mean 10 12 tera T
by its analysis and its design. Since this is a quantitative study, we must identify the 109 giga G
quantities associated with it, in what units these quantities are measured, and the basic 106 mega M
definitions and conventions to be used throughout. These are the topics we consider in 103 kilo k
this chapter:"' 10- 3 milli m
10- 6 micro f-t
10-9 nano n
10- 12 pico p
10- 15 femto f
4 Chapter 1 Introduction
Section 1.2 Charge and Current 5
important of these particles are protons (positive charges) and neutrons (neutral, no net conductor. Unless otherwise specified, all currents are to be understood as conventional
charge) found in the nucleus of the atom and electrons (negative charges) moving in currents.
orbit about the nucleus. Normally, the atom is electrically neutral; the negative charge of Current flow along a lead or through an element will be specified by two indicators:
the electrons balances the positive charge of the protons. Atoms may become positively an arrow, which establishes a current reference direction, and a value (variable or fixed),
charged by losing electrons and become negatively charged by gaining electrons from which quantifies the current flow in the reference direction. Figure 1.3(a) depicts a current
other atoms. i l flowing left to right through a lead, and Fig. l.3(b) a current of 7 A flowing from right
As an example, we may transfer negative charge to a balloon by rubbing it against to left through an element. In the latter case, 7 A of conventional (positive) charge flows
our hair. The balloon will then stick to a wall or ceiling, evidence of the attractive force right to left, which may physically consist of an equal rate of negative charges flowing
between the balloon's excess electrons and the wall, which becomes positively charged left to right. Returning to Fig. 1.3(a), note that if the variable il is positive conventional
when its surface electrons drift toward its interior due to repulsion by the net negative current flows left to right, whereas if it is negative, current is said to flow from right
charge of the balloon. to left. The arrow does not indicate the actual direction of current flow but rather, the
We now define the coulomb (C), introduced in Section 1.1, by stating that an direction of conventional current flow if the algebraic sign of the current value is positive.
electron has a negative charge of 1.6021 x 10- 19 coulomb. Putting it another way, a \ The arrow points opposite to the direction of conventional current flow if the sign of the
coulomb is the aggregate charge of about 6.24 x 10 18 electrons. These are, of course, current value is negative.
mind-boggling numbers, but their sizes enable us to use more manageable numbers, such
as 2 C, in the work to come. 7A
The symbol for charge will be taken as Q or q, the capital letter usually denoting o lJEj@~o
constant charges, such as Q = 4 C, and the lowercase letter indicating a time-varying a . b
charge. In the latter case we may emphasize the time dependency by writing q(t). This (b)
practice will be carried over to other electrical quantities as well.
FIGURE 1.3
The primary purpose of an electric circuit is to move charges at desired rates along
specified paths. The motion of charges constitutes an electric current, denoted by the
letters i or I, taken from the French word intensite. By definition, current is the time As another example, suppose that the current in the wire of Fig. 1.4(a) is 3 A. That
rate of change of charge, or is, 3 Cis passes some specific point in the wire in the left-to-right direction. In Fig. 1.4(b),
- 3 A passes right to left. These two cases result in exactly the same charge transfer
and are thus equivalent ways to denote precisely the same current flow. We may always
reverse both the algebraic sign and the reference direction of a current without changing
either its value or direction. This corresponds to reversing the sign of a number twice,
. dq
1=- (1.1) thus leaving it unchanged. Note that without knowing its reference direction, a current's
dt
numerical value alone is not enough to specify the current. Both value and reference
direction are needed.
-3A
As introduced in Section 1.1, the basic unit of current is the ampere. One ampere of ~
current is said to flow past a given cross section if the net rate of charge movement ~
crossing it is 1 coulomb per second (1 Cis). (a) (b)
In circuit theory, current is generally specified as the movement of positive charges.
FIGURE 1.4 Two representations of the same current.
This convention was proposed by the great American scientist, inventor, and diplomat
Benjamin Franklin (1706-1790), who guessed that electricity traveled from positive to
negative. While this is indeed true in some physical media, we now know that in the Figure 1.5 represents a general circuit element with a current i flowing from the
important case of metallic conductors so common in real-world circuits, the opposite is i left toward the right terminal. The total charge entering the element between time to and
o~1JEj@ 0 t is found by integrating (Ll). The result is
the case. In metals and most other conductors, free electrons with negative charge are
the current carriers, rather than Franklin's postulated positive charge flow. Fortunately
from a circuit theory point of view, there is no difference between positive charge motion
FIGURE 1.5 Current
flowing in a general
qT = q(t) - q(to) = 1t
to
i de (1.2)
in one direction and equal but opposite negative charges moving in the other. Thus we element.
shall cling to the traditional convention of positive charge motion by defining conven- We should note at this point that we are considering the network elements to be
tional current and using this to measure current regardless of the true identity of the electrically neutral. That is, no net positive or negative charge can accumulate in the
charge carrier. Conventional current is the equivalent flow of positive charges in a given element. Any positive charges entering must be accompanied by equal positive charges
6 Chapter 1 Introduction
Section 1.2 Charge and Current 7
leaving (or, equivalently, an equal negative charge entering). This property is a conse-
quence of Kirchhoff's current law, which is discussed in Section 1.3. Thus the current
..""'-----........---....-,..'",..-....--,..,.._=~
EXERCISES
- . . ----""'"-----
~""~""'"
shown entering the left terminal in Fig. 1.5 must leave the right terminal. Note that the
arrow could as well have been drawn pointing out the right side of the element as into 1.2.1. How many electrons are represented by a charge of 0.32042 pC?
the left with no change in current reference direction. Answer 2 million
1.2.2. The total charge entering a terminal of an element is given by
q = 4t 3 - 5t mC. Find the current i at t = 0 and at t = 2 s.
As an example, suppose that the current entering a terminal of an Answer -5 mA; 43 mA
element is i = 4 A. The total charge entering the terminal between 1.2.3. The current entering a terminal is given by i = 1 + rr sin 2rrt A.
t = 0 and t = 3 s is given by Find the total charge entering the terminal between t = 0 and t = 1.5 s.
Answer 2.5 C
q = 1 3
4dt = 12 C
\
Several types of current are frequently encountered, some of which are shown in
Fig. 1.6. A constant current, as shown in Fig. 1.6(a), will be termed a direct current, or
dc. An alternating current, or ac, is by definition a sinusoidal current, such as that of Charges in a conductor, exemplified by free electrons, may move in a random manner.
Fig. 1.6(b). Sinusoids are discussed in some detail in Chapter 8. Figure 1.6(c) and (d) However, if we want some concerted motion on their part, such as is the case with an
illustrate, respectively, an exponential current and a sawtooth current. electric current, we must apply an external or electromotive force (emf). Thus work is
done on the charges. In Section 1.2 we defined voltage across an element as the work
done in moving a unit charge (+ 1 C) through the element from one terminal to the other.
Voltage across an element will be designated by two indicators: a plus-minus sign
pair, which establishes a voltage reference direction, and a value (variable or fixed),
which quantifies the voltage across the element in the specified reference direction. In
Fig. 1.7(a) we see a voltage (or potential difference) of value V6 volts across the element,
>1 measured as potential at the left end of the element minus potential at the right end.
Thus, if V6 > 0 V, the left end is at the higher potential, and if V6 < 0 V, the right end
(a) (b) is at the higher potential. Note that specifying a reference direction does not by itself
indicate which end of the element is at the higher potential. We need not try to guess
the direction of potential drop when assigning a voltage reference direction for a voltage
variable. In Fig. 1.7(b) the right end of the element is +23 V higher than the left, so
there is a +23-V drop right to left across the element or, equally well, a +23-V rise left
to right across the element.
I 23 V
- +
(c) (d) o-i]-----o
f> f
(a) (b)
FIGURE 1.6 (a) Dc; (b) ac; (c) exponential; (d) sawtooth current.
A FIGURE 1.7 Specifying voltage value and reference direction.
>~ji!,
ii<V - - 5V
There are many commercial uses for dc, such as in flashlights and power sup- +
As examples, Fig. 1.8(a) and (b) are two versions of exactly the same voltage. In
plies for electronic circuits, and ac is the common household current found all over the B
(a), terminal A is +5 V above terminal B, and in (b), terminal B is -5 V above A (or
world. Exponential currents appear quite often (whether we want them or not!) when (a) (b)
+5 V below A).
a switch is actuated to close a path in an energized circuit. Sawtooth waves are useful We may also use a double-subscript notation, Vab, to denote the potential of point a
FIGURE 1.8 Two
in devices such as cathode ray tubes (CRTs) used for displaying electrical waveforms with respect to point b. In this case we have, in general, Vab = -Vba' Thus in Fig. 1.8(a),
equivalent voltage
visually.
representations. VAB = 5 V and VBA = -5 V.
8 Chapter 1 Introduction
Section 1.3 Voltage, Energy, and Power 9
1,
, - - ,,
In transferring charge through an element, work is being done, as we have said. called the instantaneous power because its value is the power at the instant of time at
Or, putting it another way, energy is being supplied. To know whether energy is being i which v and i are measured.
supplied to the element or by the element to the rest of the circuit, we must know both ~--ill 0 Note carefully the relationship between the current and voltage reference directions
the polarity of the voltage across the element and the direction of the current through + v -
of the element of Fig. 1.10. If the current reference direction arrow points into the positive
the element. If a positive current enters the positive terminal, an external force must FIGURE 1.10 Element end of the voltage reference direction, the current and voltage so defined are said to satisfy
be driving the current and is thus supplying or delivering energy to the element. The whose terminal variables the passive sign convention. In this case, vi > 0 means the element is absorbing power;
element is absorbing energy in this case. If, on the other hand, a positive current leaves satisfy the passive sign vi < 0 means it is delivering power at that instant of time. More will be said about this
the positive terminal (enters the negative terminal), the element is delivering energy to convention. in Section 1.4.
the external circuit.
As examples, in Fig. 1.9(a) the element is absorbing energy. A positive current As examples, in Fig. 1.9(a) and (b) the passive sign convention is
enters the positive terminal. This is also the case in Fig. 1.9(b). In Fig. 1.9(c) and (d), satisfied (arrow points into the positive end or equivalently out of the
a positive current enters the negative terminal, and therefore the element is delivering negative end). The element is absorbing power: p = (5)(2) = 10 W.
energy in both cases. \ In Fig. 1.9(c) and (d) it is delivering 10 W to the external circuit,
since vi = + 10 W, but these variables violate the passive sign
convention.
Before ending our discussion of power and energy, let us solve (1.4) for the energy
w delivered to an element between time to and t. We have, upon integrating both sides
(a) (b) (c)
between to and t,
(d)
Let us consider now the rate at which energy is being delivered to or by a circuit wet) - w(to) = t vi dr
1to
(1.5)
element. If the voltage across the element is v and a small charge /::;.q is moved through
the element from the positive to the negative terminal, the energy absorbed by the element,
say /::;.w, is given by
For example, if, in Fig. 1.10, i = 2t A and v = 6 V, the energy
/::;.w = v /::;.q delivered to the element between t = 0 and t = 2 s is given by
If the time involved is /::;.t, then the rate at which the work is being done, or that the
energy w is being expended, is given, as /::;.t gets smaller and smaller, by w(2) - w(O) = 12 (6)(2r) dr = 24 J
. /::;.w . /::;.q
hm - - = hm v - Since the left member of (1.5) represents the energy delivered to the element be-
M--+O /::;.t M--+O /::;.t
,or dw dq .
tween to and t, we may interpret wet) as the energy delivered to the element between
the beginning of time and t and w(to) as the energy between the beginning of time and
- = v - =Vl (1.3) to In the beginning of time, which let us say is t = -00, the energy delivered by that
dt dt time to the element was zero; that is,
Since by definition the rate at which energy is expended is power, denoted by p, we have
w(-oo) =0
If to = -00 in (1.5), we shall have the energy delivered to the element from the beginning
dw up to t, given by
p= - =vi 0.4)
dt
wet) = i~ vi dt (1.6)
We might observe that (1.4) is dimensionally correct, since the units of vi are (J/C)(C/s)
or J/s, which is watts (W), defined earlier. This is consistent with (1.5), since
The quantities v and i are generally functions of time, which we may also denote
by v (t) and i (t). Therefore, p given by (1.4) is a time-varying quantity. It is sometimes wet) = It
-00
vi dt = Ito
-00
vi dt + 1t
~
vi dt
+~
1.3.3. A two-terminal element absorbs w millijoules of energy as shown. +
L - _ - ' -_ _ _ _ _ _ _--'-----?> t(ms)
If the current entering the positive terminal is i = 100 cos 1000rrt rnA, find
o 2 8
the element voltage at t = 1 ms and t = 4 ms.
EXERCISE 1.3.3 Answer -50 V; 5 V v Vs v --==- 12 V
-~
(a) (b)
We may classify circuit elements into two broad categories, passive elements and active In other words, the voltage across an independent voltage source always equals its
elements, by considering the energy delivered to or by them. A circuit element is said specified source function, regardless of the current through it. Note that, if the terminal
to be~assive if it cannot deliver more energy than has previously been supplied to it by voltage reference direction and the source function reference direction are opposite one
the rest of the circuit. That is, referring to (1.6), at each t the net energy absorbed by a + another, a minus sign must be inserted in the element law, v = -Vs. since the terminal
passive element up to t must be nonnegative: voltage is defined in this case as the negative of the source function.
Equation (1.8a) is an example of an element law, an equation involving the ele-
wet) = f~oo per) dr = f~oo vi dr :::: 0 (1.7)
v is
ment's terminal voltage andlor current, which describes the behavior of an element. Each
element has a distinct element law, as we shall see for current sources, resistors, and each
The reference directions of v and i in this equation are assumed to satisfy the passive other type of element as it is introduced.
sign convention introduced in Section 1.3 (the arrow for current reference points into the Another symbol that is often used for a constant voltage source, such as a battery
positive end of the voltage reference direction). Only if the passive sign convention is FIGURE 1.12 with 12 V across its terminals, is shown in Fig. 1.11 (b). In the case of constant sources,
satisfied will this integral be guaranteed nonnegative for passive elements (which explains Independent current we shall use Fig. 1.11(a) and (b) interchangeably.
the choice of the name "passive sign convention"). As we shall explore later, examples source. An independent current source is a two-terminal element through which a specified
of passive elements are resistors, capacitors, and inductors. current flows. The value of this current is given by its source function (is in Fig. 1.12)
and its source function reference direction by the arrow inside the source. As with the supplying 24 W of power to the rest of the circuit. In Fig. 1. 13(b),
independent voltage source, the source function is [or is(t) in the time-varying case] is the reference directions do satisfy the passive sign convention, so
assumed known and is related to this element's terminal current variable (i in Fig. 1.12) p = vi = 24 W > 0 signifies that the battery is absorbing 24 W,
by the element law for an independent current source: as would be the case when it is being charged by a more powerful
(1.8b) source somewhere in the circuit.
If the source function is and terminal current i reference directions are opposed rather than The sources that we have discussed here, as well as the circuit elements to be con-
in agreement as shown in Fig. 1.13, a minus sign must be introduced into the terminal sidered later, are ideal elements. That is, they are mathematical models that approximate
law, i = -is. the actual or physical elements only under certain conditions. For example, an ideal
automobile battery supplies a constant 12 V, no matter what external circuit is connected
2A
-----3> to it. Since its current is completely arbitrary, it could theoretically deliver an infinite
amount of power. This, of course, is not possible in the case of a physical device. A real
\ 12-V automobile battery supplies roughly constant voltage only as long as the current
it is required to deliver is low. When the current exceeds a few hundred amperes, the
12 V 12V voltage drops appreciably from 12 V.
2A
-----3>
(a) (b)
EXERCISES
fiGURE 1.13 (a) Source delivering power; (b) source absorbing power.
1.4.1. Find the power being supplied by the sources shown.
Answer (a) 36 W; (b) 20 W; (c) -24 W; (d) -45 W
To show that independent sources are, in general, active elements,
we need only demonstrate that there exist circumstances under which
they may be made to supply rather than absorb net energy. For a
current source with source function is (t) A and element law i (t) =
lJ4A
12V 6V-=- -9V
is(t), suppose that we arrange for the voltage across the source to
be vet) = -is(t) V, with reference directions as shown in Fig. 1.12.
Then, by (1.5), pet) = v(t)i(t) = -i;(t), and the net energy w(t)
absorbed by the source must be nonpositive for each t. If is (t) is not (a) (b)
T
(e) (d)
identically equal to zero for all t, w (t) must be not just nonpositive
EXERCISE 1.4.1
but, in fact, negative for some t. This violates the requirement (1.7)
that passive elements never absorb negative net energy, in other
1.4.2. The terminal voltage of a voltage source is v = 6 sin 2t V. If the
words never supply net energy, under any circumstances. A similar
charge leaving the positive terminal is q = -2cos 2t mC, find (a) the power
argument can be used to show that a voltage source is also an active supplied by the source at any time and (b) the energy supplied by the source
element. Note that this result assumes that the source function is (t) between 0 and t seconds.
is not identically equal to zero for all time t, in which case wet) Answer (a) 24sin2 2t mW; (b) 12t - 3sin4t mJ
would also be zero for all t. Sources with zero source functions
are passive elements; all other sources are capable of producing net
energy and are thus active elements.
Independent sources are usually meant to deliver power to the
external circuit, although in a particular circuit they may be either
net suppliers or absorbers of power. In Fig. 1. 13(a), v = + 12 V is
the terminal voltage across the source and i = +2 A its terminal Let us now tum to the words circuit analysis, which are contained in the title of the
current. Since the reference directions of these terminal variables book, and the related idea of circuit design, and consider their meaning. Generally, if an
do not satisfy the passive sign convention (the arrow points into electric circuit is subjected to an input or excitation in the form of a voltage or a current
the minus end), p = vi = 24 W > 0 implies that the source is provided by an independent source, then an output or response is elicited. The input and
14 Chapter 1 Introduction
Section 1.5 Circuit Analysis and Design 15
output are voltages or currents associated with particular elements in the circuit. There cise the circuit analytically to predict its idealized input-output behavior. In addition,
may be, of course, more than one input and more than one output. computer-aided circuit design software tools such as SPICE or PSpice will prove vital to
There are two main branches of circuit theory, and they are closely linked to the all but the simplest schematic design exercises, speeding the many calculations that must
fundamental concepts of input, circuit, and output. The first branch is circuit analysis, be undertaken at each repetition of the design cycle. Without these simulation tools we
which, given an input and a circuit, is the process of determining the output. The other could not tackle designs of circuits containing many loops and nodes, that is, we would
branch is circuit design, which, given a set of inputs and their desired outputs, is the limit ourselves to a narrow selection of "plain vanilla" design exercises. With these tools
process of creating a circuit that realizes this desired input-output relationship (often in hand, the scope of accessible schematic designs will prove satisfyingly broad and
subject to additional side constraints such as size, cost, etc). practical. Our goal in this study is to present the systematic methods of analysis upon
The balance of emphasis in this book tilts strongly toward analysis. While circuit which design stands, and to offer a first exposure to the challenges and satisfactions of
design is the subject of several chapter sections entitled "Design of ... " and is the sole circuit design.
focus of the final chapter, the balance of this study is devoted to analysis. This may seem
a somewhat puzzling choice. After all, design, the creation of something new and useful,
is for most of us more satisfying and motivating than analyzing a given circuit by solving
for its currents and voltages. The emphasis on analysis perhaps merits some comment.
First consider the process of circuit design, which may be viewed as a cyclic
process. Given the desired inputs and outputs, an initial guess of a circuit is made. This We begin our study of electrical circuits by defining those electrical quantities that will
circuit is then analyzed to determine its actual outputs for the desired inputs, and the be of interest, primarily current, voltage, and power, and the selection of a consistent
result is compared to the desired input-output relationship. A judgment is made as to system of units for their measurement. The SI system chosen incorporates the metric
whether the design goals have been met satisfactorily. If not, modifications in the circuit system (meters, kilograms, seconds) and uses the coulomb, a measure of charge, as a
are made and this iterative-improvement design cycle is then repeated. fundamental unit. SI units of current, voltage, and power are amperes, volts, and watts,
Conceiving a first-cut or initial-guess circuit and selecting the specific modifications respectively.
to be made during each redesign cycle are seldom matters on which two circuit designers
would agree exactly. These choices depend heavily on individual skill, experience, and Current has both magnitude and direction. Its direction is determined by its current
intuition. Matters largely of judgment and creativity, they constitute the "art" component reference direction arrow and the sign of its algebraic value. A positive value indicates
equivalent positive current flow in the direction of the arrow, and negative current
of circuit design. They cannot be codified into explicit, mechanistic rules any more than indicates the opposite direction of flow.
the process by which a sculptor works can. Circuit design is a creative human activity,
like the sculptor's process in designing a statue. Voltage, or potential difference, also has magnitUde and direction. Its direction is deter-
Circuit analysis constitutes the complementary "science" component embedded mined by its voltage reference direction plus-minus signs and its algebraic sign. A pos-
within the circuit design cycle. Without analysis, the input-output behavior of even itive voltage indicated voltage drop in the plus-to-minus direction (higher electric poten-
tial at the plus sign), and negative value indicates the opposite direction of voltage drop.
the initial-guess circuit could not be determined; thus not a single improvement in the
first-cut circuit can be made with confidence. We cannot expect to improve steadily Passive elements cannot supply more energy on balance than they receive from other
what we don't clearly understand. Thus, while circuit analysis can be performed without elements. Active elements, such as independent sources, can.
reference to circuit design, the opposite is not true. The instantaneous power absorbed by a two-terminal subnetwork equals its current-
Since circuit analysis forms the necessary foundation for circuit design, it must voltage product if the reference directions satisfy the passive sign convention (current
be studied before meaningful design is possible. Just as a jazz musician must master arrow points into voltage plus sign or out of minus sign). Negative power absorbed is
the musical scales before creating really interesting and original music, the elements of equivalent to positive power supplied.
circuit analysis are our scales and must be mastered if we are to accept the challenge of Circuit analysis is the process of determining output currents and voltages for a specified
desig'iiing useful and interesting circuits. circuit. Circuit design is the discovery of a circuit with some required input-output
Finally, we note that circuit design may be approached on two levels. Schematic behavior.
design is the solution of a design problem by specifying a circuit diagram, while phys-
ical design is the further conversion of the schematic to actual hardware. In this study
we content ourselves with schematic design. Unlike physical design, schematic design PROBLEMS
does not require a parts inventory, measurement instruments, or access to an electronics
01.1. Einstein's formula for conversion of mass to en- 01.2. Taking the speed oflight to be c = 3.02 X 108 mls and
laboratory. The elements of schematic design are more abstract than physical elements
such as resistors, capacitors, and the like: equations to model the idealized behavior of
Iill.!l ergy is i
E = mc 2 , where E is energy, m is mass, and I!illI a year to contain exactly 365 days, how many kilometers
c = 3.02 X 108 mls is the speed of light. How many joules
the elements contained in the circuit diagram, and the mathematical methods to exer- are there in a light-year?
of energy are there in 5 micrograms of matter?
16 Chapter 1 Introduction
i(A)
g 1.3. The fastest recorded speed for an~ ground vehicle was 1.7. If J(t) in Problem 1.6 is the charge entering the el- 1.19. If VsI = 100 V and is2 = -10 A, which source is
supplying energy and which is dissipating energy? Holding
1iilll6121 mph (an unmanned rocket sled ill 1982). For many ement in coulombs and t is in seconds, find the charge
entering the element between 4 and 9 s, the current at 6.5 s, is2 fixed at -10 A, what would VsI be so that 1000 J/h was
years, the official air speed record for manned flight (a SR-
and the current at 8 s. being supplied by the current source to the voltage source?
71 "Blackbird" aircraft in 1964) was 980.4 mls. Compare
these speeds in kilometers per hour. !
1.8. If 10 J of work is required to move C of negative
01.4. J(t) is the charge in coulombs passing a certain cross charge from point a to point b, find Vab and Vba +
Iiilll section along a wire per unit time. Find (a) the total charge 1.9. If the voltage across an element is 8 V and the current V=V ,1
passing the cross section, (b) the time at which exactly 1 C i entering the positive terminal is as shown, find the power
has passed, (c) the current at t = 5 s, and (d) the current at delivered to the element at t = 7 ms and the total charge
t = 8 s. and total energy delivered to the element between 0 and o - - - - - - - - - - - - - - : 1...0 - t (s)
FIGURE P1.19
10 ms.
10 FIGURE Pl.12
1.20. If a current i = 0.4 A is entering the positive termi-
lOe- t / 4 , 0 :0; t < 8
\
1.13. Repeat Problem 1.12 for vet) = 4(t - 5) V. nal of a battery with terminal voltage v = 12 V, the battery
0.667(10 - t), 8 :0; t < 10 is in the process of being charged. (It is absorbing rather
fUl= { o , t :::: 10 i(mA) 1.14. The power delivered to an element is p =
than delivering power.) Find (a) the energy supplied to the
o ,t < 0 lOt cos 2 t W and the voltage is vet) = 2 cos t V. Find the
14 battery and (b) the charge delivered to the battery in 2 h
current entering the positive terminal and the charge be-
(hours). Note the consistency of the units 1 V = 1 J/C.
f(t) tween times 0 and 2n s.
(CIs) 1.21. Suppose that the voltage in Problem 1.20 varies lin-
10 1.15. An element has vet) = 4e- t V for t :::: 0 s. If we
early from 6 to 18 V as t varies from 0 to 10 min. If
wish this element to supply net energy of 2t J for t :::: 0,
i = 2 A during this time, find (a) the total energy supplied
what must i(t) be?
and (b) the total charge delivered to the battery.
1.22. If a current source and a voltage source are con-
L___ _ _ _ _ _ _ _ _~_~-~t(s) nected back to back as in Problem 1.19, are there values VsI
o 8 10 and is2 for which both sources are supplying equal nonzero
FIGURE P1.4
o 4 10 t (ms) power to one another? If so, what values? If not, show
why not.
FIGURE Pl.9
1.23. The power absorbed by a circuit element is shown.
g 1.5. How long w?uld it take a force of 10 N to accelerate FIGURE PUS
At what time is the net energy a maximum? What is the
Iiilll a 1000-lb mass object to a speed of 55 mph? maximum value of the net energy?
1.6. If the function J (t) is the current in amperes entering 1.16. Show that by definition of a passive element, any
the positive terminal of an element at time t (seconds), find element satisfying v = mi, where m :::: 0 and the element
1.10. If the function graphed in Problem 1.9 is the volt- peW)
(a) the total charge that has entered between 4 and 9 s, variables i and v satisfy the passive sign convention, is
age v (volts) across an element versus the time (ms), and
(b) the charge entering at 8 s, and (c) the current at 1, 5, passive.
the current entering the positive terminal is 3 rnA, find the 2
and 8 s. Take the charge to be 0 at t = O. power delivered to the element at 2 ms and at 6 ms. 1.17. The element shown satisfies v = 2i + l. Is this
element active or passive? Find the net energy between t =
f(t)
1.11. Find the power delivered to an element at t = 2 ms
if the charge entering the positive terminal is
-00 and t = 1 s if v = 0 for t < 0 and v = !V for t :::: O.
8 -+-----'----\~L--f__..L-----L-i.--..:;.. t (s)
q = lOcos 125m mC
6 -I
5
18 Chapter 1 Introduction
Problems 19
I(J)
78
to start the car in lOOF weather? In 32F weather? Assume
6 ~--- _________ ~ that each attempt lasts lOs and draws 90 A of current from
the battery.
-
20 Chapter 1 Introduction
21
Chapter Contents of elements constrains their possible currents and voltages. These laws are valid for
circuits containing elements of all types: resistors, inductors, capacitors, sources, and
III 2.1 Kirchhoff's Laws II 2.7 Practical Sources and Resistors others. They are called Kirchhoff's current law and Kirchhoff's voltage law. These
III 2.2 Ohm's Law II 2.8 Ammeters, Voltmeters, and two interconnection laws, together with element laws describing the behavior of each
Ohmmeters individual element in the circuit, are all the equations we will need for systematic circuit
III 2.3 Equivalent Subcircuits
III 2.9 Design of Resistive Circuits analysis. The element law for a resistor, Ohm's law, will be taken up in Section 2.2, and
III 2.4 Series Equivalents and Voltage systematic circuit analysis methods are developed in Chapter 4.
Division III Summary Our development of Kirchhoff's laws will not include rigorous proofs. These are
III 2.5 Parallel Equivalents and Current III Problems best supplied in the context of the study of electromagnetic fields, the theory upon which
Division Kirchhoff's laws are based. For the present circuit theory purposes, we content ourselves
with defining the laws and justifying their plausability.
III 2.6 Thevenin and Norton
A circuit consists of two or more circuit elements connected by means of perfect
Equivalents
\ conductors. Perfect conductors are idealized leads or wires that allow current to flow
without impediment (no charge accumulations or v9ltage drops along the leads, no power
or energy dissipation). For circuits so defined, the energy can be considered to reside, or
be lumped, entirely within each circuit element, and thus the circuit is called a lumped-
parameter circuit.
A point of connection of two or more circuit elements, together with all the con-
The simplest and most commonly used circuit element is the resistor. All electrical necting wire* in unbroken contact with this point, is called a node. An example of a
conductors exhibit properties that are characteristic of resistors. When currents flow in circuit with three nodes is shown in Fig. 2.1(a). Nodes are indicated in diagrams in two
metals, for instance, electrons that make up the current collide with the orderly lattice ways: by a dashed line enclosing the node, as with nodes 1 and 3, or by marking a typical
of atoms of the metal. This impedes or resists the motion of the electrons. The larger point within the node, as with node 2 of this figure. Whichever is used, it is important to
the number of collisions, the greater the resistance of the conductor. In other materials remember that a node is all the wire in direct contact with a given point, and thus any two
the charge carriers and their surrounding milieu may differ from free electrons flowing points that can be traversed by moving exclusively along connecting leads are both part
through crystal lattices, but the principle that movement of charge is resisted remains of the same node. The points marked a and b in Fig. 2.1(a) are part of the same node,
the same. We shall consider a resistor to be any device that exhibits solely a resistance. designated node 1. We are now ready to discuss the all-important laws of Kirchhoff.
Materials that are commonly used in fabricating resistors include metallic alloys and
carbon compounds and, in the case of resistors that are deposited onto integrated-circuit a Node 1
chip substrates, semiconductors.
In this chapter we first introduce two general laws governing the flow of current
and the pattern of voltages in any circuit, called Kirchhoff's current law and Kirchhoff's
voltage law. We next introduce the basic relationship linking current and voltage in
resistors, Ohm's law. Armed with these tools, we begin our study of systematic circuit
analysis. The fundamental concept of equivalent circuits is introduced. Simplifications
available through the use of equivalent circuits are then discovered in the context of Node 3
... voltages.
In this section we consider two simple laws first stated in 1847 by the German
*!he leads th~t connect eleme~ts will frequently be called "wires" even though they may take on quite
different phYSical form. For mstance, integrated-circuit chips leads are narrow, shallow conducting
physicist Gustav Kirchhoff (1824-1887), which together define how the interconnection channels called traces deposited onto the chip substrate.
(2.1)
where in is the nth current entering (or leaving) the node and N is the number of node
currents.
FIGURE 2.2 Current flowing into a node. As an example of KCL, let us find the current i in Fig. 2.3. Summing
the currents entering the node, we have
For the sake of argument let us suppose that the sum were not zero. In such a case
5+i-(-3)-2=O
we would have 5A
-----7 or i = -6 A
i] + i2 - i3 + i4 = \{J =1= 0
We note that -6 A entering the node is equivalent to 6 A leaving
where \{J has units of coulombs per second and must be the rate at which charges are
the node. Therefore, it is not necessary to guess the correct current
accumulating in the node. However, a node consists of perfect conductors and cannot
direction when assigning reference directions to current variables.
accumulate charges. In addition, a basic principle of physics states that charges can be
2A -3A We still arrive at the correct answer in the end.
neither created nor destroyed (conservation of charge). Therefore, our assumption is not -E-- ~ We may find the current i slightly more conveniently by noting
valid, and \{J must be zero, demonstrating the plausibility of KCL. For every charge
that since i flows into the node, it must equal the sum of all the
carrier entering a node, one is immediately forced out, leaving the net current entering
FIGURE 2.3 Example of KCL. other currents if they are defined to be flowing out (by reversing
the node at each instant of time exactly equal to zero.
their reference directions if necessary), yielding
There are several ways that KCL may be stated, all logically equivalent but each
phrased slightly differently. Multiplying both sides of the KCL equation by -1, we obtain i = -3 + 2 + (-5) = -6 A
(-i]) + (-i2) + i3 + (-i4) = 0 which agrees with the previous answer.
Examining this equation, each term on the left is a current exiting the node. Generalizing,
we arrive at a second form of KCL. We now move to Kirchhoff's voltage law, which states that
The algebraic sum of the currents leaving any node is zero. The algebraic sum of voltage drops around any closed path is zero.
Next, this same equation may be rearranged in the form As an illustration, application of this principle to the closed path abcda of Fig. 2.4 gives
I
i] + i2 + i4 = i3 -v] + V2 - V3 = 0
where currents whose reference directions point into the node are gathered on one side where the algebraic sign for each voltage has been taken as positive when going from +
and those exiting the node on the other. All minus signs disappear, and we have the third to - (higher to lower potential) and negative when going from - to + (lower to higher
form of KCL. potential) in traversing the element.
As with KCL, temporarily suppose the contrary, that the sum is not zero. Then
The sum of the currents entering any node equals the sum of the currents -v] + V2 - V3 = <I> =1= 0
leaving the node. wtkre <I> has units of volts and must equal the potential difference between the initial
point and endpoint of the traversal. But this is a closed path, the two points are one
-15 + v + 10 + 2 = 0
+ v _
or v = 3 V. Had we chosen to move counterclockwise from b, we
d would have
FIGURE 2.4 Voltages around a closed path. +
15 V lOV +15-2-IO-v=O
and the same, and a node can have no potential difference with itself. The supposition
yielding the same result. The sum of rises form yields exactly these
is invalid, and the plausibility of KVL is demonstrated. KVL is in fact an instance of
same two equations, the first starting from the point b and mov-
an even more general physical principle that potential differences along a closed path in
ing counterclockwise, the second from point a moving clockwise.
any conservative energy field must sum to zero. The electric fields produced in lumped- FIGURE 2.5 Circuit to illustrate KVL. Finally, equating drops and rises clockwise, we have
parameter circuits are such fields, and so (for instance) is the field of gravity. If we hike
along a path with uphill and downhill sections, the sum of all the altitude changes around
v + 10 + 2 = 15
any closed path must be zero, for we will have returned to the altitude (gravitational
potential) at which we started.
As with KCL, there are two additional equivalent statements of KVL worth noting.
Multiplying each term in the sum of drops form of KVL stated above by -1 yields the Clearly, none of these equations is more informative than the others. Their choice
sum of rises form. is left to convenience or some arbitrary convention. All other things being equal, we will
adopt the convention of using the sum of drops form and clockwise traversal.
There is a common situation, however, where a special form of the sum of drops
The algebraic sum of voltage rises along any closed path equals zero. will be useful. Suppose that there is a single unknown voltage in a loop, such as v in
Fig. 2.5, which we wish to solve for. Traversing the loop in the plus-to-minus direction
of the unknown voltage (clockwise in Fig. 2.5) and summing the drops to zero and then
The final form of KVL arises when voltages encountered plus-to-minus in the direction of moving all terms but the unknown to the other side of the equation, we have
motion around the closed path are summed on one side of the equation and minus-to-plus
on the other, as in the loop of Fig. 2.4, yielding v=+15-2-1O
from which the value of v may be immediately computed. Examining this equation, the
left-hand side is the voltage drop from the point b to the point c, while the right-hand
side is the negative sum of voltage drops from c to back to b, in other words, the sum of
The sum of voltage drops equals the sum of voltage rises along any closed path. voltage drops from b to c by this alternative (counterclockwise) route badc. Indeed, the
voltage drop between two nodes must be the same regardless of the path used to compute
it, in the present case
In general, KVL may be expressed mathematically as
or v = 15 - 2 - 10 = 3 V
(2.2)
Thus, by using the principle that the voltage drop between two nodes equals the sum of
the voltage drops between the same nodes along any path, a loop containing a single
unknown voltage can be conveniently described and the unknown solved.
where Vn is the nth voltage drop (or rise) in a loop containing N voltages. The signs of By use of similar reasoning, it is easily demonstrated that a desired current into
the terms are as noted in the illustrations, positive for plus-to-minus traversal in the sum a node equals the sum of all other currents at that node, counting outward flows as
of drops form or for minus-to-plus traversal using the sum of rises form, and negative if positive. This may be used to solve for a single unknown current at a node and in
the traversal is in the opposite sense. fact was employed to solve the final example in the earlier discussion of KCL in this
section.
EXERCISES
where i4 flows out, and 3 A and -2 A both flow in (+2 A out equals EXERCISE 2.1.1
- 2 A in). At the upper-right node,
2.1.2. In the circuit for Exercise 2.1.1, find V3.
is = 2 + (-1) = 1 A Answer 12 V
while at the lower right,
'The surface cannot pass through an element, which is considered to be concentrated at a point in
i6 = 1 + (-10) = -9 A lumped-parameter networks.
v v
EXERCISE 2.1.3 \
Georg Simon Ohm (1787-1854), a German physicist, is credited with formulating the
current-voltage relationship for a resistor based on experiments performed in 1826. In
1827 he published the results in a paper titled "The Galvanic Chain, Mathematically
Treated." As a result of his work, the unit of resistance is called the ohm. It is ironic,
(a) (b)
however, that Henry Cavendish (1731-1810), a British chemist, discovered the same
result 46 years earlier. Had he not failed to publish his findings, the unit of resistance FIGURE 2.8 (a) i-v graph for an example linear resistor; (b) i-v graph
might well be known as the caven. for an example nonlinear resistor.
Ohm's law states that the voltage across a resistor is directly proportional to the
current flowing through the resistor. The constant of proportionality is the resistance value In reality, all practical resistors exhibit nonlinear behavior because the resistance
+ of the resistor in ohms. The circuit symbol for the resistor is shown in Fig. 2.7. For
R v of any resistor is affected by conditions such as temperature, which are in turn linked to
element current i and voltage v defined to satisfy the passive sign convention introduced past and present current flow through the element (current flow causes heating). Many
in Section 1.3 (current reference direction arrow points into the plus end of the voltage materials, however, closely approximate an ideal linear resistor over a limited range of
reference direction), Ohm's law is currents and environmental conditions. In this book our interest will be exclusively on
linear elements, including linear resistors (which we will hereafter simply refer to as
FIGURE 2.7 Circuit "resistors"). Implied by the choice to concentrate on linear elements is the result that
symbol for the resistor. v = Ri (2.3)
our equations and techniques will work only over a limited operating range. A l-kQ
resistor pulled out of a lab drawer will reliably draw 1 rnA of current when placed across
where R 2: 0 is the resistance in ohms. i a I-V source, but do not expect this same I-kQ resistor to conduct I kA of current if
---3>
The symbol used to represent the ohm is the capital Greek letter omega (Q). By you should find a I-MV power supply with which to excite it. In all likelihood you will
(2.3), we have R = vii so, dimensionally, measure zero current, because after a brief transient you will be left with just a wisp of
1 Q = 1 VIA smoke and a bad smell in the room (or worse).
It was stated that Ohm's law takes on the familiar form v = Ri if the terminal
If R = 3 Q and v = 6 V in Fig. 2.7, the current is variables satisfy the passive sign convention. If we choose to state this law using variables
. v 6V that violate the convention, the form taken on by Ohm's law changes accordingly. In
1=-=-=2A Fig. 2.9, suppose that we use the variables i and VI, rather than i and vas used in Fig. 2.7,
R 3Q
If R is changed to 1 kQ (using the power of 10 prefixes of Table 1.1), FIGURE 2.9 Resistor with
as the element current and voltage variables, where VI = -v. Then, substituting VI for
then with the same voltage the current becomes terminal variables i and Vj v in v = Ri, Ohm's law becomes VI = -Ri. Both v = Ri and this last equation are
6V not satisfying the passive valid expressions of Ohm's law. To avoid the minus sign, we will almost always choose
i=--=6rnA sign convention. to define resistor current and voltage variables that satisfy the passive sign convention.
I kQ
EXERCiSES
2.2.1. The terminal voltage of a lO-kQ resistor is 5 V. Find (a) the
conductance, (b) the terminal current, and (c) the power dissipated.
Answer (a) 0.1 mS; (b) 0.50 rnA; (c) 2.5 mW
The SI unit for conductance is the siemens, symbolized by S and named for the brothers v sn lOn 2.2.2. A long cable carries 3 A of current while dissipating 72 W. What
Werner and William Siemens, t~o noted German engineers of the late nineteenth century. are the resistance of the cable and the voltage drop across its length?
Thus 1 S = 1 AN. (Another umt of conductance, widely used in the United States is the Answer R = 8 Q; v = 24 V
mho, ~~ich is ohm spelled backward. The symbol for the mho is an inverted o:Uega.) 2.2.3. Find (a) the current i and (b) the power delivered to the resistors
CombInIng (2.3) to (2.5), we see that Ohm's law and the formulas for instantaneous if v = -100 V.
power may be expressed in terms of conductance as Answer (a) +10 A; (b) 1 kW to the lO-Q resistor, 2 kW to the
EXERCISE 2.2.3 5-Q resistor
and
i (terminal cnrrent) A generally useful strategy in analyzing electric circuits is to simplify wherever possible.
~
We shall see how to remove a part of a circuit, replacing it with a simpler subcircuit
+ containing fewer elements, without altering any current or voltage outside that region.
(2.7) v The simpler circuit can then be analyzed, and the results will apply equally to the original,
(terminal more complex, circuit. Such a beneficial trade is possible only when the original and
voltage)
replacement subcircuits are equivalent to one another in a specific sense to be defined.
. The important concepts of short circuit and open circuit may be defined in terms of A subcircuit is any part of a circuit. A subcircuit containing any number of in-
reSIstance. A short circuit ~s a resistor of zero ohms resistance, in other words a perfect terconnected elements, but with only two accessible terminals, is called a two-terminal
conduc~or capable of .Ca.n.:I~g any .amount of current without sustaining a voltage drop FIGURE 2.11
subcircuit. Examples include two-terminal elements, as illustrated in Fig. 1.10, and any
across I~. An open cIrcuit IS a reSIstor of zero siemens conductance, in other words a Two-terminal subcircuit. arbitrary circuit in which a pair of leads is made accessible for measurement or inter-
perfect I~sulato~ capable of supporting any voltage without permitting any current flow The shaded region may connection with other elements. The voltage across and current into these terminals are
through It. In hght of (2.5), an open circuit is equivalent to an infinite resistance or a contain any intercon- called the terminal voltage and terminal current of the two-terminal subcircuit, as in
break in the wire. Two points may be short-circuited by joining them with a wire, ~hich nection of elements. Fig. 2.11.
v v
V2 +V- V3 - VI = 0
and rewriting this equation in terms of currents by applying the terminal laws yields
5i 2 + /J (i) - 5 - 4il = 0
(a) (b) for (a) and
FIGURE 2.12 Independent source elements.
tions and V, i the terminal variables.
VSI is are the source func- 5i2 + h(i) - 5 - 4iJ = 0
for (b).
The terminal laws for ideal voltage and current sources are specified in terms of
only one terminal variable. All other elements, such as the resistor, as well as subcircuits ~ + V2 ~ + V2
with several elements, have terminal laws that are equations linking their two terminal
variables. The terminal law is what is needed to understand the action of elements and
i lt 5Q
i lt 5Q
subcircuits.
+ 2A + + +
We come to the central point of this discussion. Two two-terminal subcircuits are
said to be equivalent if they have the same terminal law. A subcircuit with terminal law
VI /' + v VI .R V
V4
v = fl (i) and a second with terminal law v = h(i) are equivalent if /J (i) = h(i) for
each i, or if gl (v) = g2(V) for every v.
Let's find the terminal laws for the two subcircuits of Fig. 2.13. + +
That for Fig. 2. 13 (a) is clearly v = 2i. In (b), by Ohm's law applied
to each resistor, V21 = i, V22 = i, and by KVL, v = V21 + V22 or (a) (b)
v = i + i = 2i. Since the terminal laws are identical, the subcircuits FIGURE 2.14 Exchanging equivalent subcircuits.
are indeed equivalent.
Clearly, these equations are identical if and only if t~e c~cuits are equivalen~.
Since corresponding equations for the two equivalents ~e all1dentlcal, so must be th~lr
'By constant we mean not dependent on i. The source functions Vs and is might in fact vary with time, solutions. We conclude that equivalent circuits result m equal currents and voltages m
Vs= vs(t) and is = is(t).
any circuit to which they are connected.
34 Chapter 2 Resistive Circuits
Section 2.3 Equivalent Subcircuits 35
The notion of equivalent circuits will prove very helpful in simplifying circuit
problems. We will proceed by replacing more complex subcircuits by simpler ones. E DIVISION
Classes of equivalents to be explored in this chapter include series-parallel and Thevenin-
Norton equivalents. The test for equivalence will always be the identity of the terminal Now that the laws of Kirchhoff and Ohm and the notion of equivalent circuits have been
laws of the two subcircuits. introduced, we are prepared to analyze resistive circuits. We b~gin with a :imple circuit,
While the currents and voltages external to equivalent subcircuits will not be which we define as one that can be completely described by a smgle equatIOn. One type,
changed when one is exchanged for the other in any circuit, the internal behavior of which we consider in this section, is a circuit consisting of a single closed path, or loop,
the equivalents may be quite different. For instance, in Example 2.6 there is no element of elements. By KCL, each element has a common current, say i. T~en Ohm'~ la,:"
in Fig. 2.14(b) with the same terminal voltage as the element in Fig. 2. 14(a). All that and KVL applied around the loop yield a single equation in i that descnbes the CircUlt
is guaranteed by their equivalence is that they can be freely interchanged in any cir- completely. . .,
cuit without affecting any currents or voltages measurable outside their own boundaries. Two adjacent elements are said to be connected III serzes If th~y sha.re ~ common
Based on their effects (resulting currents and voltages) on the rest of the circuit, there is node that has no other currents entering it. Nonadjacent elements are III senes If they are
no way to tell which of two equivalents is in the circuit. each in series with the same element. So chains of series elements may be formed of any
Finally, while there is a useful distinction between the terms subcircuit and circuit, \ length. For instance, in Fig. 2.15(a) the resistors R] and R2 ar~ in s~ries s.ince they share
that of part versus whole, it is often clear enough which we are talking about from the exclusive access to a common node; similarly, R2 and R3 are III senes. SIllce R] and R3
context. Wherever this is the case, we will freely use the term circuit to refer to either are both in series with the same element R 2 , they are in series with one another as well.
a complete circuit or a subcircuit; thus we will use terms such as equivalent circuit and Note that, by KCL, elements connected in series must all carry the same current.
two-terminal circuit. Only where it is thought necessary to emphasize the incompleteness
of a part of a circuit will the term subcircuit be used.
v v
2.3.1. Find the tenninal laws for the two circuits sketched.
Answer (a) v = 9i; (b) v = (2 + R)i
I,:, '>',," '''".
1";-'-'-, ___ ..:._1
90 (a) (b)
20
+
FIGURE 2.15 (a) Single-loop circuit; (b) equivalent circuit.
+
v 180 90 v The single-loop networks of this section consist entirely ~f elements c.onnect~d in
R
series. Consider first the circuit of Fig. 2.15(a). The first step III the analysIs of thIS or
any circuit is the assignment of current and voltage variables, bo~ na~es and reference
directions to each element. Since all elements in the loop are III senes, they carry a
(a) (b) common ~urrent, which we will note as i and assign a clockwise reference dire~tion. We
could have assigned its reference direction counterclockwise had we preferred; III t,?~ en~
EXERCISE 2.3.1
we would draw consistent conclusions. It is never necessary somehow to guess the nght
reference direction in order to arrive at a correct answer. The opposite choice of reference
2.3.2. Find a value for R for which the two circuits of Exercise 2.3.1 direction for i will just result in the opposite sign when we compute its value. Si~ce, for
are equivalent. any number I, I amperes clockwise is the same as - I amperes counterclockwIse, the
Answer R =7 n two results must be identical. .
2.3.3. Find an equivalent to the two-tenninal circuit shown in Exer- We next make the voltage variable assignments V], V2, and V3 for the resistors Rj,
cise 2.3.1(a) containing just a single element. Rand R respectively. Note that the voltage reference directions have been chosen
2, 3, d' . al d
Answer 9-11 resistor to satisfy the passive sign convention jointly with the current reference IrectIOn ~ea y
chosen for their common current i. This will permit us to use Ohm's law for the reSIstors
Then, knowing the source function v, we may solve for the loop current i:
(2.9)
Equation (2.8) or (2.9) also serves as the terminal law for the two-terminal subcircuit 12 V
shaded in blue in Fig. 2.15(a), since it links its terminal current i and terminal voltage
V (by KVL, v is both the source function of the voltage source and the terminal voltage
into the shaded region). Examining Fig. 2.15(b), the terminal law of its shaded region is
v = Rsi FIGURE 2.16 Voltage-divider example.
Comparing this result with (2.8), we see that the terminal laws are identical; hence the
two shaded circuits of Fig. 2.15 are equivalent if Rs is the sum of the three resistances: As another example, suppose that v = 120sint V in Fig. 2.17. Let
us determine the loop current, resistive voltages, and the element
Rs = RI + R2 + R3 power deliveries or dissipations. Replacing the three series resistors
Generalizing in the obvious way, for N resistors connected in series, by a single 60-Q equivalent, the loop current is, by Ohm's law,
120 sin t . A
i= =2smt
60
N
Rs = Lr;
;=1
(2.10)
and the voltages are
VI = Rli = 60 sin t V
v
V2 = R2i = 40 sin t V
A chain of series resistors is equivalent to a single resistor whose resistance is the sum of
V3 = R3i = 20 sin t V
the series resistances. Less formally, series resistance adds. This equivalent resistor can
be freely substituted in any circuit without changing any current or voltage external to Although we did not explicitly use the voltage-divider principle, note
the equivalent circuits themselves. fiGURE 2.17 Single-loop circuit. that these three voltages are in proportion to their resistances.
Continuing our analysis of Fig. 2.15(a), we substitute the current from (2.9) into We next compute the powers. Since VI. V2, and V3 each sat-
Ohm's law VI = Rli for the first resistor, yielding isfies the passive sign convention relative to i, the power dissipated
RI by these elements is just their vi products, or
VI =-V
Rs PI = VIi = 120sin2 t W
Similarly,
P2 = v2i = 80 sin2 t W
R2
V2= - v P3 = v3i = 40 sin2 t W
Rs
R3 Note that the power dissipations are also in proportion to the re-
V3 =-V sistances. This follows from the voltage-divider principle and the
Rs
fact that series elements carry a common current. Finally, V and
Note that these voltages are in proportion to their resistances. This is the principle of i together violate the passive sign convention for the source (the
voltage division: the voltage across series resistors divides up in direct proportion to their current reference direction points out of, not into, the plus end of
resistances. For this reason, the circuit of Fig. 2.15(a) is called a voltage divider. Larger the voltage reference direction). Thus a positive vi product means
resistances have larger voltage drops in a voltage divider. instantaneous power supplied, not dissipated, by this element. Then
38 Chapter 2 Resistive Circuits Section 2.4 Series Equivalents and Voltage Division 39
the power delivered by the source is
vi = 240 sin2 t W
v = Vs! + Vs2 + ... + VsN fiGURE 2.19 (a) Single-loop circuit; (b) equivalent circuit.
This is also the terminal law for the two-terminal circuit. Comparing the terminal law Finally, consider the series interconnection of the remaining type of element found
v = Vs for Fig. 2.18(b), we conclude that the two are equivalent if in resistive circuits, the current source. From Fig. 2.20(a) we have is! = is2 = ... = i sN ,
and at the top node i equals this common value i = is!. From Fig. 2.20(b) we see that
Vs = Vs! + Vs2 + ... + VsN
i = is, and the circuits are equivalent if the single source function is equals the common
A chain of series voltage sources is equivalent to a single voltage source whose source value of the series current source functions.
function is the algebraic sum of the series source functions. More concisely, series voltage
sources add.
+
i
----';>-
~ :VSI +
v t is
Vs2
v - v + Vs
fIGURE 2.18 (a) Series voltage sources; (b) equivalent circuit. But what if the series current source functions we are given are not all equal?
Apparently, charge will be accumulating at the node connecting two unequal sources and
KCL will be violated. Yet we know this cannot happen, since we have agreed that KCL
Find i in Fig. 2.19(a).
will not be violated by any lumped-parameter circuit. This is a case of inconsistent math-
Since the three voltage sources are in series, we may replace
ematical assumptions. Either the given element laws are incorrect or KCL is violated.
them by a single equivalent source. With reference directions as
Once a mathematical inconsistency is discovered in a problem statement, there is nothing
indicated in Fig. 2.19,
more to be done than to note this fact and stop. Inconsistent equations cannot be solved.
= Vba = -7 + 18 + 5 = 16 V
Vs In the real world, if we were to series-connect two current sources with unequal source
functions, they would not behave like ideal current sources whose currents were inde-
Then VR = 16 = 4i, so i = 4 A. Note that we could have assigned pendent of their voltages. The "give" would be on the part of the element law, bringing
the opposite reference direction for the source function vs , in which the assumptions back to consistency.
case we would have had Vs = -16 V and VR + Vs = 0, so VR = We conclude that a chain of current sources in series must have equal source func-
+ 16 V and once again i = 4 A. It is not necessary to guess the best tions, in which case the chain is equivalent to anyone of them. If the given source functions
reference direction for an equivalent source function; either will do. are not all equal, there is a mathematical inconsistency and the circuit cannot be analyzed.
40 Chapter 2 Resistive Circuits Section 2.4 Series Equivalents and Voltage Division 41
EXERCISES
2.4.1. Reduce the circuit to two equivalent elements and find i and the
the instantaneous power dissipated by the 50-V source.
Answer 2 - 4 cos 2t A; 200 cos 2t - 100 W
(a) (b)
Designate the common voltage of these elements v and define resistive currents to
satisfy the passive sign convention relative to v. Applying KCL at the upper node, we
obtain
EXERCISE 2.4.1
2.4.2. Find R in the circuit shown. Using Ohm's law in the form i = Gv for each term on the right and gathering the
10Q Answer 6 Q common voltage factor yields
2.4.3. Find values for isl and Vs2 so that v = 4 V.
Answer -6 A; 0 V (2.ll)
+
IV R 6V where each conductance G i is the inverse of the corresponding resistance R i . Then,
knowing the current source function i, we may solve for v:
+
6Q v=------ (2.12)
G, +G 2 +G3
EXERCISE 2.4.2 6A v
Equation (2.11) or (2.12) also serves as the terminal law for the two-terminal circuit
2Q shaded in blue in Fig. 2.21(a), since it links terminal current i and voltage v. The terminal
law for the shaded circuit in Fig. 2.21(b) in terms of the conductance G p = 1jRp is
EXERCISE 2.4.3
Comparing this with (2.11), the two shaded circuits of Fig. 2.21 are equivalent if G p is
the sum of the three conductances:
Generalizing in the obvious way for N resistors connected in parallel, a set of parallel
resistors is equivalent to a single resistor whose conductance is the sum of the parallel
Another important simple circuit is the single-node-pair resistive circuit. Just as single- conductances. Informally, parallel conductance adds. By denoting the conductance of
loop circuits lead naturally to series interconnections and voltage dividers, single-node- this equivalent resistor as
pair circuits will be shown in this section to lead to the equally useful ideas of parallel
interconnection and current division.
Two elements are connected in parallel if together they form a loop containing no
other elements. By KVL, elements in parallel all have the same voltage across them. (2.13)
The single-node-pair circuit of Fig. 2.21 contains three resistors and a current source all
in parallel, since any pair of the four elemtnts forms a loop containing no other elements.
. R2 .
ll= 1
Rl +R2
(2.14) . R1
l2 = I (2.17b)
Rl +R2
For two resistors in parallel, current divides inversely with their resistances. Note that
the same caution holds with this as with the product-by-sum rule: it works only in the
A set of parallel resistors is equivalent to a single resistor whose resistance is the inverse
case of exactly two resistors in parallel.
of the sum of the inverses of the parallel resistances.
This inverse-inverse law has an even simpler form when N = 2. In this case (2.14)
becomes
Let us determine how the 18 rnA delivered by the source divides
among the four resistors in Fig. 2.22. Given their resistances, we
note their conductances to be Gl = 4 mS, G 2 = G3 = 2 mS,
and G4 = 1 mS. Then the currents will be, by the current-divider
principle, in the proportions 4 : 2 : 2 : 1, with Rl having the largest
(2.15) current and R4 the smallest. The equivalent parallel conductance is
or
the sum
The equivalent resistance of two resistors in parallel is the product of their resistances
divided by the sum. Unfortunately, this product-by-sum rule does not directly apply to G p = 4 + 2 + 2 + 1 = 9 mS
sets of more than two resistors in parallel. In this case, (2.13) or (2.14) should be used.
and by (2.16), for the case of four parallel resistors,
Note by (2.13) that adding parallel resistors can only increase the equivalent con-
ductance, in other words decrease the equivalent resistance. Putting resistors in parallel
G1 4
reduces the overall resistance below that of any of them individually. For N equal R-Q 11 = - i = -(18 rnA) = 8 rnA
Gp 9
resistors in parallel, by (2.14), Rp = R/N; in other words, putting N equal resistors in
parallel reduces the resistance by a factor of N. G2 2
l2 = - i = -(18 rnA) = 4 rnA
Continuing our analysis of Fig. 2.21(a), we substitute the voltage from (2.12) into Gp 9
Ohm's law in the form il = Gl v for the first resistor, yielding
G3 2
l3 = - i = -(18 rnA) =4 rnA
Gp 9
(2.16a)
G4 1
l4 = - i = -(18 rnA) = 2 rnA
Gp 9
Similarly,
. G2. Note that the currents are in the predicted proportions (4: 2 : 2 : 1).
l2=-l (2.16b)
Gp
. G3.
13=-l (2.l6c) +
Gp
Note that these currents are all in proportion to their conductances. This is the principle 18mA 250 n 500 n 500 n 1000 n v
of current division: the current through parallel resistors divides up in direct proportion
to their conductances. For this reason, the circuit of Fig. 2.18(a) is called a current
divider. Smaller resistances (larger conductances) have larger current flows in a current
divider. FIGURE 2.22 Current-divider example.
44 Chapter 2 Resistive Circuits Section 2.5 Parallel Equivalents and Current Division 45
Continuing the example, let us see how the power divides.
The voltage v is found from the equivalent
+ +
18 rnA
i
v=-=--=2V
Gp 9 mS
v v
or by dividing any of the resistive currents computed above by the 6Q 18 Q
conductance of that resistor. Then noting that v and that each re-
sistive current in Fig. 2.22 satisfies the passive sign convention, the
dissipated power in RI is (a) (b)
PI = vii = (2 V)(8 rnA) = 16 mW fiGURE 2.24 (a) Circuit; (b) its equivalent.
Thus the 36 mW produced by the source is divided among the resis- R _ (18)(6) _ ~Q
tors comprising the current divider in the same ratio as the currents eq - 18 + 6 - 2
themselves, and the power supplied equals the power dissipated over
Then v = Reqis = (9/2)(20) = 90 V and i l = -v/I8 = -5 A.
the entire circuit.
Note the minus sign in Ohm's law here, since i l flows out of the
positive end of v and this pair of variables violates the passive sign
We next consider current sources connected in parallel. Examining Fig. 2.23(a), convention.
i = isl + is2 + ... + isN
while, for Fig. 2.23(b), i = is. The terminal laws are identical if Finally, voltage sources in parallel are vulnerable to the mathematical inconsistency
problem that we noted in Section 2.4 with current sources in series. If two or more voltage
is = isl + is2 + ... + isN
sources in parallel are all equal, they can be replaced by an equivalent single source of
A set of parallel current sources is equivalent to a single current source whose source the common value, and if they are not all equal, there is an inconsistency between KVL
function is the sum of the parallel source functions. Concisely, parallel current sources and the element laws for these sources. A set of voltage sources in parallel must have
add. equal source functions, in which case anyone of them is equivalent to the entire set. If
the given source functions are not all equal, there is a mathematical inconsistency and the
i i circuit cannot be analyzed.
---7 ---7
+ +
v v is
46 Chapter 2 Resistive Circuits Section 2.5 Parallel Equivalents and Current Division 47
2Q Solving the last for v yields
From our previous discussion of equivalents in Section 2.3, we know that two
equivalent subcircuits can be freely exchanged in any circuit without altering any current
or voltage external to the subcircuits. If it is to our advantage, we may replace a Thevenin
form by a Norton form (or vice versa) before computing a desired current or voltage.
The following example shows how it may be to our advantage to do so.
The series and parallel equivalents described thus far are l~mited to el~ments of the same
type: resistors in series with other resistors, current source m parallel With c~ent sources,
and so on. In this section we develop a pair of equivalents, called Thevemn and ~.ort~n Find i I in Fig. 2.26.
forms, containing both resistors and sources. They will prove to be of great utilIty m First, replace the Norton form to the right of the dashed line
simplifying many circuit analysis problems. ., . by its Thevenin equivalent. Since RN = 4 Q and iN = 1 A, to
Consider the pair of two-terminal circuits shown m FIg. 2.25. Applymg KVL to (a) be equivalent we require that RT = 4 Q and VT = (4)(1) = 4 V.
gives Performing this replacement, we have the circuit of Fig. 2.27(a). This
simple single-loop network may be further simplified by combining
v = -RTi + VT (2.18)
series resistors and series sources, yielding the circuit of Fig. 2.27(b).
while KCL at the upper node of (b) yields From this circuit, i l = 12/6 = 2 A. Note that these additional
16V 4Q lA
series simplifications were not possible before the Thevenin-Norton
(2.l9a) transformation.
(a) (b)
(a) (b)
FIGURE 2.27 (a) Circuit of Fig. 2.26 after Thevenin-Norton transfor-
FIGURE 2.25 (a) Thevenin and (b) Norton subcircuits. mation; (b) combining series elements.
i i
---7 RT ---7
+
+ +
(b)
(a) (b)
FIGURE 2.29 Circuit for Example 2.13.
FIGURE 2.28 (a) Two-terminal circuit; (b) Thevenin form.
single-node-pair type, with KCL
Equation (2.21) is satisfied by selecting Rr and Vr so that the slope of f (i) equals
26_2=vo Vo
- Rr and the intercept is Vr. The resulting Thevenin form, Fig. 2.26(b), is then equivalent 2 + 14
to the given circuit, Fig. 2.26(a). The Norton equivalent may be found by first finding the
Thevenin equivalent and then performing the Thevenin-Norton transformation described or Vo = 42 V.
by (2.20). Or the Norton equivalent may be found directly by matching the slope of f(i)
to -RN and the intercept to RNiN' As Example 2.13 illustrates, a circuit with many elements will have Thevenin and
Norton equ~valents with only two. Replacing the original by its two-element equivalent
The voltage Vo in Fig. 2.29(a) is required. We first simplify the has the desIred effect of simplifying the circuit.
circuit by replacing the circuit to the right of the dashed line by its Given the terminal law for a circuit, it is straightforward to find its Thevenin and
Norton equivalent. To do so, we will compute its terminal law. By Norton equivalents. In the previous examples, we derived the terminal law from the
KCL, circuit diagram by applying Kirchhoff's laws and the element laws. It would be more
convenient to determine the equivalents directly from the circuit diagram. Setting i = 0
i] = 10 + i - 6 = i +4 in (2.21) yields
and by KVL f(O) = Vr (2.22)
v = -4i - 8i] - 2i +4 With terminal law v = f(i), f(O) must be the terminal voltage of the given circuit when
Combining, we have the terminal law its terminal current i is zero. Since no current can flow if the terminals of the circuit
are open circuited, f (0) is called the open-circuit voltage v = Voe of the circuit, and we
v = -14i - 28 see by (~.22~ that t?e Thevenin equivalent voltage Vr is equal to the open-circuit voltage
of the CIrcUlt. If Instead of open circuiting the terminals, we short circuited them, a
The equivalent Thevenin form then has Rr = 14 Q and Vr =
short-circuit current i = ise would flow. Since v = f(i), and across a short circuit there
-28 V, and the equivalent Norton form has RN = Rr = 14 Q and
is no voltage v = 0, (2.21) in the short-circuit case becomes
iN = Vr/RN = -2 A. Choosing the Norton form, the circuit after
replacement is shown in Fig. 2.29(b). This is a simple circuit of the (2.23)
+ +
Given a subcircuit with open-circuit voltage Voe and short-circuit current ise , its !
I/;{
VI = G~) 24 = 18 V Equation (2.25) suggests that RT (or R N) can be computed as the resistance looking into
the terminals of the subnetwork when all internal sources have been killed. The sign
i 6Q 4Q
-E--
and by KVL around the left loop V = Voe = 0 + 18 = 18 V. So the change i ' = -i was necessary to create a pair of terminal variables (ii, v) that together
+ Thevenin equivalent voltage source is VT = Voe = 18 V. To find ise , satisfy the passive sign convention so that Ohm's law will hold without a minus sign.
we assume that the terminals are short circuited, v = O. Then the Note that killing a source means setting its source function to zero. Since a 0- V
v 24V two leftmost resistors are in parallel (they form a loop containing no voltage source is simply a short circuit, independent voltage sources are killed by replacing
other voltage drops), and by the product-by-sum rule, them by short circuits. Similarly, since a O-A current source is an open circuit, kill
independent current sources by replacing them with open circuits.
_ (12)(6) _ 4 Q
Req - 12+6 -
FIGURE 2.30 Circuit for Example 2.14.
Consider the circuit of Fig. 2.30. After killing internal sources, we
In this case VI is, by voltage division of 24 V between the two 4-Q
have the circuit shown in Fig. 2.32. The resistance looking into its
resistors,
terminals is found by combining the parallel 12- and 4-Q resistors
4
VI = 824 = 12 V
i 6Q
~
4Q
Since VI is also the voltage across the 6-Q resistor under these short-
circuit conditions, by Ohm's law VI = 6i. Then, solving for i = ise +
gives
. VI 12 v 12Q
Zse = - = - =2 A
6 6
Hence RT = voe/ise = 18/2 = 9 Q. The Thevenin form and
corresponding Norton form (RN = RT = 9 Q, VT = Voe = 18 V,
iN = ise = ~ A) are shown in Fig. 2.31. fiGURE 2.32 Circuit for Example 2.15.
EXERCISES
2.6.1. Find the Thevenin equivalent of the circuit to the right of line cd.
Answer VT = 6 V; RT = 1.5 kQ
lkQ lkQ
,-
12V 4mA IkQ
EXERCISE 2.6.1 FIGURE 2.33 Practical sources connected to a load resistor: (a) voltage
source; (b) current source.
2.6.2. Repeat Exercise 2.6.1 for the circuit to the left of line abo
Answer VT = 14 V; RT = 2 kQ
2.6.3. Use the results of Exercises 2.6.1 and 2.6.2 to find VI. The effect of the source resistance Rs in the practical voltage source model of
Answer -1 V
Fig. 2.33(a) may be seen by plotting load voltage VL and current iL versus load resistance
RL as shown in Fig. 2.34. For large values of RL the source is lightly loaded (little current
is being drawn), and the source produces values closely approximating the ideal voltage
source Rs = 0, as shown. As RL converges toward zero, VL decreases from its ideal
value, as does i L. The power dissipated by the load is
L---------------------------~RL
see in the following example.
\
(a) (b)
The open-circuit voltage across the terminals of a fresh C cell is
FIGURE 2.34 Comparison of practical and ideal voltage sources: (a) vL
measured to be 1.55 V. When a 29-Q resistor is placed across it,
versus RLi (b) it versus RL its terminal voltage drops to 1.45 V. Find the Thevenin and Norton
equivalents of the C cell, and the maximum current that can be drawn
Similarly, the load current and voltage comparisons for ideal and practical current from this battery.
sources are shown in Fig. 2.35. The practical values are reduced for current sources with The Thevenin equivalent voltage VT = VOG> its open-circuit
RL large. The power to the load is, noting the current divider in Fig. 2.33(b), voltage. This is 1.55 V as shown in Fig. 2.36(a). Noting that RT
and 29 Q form a voltage divider, we find that
P = iIRL = ( Rs is)2 RL (2.27) 29 ) (1.55) = 1.45
RL +Rs (
29+R T
For an ideal current source with Rs infinite, P goes to infinity with RL. For a practical
or RT =2 Q
current source with finite source resistance Rs > 0, as RL goes to infinity in (2.27),
P goes to zero. The source resistance Rs serves to limit the open-circuit voltage the The Thevenin equivalent of the C cell is shown in Fig. 2.36(b).
practical current source may produce, and thus its power to the load. The comparisons Maximum current will flow if the device is short-circuited, Vab = O.
for ideal and practical current sources are shown in Fig. 2.35. In this case
1.55
i max = ise = 2 = 775 rnA
..
.....
Ideal current source
iL=is
Ideal current source
2Q
...... "",,--va
..
.. VT =1.55 V 29 Q 1.55 V
'-----ob
(a) (b)
VL = Rs+RL Is
~------------------------~RL OL---------------------------~RL FIGURE 2.36 (a) Circuit for the examplei (b) practical source.
(a) (b)
Note that while ideal sources can supply power to a circuit without themselves
FIGURE 2.35 Comparison of practical and ideal current sources: (a) VL dissipating any fraction, this is not the case for practical sources. For instance, in Ex-
versus RLi (b) it versus RL ample 2.16, if the C cell is supplying 100 rnA, there will be P = 0.1 2 (2) or 20 mW
56 Chapter 2 Resistive Ci rcu its Section 2.7 Practical Sources and Resistors 57
r--<-
I
power dissipated by the internal source resistance of the cell. Sources feel warm to the We next tum to practical resistors. Resistors are manufactured from a variety of
touch when they are operating because they always drop some of the power they produce materials and are available in many sizes and values. Their characteristics include a
internally across their source resistance. This power is not available to supply to other nominal resistance value, an accuracy with which the actual resistance approximates the
elements. nominal value (known as tolerance), a power dissipation, and a stability as a function
of temperature, humidity, and other environmental factors. They may be produced as
discrete elements or deposited onto substrates as part of an integrated circuit containing
A practical current source with a I-A short-circuit current and Rs =
100 Q look-in resistance is to be used to deliver A to a 100-Q i2 other elements such as transistors and capacitors.
The most common type of discrete resistor found in electrical circuits is the carbon
load R L . Determine an external resistance RSH that can be placed in
composition or carbon film resistor. The composition type is made of hot-pressed carbon
"shunt," or across the source terminals, to achieve this result. Then
granules. The carbon film device consists of carbon powder that is deposited on an
combine RSH with the original current source in a Norton source
insulating substrate. A typical resistor of this type is shown in Fig. 2.38. Multicolored
model, and determine the maximum power dissipation internal to
bands, shown as a, b, c, and % tolerance, are painted on the resistor body to indicate the
this new practical source under all possible load conditions.
\ nominal value of the resistance. The i~olor code for the bands is given in Table 2.1.
We first use current division to determine RSH Let GSH =
1/RSH be the conductance. By current division, to achieve the pre- % Tolerance
scribed load current
iL = (
0.01 ) (1) = -1
0.01 + 0.01 + GSH 12
and solving for GSH, we obtain iii
abc
GSH = (12)(0.01) - 0.02 = 0.1 S
FIGURE 2.38 Carbon resistor.
So RSH = 10 Q. Then the Norton equivalent of the left of ab in
Fig. 2.37(a) has Norton equivalent resistance
% Tolerance Band
,.----,---0 a
a
Gold 5%
Silver 10%
9.09&1
aThese colors apply to band conly.
b "--_--'-_--0 b
Bands a, b, and c give the nominal resistance of the resistor, and the tolerance
band gives the percentage by which the resistance may deviate from its nominal value.
Referring to Fig. 2.38, the resistance is
Original current
source R = (lOa + b)10 % tolerance
C (2.28)
(a) (b)
by which we mean that the % tolerance of the nominal resistance is to be added or
FIGURE 2.37 (a) Circuit for Example 2.17; (b) Norton source model. subtracted to give the range in which the resistance lies.
resistance of the electrical coiL Clearly, a voltage appears across the ammeter terminals
as a result of the current i flowing through RM. RM is usually a few ohms, and the
terminal voltage for a full-scale current is typically from 20 to 200 mY. FIGURE 2.42 Ohmmeter circuit.
The d' Arsonval meter is an ammeter suitable for measuring dc currents not greater The current sensitivity of a voltmeter, expressed in ohms per volt, is the value
than the full-scale current Ips. Suppose that we wish, however, to measure a current that obtained by dividing the resistance of the voltmeter by its full-scale voltage. Therefore,
exceeds Ips. We must not allow a current greater than Ips to flow through the device
without risking damage to the delicate device. A circuit to accomplish this is shown in QfV
.
raung =
Rs + RM ~ -
Rs
Fig. 2.41(a), where Rp is a parallel resistance which reduces the current flowing through Vps vps
the meter coiL (Note: "~" means approximately equal to.)
From current division we see that A simple ohmmeter circuit employing a d' Arsonval meter for measuring an un-
known resistance Rx is shown in Fig. 2.42. In this circuit, the battery E causes a current
i to flow when Rx is connected into the circuit. Applying KVL, we have
-E + (Rs + RM + Rx)i = 0
where ips is the current that produces Ips in the d' Arsonval meter. (Clearly, ips is the from which
maximum current the ammeter can measure.) Solving for Rp, we have E
Rx = --:- -
I
(Rs + RM)
We select E and Rs such that for Rx = 0, i = Ips. Therefore,
E
A dc voltmeter can be constructed using the basic d' Arsonval meter by placing a Ips = Rs + RM
large resistance Rs in series with the device, as shown in Fig. 2.41(b). It is obvious that the Combining the last two equations, we find that
full-scale voltage, v = Vps, occurs when the meter current is Ips. Therefore, from KVL,
Rx = (I~s - 1) (Rs + RM)
A very popular general-purpose meter which combines the three circuits described
from which previously is the YOM (voltmeter-ohmmeter-milliammeter). In the YOM, provisions are
made for changing Rp and Rs so that a wide dynamic range of operation is provided.
Scales may then be laid out on the face of the instrument to correspond with the resistance
values Rx given in the preceding equation for each available Rs.
Electronic volt-ohm-ammeters employ active circuits, such as precision op amp
R, stages, instead of the passive d' Arsonval movements. They can be made to measure
with precision unavailable using the d' Arsonval meter, and have the added advantage of
coupling naturally with digital readouts.
EXERCISES
(02)2
3 I I S
N G = 16 - "8 = 16
P = 1~ (470) = 0.188 W
or RN = 16 Q. The Norton source elements are IN = 3 A and
RN = 16 Q, so the Thevenin source has VT = (3)(16) = 48 V
well below the ~ W permitted. The power dissipated in Re2 divides Source Load
and RT = 16 Q. This source will deliver 16 V to an 8-Q load and
\
proportionally with the conductance of the parallel elements, so each never produce more than 3 A, as required.
FIGURE 2.44
of the two 147-mS conductances dissipate more power than the other
conductances. By current division, the power dissipated in each
147-mS (6.8 Q) element is One very practical class of resistive circuits are those that. supp~y r~lativel.y fixed
Our goal is to design a battery charger circuit that will deliver con-
trolled levels of dc current to charge batteries of different sizes. We
wish to deliver 10 rnA of current to larger batteries, which will be
Note that this solution is hardly unique. We could have begun
connected between terminals a and a', and 1 rnA to smaller batter-
by placing seven 6.8-Q resistors in series rather than ten 470-Q
ies being charged at terminals b-b'. At any terminal voltage with
resistors in parallel. This may seem to save resistors, but minimizing
batteries from zero to full charge 1.5 V, and any battery internal
the number of resistors used was not an explicit design specification.
resistance from zero up to 30 Q, the charging currents must remain
Among distinct solutions to a design problem it is common sense to
reduce the component count where possible, and this may seem to within 5% of the target values.
We begin the design by noting that the circuit shown in
favor the series alternative. Note, however, that had we done so, the
Fig. 2.45(a) will deliver
entire current 1 = 0.2 A needed to meet the 2-W power requirement
would flow through each 6.8-Q resistor, yielding Vs - VB
1=--- (2.29)
Rs+RB
P = (0.2 2 )(6.8) = 0.272 W
where Vs and Rs are the source voltage and resistance of the charger
which would exceed the ~-W limit per resistor. We could work circuit, VB the battery voltage, and RB its internal resistance. Since 1
around this problem by substituting a pair of parallel 6.8-Q resistors must be relatively constant as VB and RB vary, we will set Vs VB
in series with another such pair for each of the seven proposed
6.8-Q resistors, since each four-resistor array would have the same I
equivalent resistance of 6.8 Q, but each resistor would dissipate only -----7
v, 30V
Design a practical electrical source (by specifying its Thevenin equiv-
alent components) which delivers at least 16 V dc to an 8-Q load
but is current limited to 3 A.
Battery
We will determine the Norton source, then transform to get the
(a) (b)
desired Thevenin values. Connecting the source and load as shown
in Fig. 2.44, we wish h to be less than or equal to 3 A regardless of FIGURE 2.45 (a) Source and battery; (b) final circuit.
Once again the range of errors (3%) is well within the 5% tol-
erance. The final design is shown in Fig. 2.45(b).
---~--1i'::;;I----_____ ~
KCL asserts that the net current entering any node or region is zero.
\ +
KVL asserts that the net voltage drop around any closed loop is zero.
FIGURE P2.1
The i-v law for resistors, Ohm's law, is v = Ri, with R the resistance in ohms, and is
valid as long as i and v satisfy the passive sign convention.
2.2. Write KCL for this node:
Resistors are passive elements; that is, they cannot supply power. (a) In sum-entering-equals zero form.
Two subcircuits are equivalent if they have the same i-v law. Equivalent subcircuits (b) In sum-leaving-equals zero form.
may be freely interchanged without changing any external variable. Examples include (c) In sum-entering-equals-sum-Ieaving form.
FIGURE P2.5
series-parallel and Thevenin-Norton equivalents.
Series resistance adds, parallel conductance adds.
Parallel resistance obeys the reciprocal-reciprocal law: the equivalent resistance is the
2.6. Assign names and reference directions for al~ vol~
reciprocal of the sum of the reciprocals of the parallel resistances.
ages, then write KVL for each of the seven loops m thIS
Series voltage divides in proportion to resistance, parallel current divides in proportion circuit.
to conductance (or inversely proportional to resistance).
A Thevenin form is a two-terminal series interconnection of voltage source and resistor,
a Norton form is a parallel interconnection of current source and resistor.
FIGURE 1>2.2
Any two-terminal subnetwork has a Thevenin equivalent and a Norton equivalent form.
These may be found by determining the open-circuit voltage, short-circuit current, and
resistance looking into the terminal pair. 2.3. KCL equations for a certain three-node circuit are:
Node 1: it + i2 - i3 = 0
Thevenin and Norton transformations are useful in replacing a selected region of a Node 2: -il - i2 - i4 = 0
circuit to be analyzed by a simpler one which will not result in any changed currents Node 3: i3 + i4 = 0
or voltages outside its own boundaries. Sketch the circuit, indicating current reference directions. FIGURE P2.6
Practical voltage and current sources may be modeled accurately by their Thevenin and 2.4. Write KVL around each of the three loops.
Norton forms.
Ammeters, voltmeters, and ohmmeters based upon the d' Arsonval movement are in 2.7. Supply reference directions so that the~e ~CL ~qua
wide use. The force between a fixed and variable magnetic field moves a pointer. tions are all valid and the passive sign convention IS satisfied
+ everywhere.
While KCL, KVL, i-v laws, and equivalent subcircuits are introduced here in the context il + i4 - i2 = 0
of resistive circuits studied in the time domain, these fundamental ideas will apply equally h + i8 - i5 = 0
well as we broaden our scope in later chapters to include other types of elements and il + i3 - i5 = 0
other domains of analysis. i6 - i4 - i8 = 0
FIGURE P2.4 i2 + i3 - i6 - h =0
70 Chapter 2 Resistive Circuits
Problems 71
2.15. Find v for each subcircuit. In each case, R = 1 kn. 2.19. By what factor would we increase the source func-
tion in the circuit of Problem 2.17 if we wished to double
lOrnA the power it supplies to the rest of the circuit? Justify.
~
R v R v R v R 4Q
+
+ - lOrnA + lOrnA
~ -<E--
FIGURE P2.12
FIGURE P2.7 FIGURE P2.15
\
2.16. Find VI, i2, and the power produced (or dissipated)
2.13. Assign voltage reference directions for each resistor FIGURE P2.20
by the 7-V voltage source.
so that the passive sign convention is satisfied. Determine 2.21. Find the terminal law v
each voltage if i2 = 2 A and i4 = -4 A. g(v).
J(i). Rewrite as i =
2.S. In a given circuit, suppose that we write KCL at each 0t A
IQ
I
-<E-- 2Q 2Q Hl
node in the sum-entering-equals-zero form.
+
(a) In how many equations will a given element current 4Q
appear? With what signs? Justify.
(b) If all these equations were added together, what would 4>2 4>2 2>2
we get?
6Q
(c) Use part (b) to show that the last node equation is just
the negative sum of all the previous ones. 8Q 2A t
2.9. Assign names and reference directions; then write lOQ FIGURE P2.21
KCL at each node and KVL around each loop. 2.22. Find the terminal law for this circuit. Note that only
one of the two forms v = J (i) and i = g (v) exists in this
case. 4 cos t A
FIGURE P2.13 2Q
+
2.14. Assign voltage reference directions for all five ele- FIGURE P2.16
ments so that the voltage drop across each element is posi- 6V
tive in the clockwise direction; then assign current reference 2.17. Find all currents and voltages.
directions that satisfy the passive sign convention.
FIGURE P2.9
IQ FIGURE P2.22
+ +
5sin 2tY
2.23. A circuit has 4 nodes. The resistor connecting nodes a
and c is Rae= 360ft Similarly Rad = 180n, Rbc =540V, Rbd
2.10. A toaster is essentially a resistor that becomes hot 2Q
= 540n. Open means no other connections. (a) Find the
when it carries a current. If a toaster is dissipating 960 W equivalent resistance looking in terminals a -b if terminals
at a voltage of 120 V, find its current and its resistance. c-d are open, and if terminals c-d are shorted together. (b)
FIGURE P2.17
2.11. Find the energy used by a toaster with a resistance Find the equivalent resistance looking in terminals c -d if ter-
4Q 3Q
of 12 n, which is operated at 120 V for 10 s. minals a -b are open, and if terminals a -b are shorted to-
2.1S. Determine the power supplied by, or dissipated by,
2.12. Find ii, i2, and v. FIGURE P2.14 gether.
each of the four elements in the circuit of Problem 2.17.
48 fl
fiGURE P2.27
+
2.28. If all resistors in a voltage divider are scaled by
v
the same factor a, how does this affect the voltages across 24V 30 fl 48 fl
them? 'The currents through them? The power? Use this
result to design a voltage divider using a +5-V source with
FIGURE P2.31
outputs of +4.0, +3.0, and +1.5 V (all with respect to
a common negative terminal) and a power dissipation of FIGURE P2.40
1 mW. Start by taking RI to be 1 Q; compute the other R's 2.32. Determine the equivalent resistance of the circuit of fiGURE P2.36
and scale. Problem 2.31.
2.33. If the conductances of all resistors in a current di- 2.37. A current divider is to be constructed using 2.41. Use a series of Thevenin-Norton and series-parallel
vider are scaled by the same factor a, how does this affect a +2-mA constant current source and 1-MQ resistors. De- transformations to reduce to a simple two-node circuit the
R4
the currents through them? The voltages across them? The sign a current divider with a + l.5-rnA output. Design an- circuit shown. Which variables from the original circuit
+ power? Use this result to design a current divider using other with a + 1.51-rnA output. remain in the simple circuit?
R3
+
a +l-A source with outputs of 0.4,0.3, and 0.1 A and a 2.38. Find the Thevenin equivalent circuit.
+5 V power dissipation of 1 mW. Start by taking GI to be 1 S;
Rz
R]
+
1.5 V
+
} +30V
1 +4.0 V compute the other G' s and scale.
2fl 2fl
+
+
ti in the range 0 to 12 V and Thevenin equivalent impedance
in the range 1 to 10 Q. Design a resistive circuit limiting
(a) Sketch the circuit, labeling currents ii, i2, ... , h and
voltages VI, ... , V7. Define the voltages for each ele-
+ the power supplied to any battery in these ranges to lOW
1 kQ v lQ v ment to satisfy the passive sign convention.
while permitting the maximum number of batteries to be
(b) Write three distinct KVL equations for this circuit.
charged at the same time.
2.60. (a) Find the i-v law V2 = !2(i2) for the circuit
shown in (b). (b) If VI = 3il - sint, nodes a-a' are con-
More Challenging Problems nected together, and nodes b-b' are connected together, find
FIGURE 1'2.45 FIGURE 1'2.49 2.57. Find i and R. il and V2.
4Q
b b'
(a) (b)
FIGURE P2.60
fiGURE P2.62
2.61. A total current of 20 A divides among three resistors
as shown. If the largest is R[ = 1 kn, find R2 and R3 so \
that 12 A flows through R3 and R2 = 2R3.
20A
--7
modest, low-paying . . .
threw himself into. his . ele6trl6al: riesearch
heavy.teaching duties,.andhisc . : in
the misplaced criticisms. ofliis work,durlnghlsiifetimeOiu';l received the fanxe
that was due him. The Royal Spcieiyof Londbn awardedhltn the Copley Medal
in 1841, and the Univ'ersityof'Munichgave him itsc:Professor or Physics .chait
in 1849. He was also honored after his death when ilie ohm was chosen as the
unit of electrical resistance.
III
3.6
3.7
Interconnecting Op Amp
Building Blocks
Practical Op Amps "'+
;,t
(a) (b)
A voltage-controlled current source (VCCS) is controlled by a voltage and a current-
controlled current source (CCCS) by a current. The symbol for a dependent current
source with value (source function) is is shown in Fig. 3.1 (b).
Figure 3.2 illustrates the four types of linear controlled sources and shows the volt-
age or current on which they are dependent. The quantities fL and f3 are dimensionless
3.3 Operational Amplifiers II 3.B Design of Simple Op Amp constants, commonly referred to as the voltage gain and current gain, respectively. The
II
Circuits FIGURE 3.1 constants rand g have dimensions of ohms and siemens, units of resistance and conduc-
III 3.4 Role of Negative Feedback (a) Dependent voltage tance, and are thus called the transresistance and transconductance, respectively. The
3.5 Op Amp Building Block II Summary source; (b) dependent prefix trans is to remind us that the current and voltage are not measured at the same
III
Circuits III Problems current source. location. These four parameters each scale the ratio of the controlled variable produced
per unit controlling variable.
0-------0
+ + +
The voltage and current sources of Chapters 1 and 2 are independent sources, as defined
in Section 1.4. We may also have dependent sources, which are very important in circuit
theory, particularly for understanding electronic circuits (circuits containing elements
such as transistors or vacuum tubes that need to be coupled to power supplies). In this (a) (b)
chapter we shall define dependent sources and consider an additional circuit element, the
operational amplifier or op amp. Dependent sources are central to the design of electronic ~l
amplifiers and to a wide variety of other circuits of great practical interest. Op amps 0-------0
behave like dependent sources, and op amps are a convenient way to fill the requirement +
for dependent sources in these circuits, and the design of simple op amp circuits.
We shall first analyze a few simple circuits containing resistors and sources, both
independent and dependent. As we shall see, the analysis is very similar to that performed
in Chapter 2. We shall then turn our attention to the op amp and its behavior as a linear
circuit element. Ideal and nonideal op amp models will be presented and the uses of these
models in analyzing op amp circuits explored. A set of op amp building block circuits (e) (d)
will be introduced, easily analyzed circuits from which complex circuit behaviors can
FIGURE 3.2 (a) VCVS; (b) CCVS; (c) VCCS; (d) CCCS.
be built by combination. The chapter concludes with a look at practical op amps in the
real world of physical packages, finite limitations, and environmental constraints, and the Nonlinear dependent sources are those in which the controlled variable is not simply
design of simple op amp circuits. proportional to the controlling variable as in the linear dependent sources above. In
the present book we are interested only in studying linear circuits, and thus nonlinear
controlled sources will not be considered further.
As an example of a circuit containing a dependent source, in Fig. 3.3 we have an
independent source, a dependent source, and two resistors. The dependent source is a
CCVS with controlling current it and transresistance r = 0.5 Q. Note that if it were
A dependent or controlled voltage source is a voltage source whose terminal voltage redefined to be the current flowing downward through the rightmost arm of this circuit
depends on, or is controlled by, a voltage or a current defined at some other location rather than the middle arm, the controlled source would then be an element whose voltage
in the circuit. Controlled voltage sources are categorized by the type of controlling was 0.5 times its own current, and thus it would be identical to an ordinary O.5-Q resistor.
variable. A voltage-controlled voltage source (VCVS) is controlled by a voltage and a This illustrates that only if the controlling variable is located elsewhere in the circuit does
current-controlled voltage source (CCVS) by a current. The symbol for a dependent the element truly have to be represented as a controlled source.
voltage source with source function Vs is shown in Fig. 3.I(a). Dependent sources are essential for producing amplifiers, circuits that produce out-
A dependent or controlled current source is a current source whose current depends puts more powerful than their inputs. They are also integral to active filters (to be studied
on, or is controlled by, a voltage or a current defined at some other location in the circuit. in Chapter 14) and to electronic circuits of all kinds. Among their many important uses,
Find the current i in the circuit of Fig. 3.4. The dependent source
is a VCVS with voltage gain of +3 controlled by the voltage VI ~ ~!;~~e~ sou~ce ha; .the important capability of generating a current or voltage that is
Applying Kirchhoff's voltage law around the circuit, we have this cap~c~ty ::~~on 0ItS. controlling variable. One practical circuit element embodyin~
(3.1) . e operatIOnal amp (op amp), defined as an electronic device that under
ZQ ~~o~er (:lrcu;:tan~es ~ehaves like a voltage-controlled voltage source (VCVS) with very
and by Ohm's law 19 gam. e ClfCUlt symbol fOf the op amp is shown in Fig. 3.5(a). The termin~1
+
VI = 2(-i) = -2i (3.2)
marked by a plus sign in the circuit symbol is called the noninverting input, that by the in general have different gain (ratio of output to input) than the open-loop op amp (no
minus sign the inverting input, and the unmarked terminal is the output. The voltages feedback), whose open-loop gain is simply the VCVS voltage gain A.
at these three nodes, Vii, Vi2, and vo, respectively, are all defined relative to ground, a To use this model, we simply replace the op amp circuit symbol by the model and
node that is usually omitted from the op amp circuit symbol but is part of every op amp apply the methods of Section 3.2 to analyze the resulting circuit.
circuit. The op amp input voltage, or voltage across the op amp input terminals, is
Suppose that we connect an independent source across the input
terminals of an op amp, and a load resistor between output and
Sometimes a fourth wire is added to the circuit symbol as shown in Fig. 3.6 to show ground. The source has a Thevenin equivalent voltage of 1 mV
+
and Thevenin equivalent resistance of 10 kQ. The load resistance is
the connection to the ground node explicitly. We will adopt the three-wire convention of
1 kQ, and the op amp has an open-loop gain of 100,000. Find the
Fig. 3.5(a) when drawing op amps in circuits.
Since the op amp is an electronic device, a power supply of the right ratings must source current is and load voltage VI.
Examining Fig. 3.8(b), there can be no current (is = 0 A)
be correctly connected to the op amp in order to energize it. This is one of the "proper
through the lO-kQ resistor since it is in series with an open circuit.
circumstances" required in the definition to achieve useful op amp behavior, and these
Then, by KVL, Vin = 0.001 V. Thus the voltage of the VCVS is
considerations will be briefly discussed in Section 3.7. Although well-behaved op amp
fiGURE 3.6 Alternative 100,000 x 0.001 = 100 V, which by KVL is also VI. We see that
circuits cannot be physically built and operated without due consideration of how these
circuit symbol for the op the source voltage has been amplified by the open-loop gain and
proper circumstances can be assured, the behavior of op amps as linear circuit elements
amp. may be analyzed assuming that this has been done. A satisfactory discussion of these transferred across to the load. Note also that the power delivered to
requirements and how to meet them must be deferred to an electronics course, which often the load did not come from the independent source, which produces
follows basic circuits courses such as this. For the present purpose of understanding how no power since is = 0 A. It must have come from the op amp or,
op amps behave in linear circuits, we will assume that the proper environment for this more specifically, from the power supply energizing the op amp,
behavior has been established in all op amp networks we will analyze. In particular, since
even though this necessary source of power is not shown explicitly.
r~- _________ _
power supplies help establish these circumstances but otherwise do not influence linear lOkS"l , lOkS"l
op amp behavior, they will generally be omitted from op amp circuit diagrams.
Under the proper operating conditions, an op amp presumably behaves like a VCVS ,,
with high gain. The simplest circuit model for an op amp, called the ideal voltage ,,
,
amplifier model and shown in Fig. 3.7, consists of nothing more than a VCVS and some ...: _________ .J
op amp. This nomenclature is related to the concept of feedback discussed in the next (a) (b)
section. As we shall see, op amps are frequently used in configurations in which the
op amp output is connected back to its input. Such circuits are called closed loop and FIGURE 3.8 Circuit for Example 3.2: (a) circuit modules; (b) ideal voltage
amp model substitution.
:........... Reflection upon the ideal voltage amplifier model of Fig. 3.7 reveals some apparent
:, - ~, - ...... shortcomings. First, the model predicts that an op amp draws no current at all from the
+ 0--+-0
Vjo :
~nput source. (the.input terminals are open circuited) and thus no power, yet performs its
- 0----'--0
I __ _
+ Job of amplIficatlOn. But basic physics suggests that any amplifier must draw at least
,
:...........
\ a little p?w~r from its sour~e in order to sense the signal it is supposed to amplify.
CommulllcatlOn always reqUIres some minimal expenditure of power by the source in
order to transmit the message, yet the model predicts that no power is needed. Second,
an equally questionable prediction is apparent on the output side of the model. If a load
resistor is applied across the output, by decreasing its resistance further and further the
fiGURE 3.7 Ideal voltage amplifier op amp circuit model. current and thus the power delivered to the load by the op amp may be made arbitrarily
large (the load voltage is fixed by the input voltage and open-loop gain, while power to
capability. +
An improved op amp model, which remedies these shortcomings by the addition Vin IMQ
of two resistors to the ideal voltage amplifier model, is shown in Fig. 3.9. Ri is called +
VI 60kQ
the input resistance and Ro the output resistance. The ideal voltage amplifier model may IMQ
+
be seen as a special case of the improved model with Ri = infinity and Ro = O. By IDS llin ll~
requiring these values to be both nonzero and finite in the improved model, we guarantee
that some power will be expended by the source and that a finite maximum current and :b
power will be developed across the load, no matter how small the load resistance is. (a) (b)
Typical values for common op amps are Ri = 1 MQ and Ro = 30 Q. Just as with the
previous model, the improved model is used by substituting it for the op amp circuit FIGURE 3.10 (a) Circuit for Example 3.3; (b) after replacement of the op
symbol in the circuit to be analyzed. amp by the improved op amp model.
20kQ I MQ I MQ
2
-VI
3
Inverting
input
EXERCISE 3.3.1
~~ ____~____~______~______________~~______~~~Eo
3.3.2. Repeat Exercise 3.3.1 with the improved op amp model using
R; =1 MQ, Ro = 30 Q. Are the voltages within 1% of Exercise 3.3.1?
Within 0.1%?
Answer 0.200 V; 2.00 V; yes; yes
Noninverting 3.3.3. Using the circuit shown in Exercise 3.31 and the improved op
input 30pF amp model of Exercise 3.3.2, suppose that a load resistor RI was connected
across Vb. For what value of RI would the current delivered to RI be a
R5 39k!:2
maximum? What would that maximum current be?
Answer 0 Q; Is A
3.3.4. Redo Example 3.3 using the ideal voltage amplifier model for the
op amp. Do the results agree with the example in which the improved op
amp model was used?
Answer V2 = ~VI; yes
Typical op amps may have open-loop gains in the range of 105 to 106 or more, yet
relatively few circuits require such large voltage amplification. Why is the op amp
Offset null
required to behave like a VCVS of very high gain, since this gain is seldom used to create
fiGURE 3.12 741 op amp. (Courtesy of Fairchild Semiconductor Corp.) such extreme voltage amplification? To understand how open-loop gain can be converted
'I
I
88 Chapter 3 Dependent Sources and Op Amps Section 3.4 Role of Negative Feedback 89
to other benefits no less important to successful circuit design than raw amplification, we would be unaffected (it is determined solely by VI). Since Vin = Vii - Vi2, the increase
will consider the concept of negative feedback. in the op amp output V2 will indeed result in an algebraic decrease in the op amp input
Negative feedback is said to exist within an electrical circuit, or dynamic system of Vin, the definitive sign of negative feedback.
any kind, if any change of the output has the algebraically opposite effect on the input. In Next we compare the behavior summarized in (3.6) with the op amp circuits of
a negative feedback system, an increase in the algebraic value of the output (more positive Section 3.3 in which negative feedback was not applied. When the voltage VI was applied
or less negative) yields a decrease in the algebraic value of the input (less positive or directly across the op amp input terminals with no feedback, the op amp output voltage V2
more negative), and vice versa. Consider an athlete running a marathon. Define as input was AVI. With negative feedback the ratio of output V2 to circuit input VI, or the voltage
the concentration of oxygen in the blood and as output the athlete's resulting running transfer ratio, has dropped from A to Aj(A + 1). For A = 100,000, this is a drop from a
speed. If at some point it should be decided to pick up the pace (increased output), voltage transfer ratio of 100,000 to just a bit less than unity. Indeed, for any reasonably
oxygen will be consumed more rapidly and blood oxygen levels will decrease in turn large value of A, the voltage transfer ratio is quite close to unity, whence the name
(decreased input). If, instead, the pace is slowed, less oxygen is burned per unit time voltage follower (the output V2 "follows" or matches the input VI without amplification
and its blood concentration will increase (increased input) due to the decreased pace or any other changes).
(decreased output). This is an example of a negative feedback system. Changes in the \ By making the negative feedback connection, it appears that we have thrown away
output of a negative feedback system lead to opposite input changes, which tend to return the A = 100,000 open-loop op amp gain, instead settling for a voltage transfer ratio
the output toward its initial value. By this process, negative feedback systems tend to (with feedback) of approximately unity. What have we gained in compensation for the
resist change and thereby to stabilize their own outputs. tremendous reduction in available amplification? Suppose that over time, as the circuit
Negative feedback may be achieved in an op amp circuit by supplying a current ages, the open-loop gain of our op amp declines 10%, from 100,000 to 90,000. In the
path between the output terminal and the inverting input terminal. Consider the circuit circuit lacking negative feedback, the voltage transfer ratio will change between the same
of Fig. 3.13(a), called a voltage follower circuit for reasons that will soon be apparent. two values, or change 1 part in 10. In our voltage follower, Aj(A + I) will go from
Replacing the op amp by its ideal voltage amplifier model as in Fig. 3.8(b), KVL around 0.999990 to 0.999989 as A declines from 100,000 to 90,000, a change of I part in 106 .
the left loop yields Thus, at the cost of a five-order-of-magnitude reduction in amplification, we have gained
a five-order-of-magnitude increase in the constancy (decrease in the sensitivity) of the
(3.5) circuit's amplification factor.
Constancy of gain is essential in many applications and useful in most others. Using
or Vin = vJ/(A + 1). The right loop shows that V2 = AVin. Eliminating Vin from these modern integrated-circuit technology, it is easy to build an op amp with high open-loop
last two equations and solving for V2 yields the result gain and do so inexpensively, but very difficult to regulate the value of the gain with
high precision. Suppose that two op amps coming off the production line have open-loop
A gains that differ by 10%. If otherwise identical circuits in which these op amps are used
V2= - - V I (3.6)
A+l have outputs that differ by 10% also, the performance of the resulting electronic end
product could not be guaranteed to any degree of precision. By using op amps only in
First let us verify that this is truly a negative feedback circuit. If the op amp output
stabilizing negative feedback configurations, the copy-to-copy end circuit variations can
voltage V2 were somehow to increase slightly, the voltage at the inverting input terminal
be kept very low even when the copy-to-copy op amp variations are rather high. Other
Vi2 would be increased by the same amount, while the voltage at the noninverting input
gain variations compensated by negative feedback include those due to temperature, aging
of other circuit elements connected to the op amp, and variable load resistance. Negative
feedback is also essential to decrease undesirable variations of gain with frequency, that
is, increase the stage bandwidth, as discussed in Chapter 14.
In summary, using negative feedback trades raw gain for constancy of gain. The
resulting circuits are highly insensitive to the numerical value of the open-loop op amp
gain A, provided only that A can be counted on to be reasonably large. The voltage
transfer ratios across these circuits may then be set by selecting the circuit type and the
values of any external elements (e.g., resistors as in Fig. 3.15), matters much more easily
controlled by the circuit designer than the open-loop op amp gains. Negative feedback
configurations are sometimes referred to as closed-loop circuits, since the feedforward
(al (b)
path through the op amp and the feedback path together form a closed loop. Because of
fiGURE 3.13 Voltage follower: (a) circuit diagram; (b) using ideal voltage its extreme usefulness in creating reliable circuit designs, all op amp circuits analyzed in
amplifier model. the remainder of this book are of the negative feedback type.
90 Chapter 3 Dependent Sources and Op Amps Section 3.4 Role of Negative Feedback 91
,------1+ 3.4.1. Determine the voltage transfer ratio V2/V1 of a voltage follower
using the improved op amp model with A = 100,000, R; = 1 MQ, and +
+
Ro = 30 Q and with a load resistor of R1 = 100 Q connected across V2. Is
it within 1% of that computed using the ideal voltage amplifier model? V2
V2
Answer 1.00; yes
50kQ
3.4.2. Determine the voltage transfer ratio V2/V1. Use the ideal voltage
+ amplifier model.
Answer 2A/(A + 2), or in the high open-loop gain limit, V2/ V 1 Voltage transfer equation: V2 = -RF/RAVI (3.8)
EXERCISE 3.4.2 =2
FIGURE 3.15 Inverting amplifier.
\
+ +
In this section we present several simple op amp circuits that are frequently used as
modules or building blocks in the design of more complex circuits. Each building block
+
performs one basic operation, such as the amplification of its input or the addition of two
or more inputs. By interconnecting these building blocks, more complex results may be
built up, in the same way that an arbitrarily complicated mathematical expression can be +
evaluated by the simple step-by-step application of add, subtract, multiply, and divide (a) (b)
operations or a wall built by properly stacking building blocks made of brick or stone.
The first building block circuit is the voltage follower shown in Fig. 3.14. For this FIGURE 3.16 Analyzing the inverting amplifier: (a) using the ideal voltage
and the other building blocks to be described in this section, what the circuit "does" is amplifier model; (b) redrawn for clarity.
summarized in its voltage transfer equation. The voltage transfer equation specijzes the
output voltage in terms of the input voltage (or voltages) when the output port is open in Fig. 3.14) are defined in terms of the input to and output from the circuit, not the input
circuited. The voltage-follower circuit was analyzed in Section 3.4, where the voltage to the op amp that is contained in the circuit.
transfer equation was found to be simply V2 = VI. Thus what the voltage follower does The voltage transfer equation for the voltage follower was derived using the ideal
is produce an output voltage that equals (follows) its input voltage. This may seem like voltage amplifier model. The same result may also be derived using the improved op
a rather modest achievement, but as we will see in Section 3.6, voltage followers are of amp model of Fig. 3.9, as shown in Exercise 3.5.1. For each building block, use of the
the utmost practical benefit. Note that the input voltage to this circuit, VI, should not ideal voltage amplifier model and use of the improved op amp model leads to the same
be confused with the input voltage to the op amp, Vin. The voltage transfer equation voltage transfer equation in the limit as A goes to infinity (which we use to define the
and corresponding voltage transfer ratio (ratio of output voltage to input voltage, V2/ V I voltage transfer function since we know that A will be very high for practical op amps).
Because analysis is a bit more laborious using the improved op amp model with its two
additional resistors, but the results do not differ in the high-op-amp-gain limit from those
~I
0
+ 0 derived using the simpler ideal voltage amplifier model, we will use the latter to study
+ each building block circuit.
Vj The second building block is the inverting amplifier of Fig. 3.15. Note that this
V2
circuit incorporates negative feedback, as evidenced by the current path from the output
back to the inverting input. Indeed, all the building blocks will incorporate negative
0 0
feedback. Replacing the op amp by its ideal voltage amplifier model results in the circuit
Voltage transfer equation: V2 = VI
of Fig. 3.16(a).
(3.7)
Examining Fig. 3.16(b), we see that
fiGURE 3.14 Voltage follower.
I
I 92 Chapter 3 Dependent Sources and Op Amps
Section 3.5 Op Amp Building Block Circuits 93
RI
Solving for i and using the fact that V2 = AVin yields
RF
-(A + 1) -(A + 1) V2
i = Vin = ---- (3.9) RT RF
RF RF A
VI +
Applying KVL around the outer loop gives
V2 = VI - (RA + RF)i (3.10)
Allin tJo
V2 = VI + RA + RF -A -+V12 (3.11)
RF A
Gathering V2 terms and solving for V2 gives the result (a) (b)
-RF \ FIGURE 3.18 Analyzing the inverting summer: (a) using ideal voltage
V2 = VI (3.12)
RA + I/A(RA + R F ) amplifier model; (b) after Thevenin transformation.
In the high-op-amp-gain limit as A becomes arbitrarily large, we have the voltage transfer
equation for the inverting amplifier: In Fig. 3.18(b),
-RF
V2 = --VI (3.13) (3.15)
RA
From (3.13) we see that the voltage transfer ratio is always negative for the inverting
amplifier, whence its name. This building block circuit functions to create an inverted,
scaled copy of the input with the scale (amplification) factor set by the designer's choice
VT = RT(~
R,
+~R2 +~)
R3
(3.l6)
of external resistances RF and R A . Just as with all other building blocks, the voltage To fix ideas, exactly three inputs have been shown in Fig. 3.18(a) and used in the
transfer equation shows no dependence on open-loop gain A, a consequence of our calculations above. Figure 3.18(b) contains a single-input inverting amplifier with input
insistence on negative feedback. voltage VT and voltage transfer ratio - RF / R T . Applying voltage transfer equation (3.13)
The next building block circuit is the inverting summer of Fig. 3.17. It differs from for this circuit with input VI = VT and input resistance RA = RT yields the voltage transfer
the inverting amplifier by having more than one input. To analyze this circuit, we first equation for the three-input case, or in general (by adding the ellipsis, " ... ") the voltage
convert it to an equivalent inverting amplifier and then use the results just derived for this transfer equation (3.14). The inverting summer output is the sum of its inverted and
circuit. The Thevenin equivalent of the circuit to the left of the dashed line in Fig. 3.18(a) scaled inputs.
is conveniently evaluated by first replacing each series combination of voltage source and If it is desired to amplify without sign change, the noninverting amplifier shown
resistor by its Norton equivalent and then adding current sources and combining parallel in Fig. 3.19 may be used. Replacing the op amp with its ideal voltage amplifier model
resistors. The result is shown in Fig. 3.18(b). results in Fig. 3.20. Examining the output circuit, RF and RA are in series across V2, or
Voltage. transferequatiQn:
. .. . .....
Vo=
-Rp . .. --:-RF -:-RFi .
- ..-Vl + -.-. V2 + -.-V3 + ...
R l R2 R3 ..
(3.14) Voltage transfer equation: V2 = (1 +~: ) VI (3.17)
(3.19a) with VT given by (3.16). While (3.16) is specific for three inputs, the ellipsis .. in
Substituting V2/ A for Vin and (3.18) for vRI gives (3.20) once again indicates the immediate generalization to any number of inputs VI. V2,
... , This building block has the effect of scaling and summing its inputs (without sign
V2 RA
VI = - + V2 (3.19b) inversion).
A RA +RF These five circuits constitute the bulk of the building blocks we will use. One
The first term goes to zero in the high-op-amp-gain limit, leaving the voltage transfer additional one will be introduced later. It is somewhat surprising, given the simplicity
equation (3.17). For instance, RF = 10 kQ, RA = 5 kQ results in a voltage transfer of these circuits, that they are quite sufficient to build up many circuits of considerable
ratio of +3. Note that whereas the inverting amplifier has no restriction on voltage practical significance. We need only consider how they may be interconnected, as in
transfer ratio magnitude, the noninverting voltage transfer ratio must be greater that unity Section 3.6, and allow inductors and capacitors in the building blocks to be ready to
(for passive resistors with nonnegative resistance). To achieve values less than unity, a understand and to design many interesting circuits such as filters and analog computers.
voltage should first be divided (see Exercise 3.5.2) before being input to a noninverting
amplifier.
The noninverting summer of Fig. 3.21 may be analyzed exactly as the previous
(inverting) summer. First convert to an equivalent single-input circuit by Thevenin
EXERCISES
Rp 3.5.1. Using the improved op amp model (Fig. 3.9) with Ri = I MQ and
Ro = 30 Q, first determine the output V2 of the voltage follower (Fig. 3.14)
RI in terms of op amp gain A and then in the high-op-amp-gain limit. Does
the voltage transfer equation agree with that calculated in (3.7)?
+ +
+ A + 0.00003
Answer A + 1.00003 VI; VI; yes
VI R2 VR, 3.5.2. This exercise shows how to get a positive voltage transfer ratio
Vo
of less than 1. Using the ideal op amp model, determine the voltage VR2
FIGURE 3.20 Substituting the ideal voltage amplifier model.
Then determine vo, noting that VR2 is the input to the noninverting amplifier.
Specify values for the resistances yielding an overall voltage transfer ratio
EXERCISE 3.5.2 of ~ (answer not unique).
(3.20a)
NG<SlOCKS
where
Op amp building blocks are simple modules that may be connected to produce more
C3.20b) complex circuit behaviors. In this section we consider one particularly useful way to
interconnect these modules and how to predict the behavior of the interconnected circuit
fiGURE 3.21 Noninverting summer.
from that of the modules that are its parts.
96 Chapter 3 Dependent Sources and Op Amps Section 3.6 Interconnecting Op Amp Building Blocks 97
A port in a circuit is a pair of wires to which another subcircuit may be attached. Op IOOkQ
amp building block circuits each have one output port and at least one input port. Circuits
containing a single input and a single output port, for example the voltage follower and
inverting and noninverting amplifiers of Section 3.5, are called two-port circuits. Those
with additional ports, such as the inverting and noninverting summers, are n-port circuits.
The most direct way to connect a pair of two-port circuits is to tie the output port
of one to the input port of the other. This is called the cascade interconnection of two-
ports. Cascading two (or more) two-ports results in an interconnected circuit that is also
a two-port, as shown in Fig. 3.22. Here VI is the input voltage to the overall cascaded
circuit and V3 the output voltage.
FIGURE 3.23 Circuit for Example 3.4.
Suppose that i2 is not equal to zero in Fig. 3.22. The ideal voltage amplifier model
of the op amp predicts that the output voltage of each building block must be independent
of i2 since it is just the voltage across a VCVS (see Fig. 3.7), whose source function
does not depend on the output current i2. So even though open-circuit output is assumed
in its definition, the voltage transfer equation continues to apply in computing the output
voltage for a building block circuit even when the output is not open circuited. This
analysis using the ideal voltage amplifier model suggests that connecting a load, such as
FIGURE 3.22 Cascade interconnection of two op amp building block units. another op amp building block, to the output port of a previous building block will not
affect its output voltage.
Every op amp building block is characterized by its voltage transfer equation, Thus for building blocks in cascade, the overall voltage transfer ratio is simply
which specifies the output voltage as a function of the input voltage or voltages under the product of the individual voltage transfer ratios, provided only that the ideal voltage
the condition that the output port is open circuited. When op amp building blocks are amplifier model of the op amps may be accurately used. The same principle applies to
interconnected, it is the voltage transfer function across the entire interconnected circuit n-port op amp building blocks as well, as illustrated in the next example.
that describes its behavior. Thus for two-port op amp building blocks in cascade, we
will determine the overall voltage transfer equation. Find the voltage transfer equation for the circuit of Fig. 3.24. Use
Referring to Fig. 3.22, take the output port (V3) to be open circuited as required the ideal voltage amplifier model for the op amps.
by definition of the voltage transfer equation. Applying the voltage transfer equation for
building block B, V3 = KBV2, where KB is the voltage transfer ratio. Then, to find the
lOOkQ
voltage transfer equation V3 = KcvI for the overall cascade, we need only express V2 in
terms of VI. If i2 = 0, the output of building block A is also open circuited, and V2 is
+
found by applying A's voltage transfer equation V2 = K A VI. The overall result is then
V3 = KAKBvI; that is, Kc = KAK B .
Find the voltage transfer equation (input VI, output V3) for the circuit
of Fig. 3.23. Use the ideal voltage amplifier model for the op amps.
The rightmost building block circuit is a noninverting amplifier
with V3 = (1 + RF / RA)V2 or V3 = 3V2. i2 = 0 since the the ideal
FIGURE 3.24 Circuit for Example 3.5.
voltage amplifier model for its op amp has an open-circuited input.
The leftmost building block is an inverting amplifier with voltage
The rightmost building block is a two-input noninverting sum-
transfer ratio - R F / R A = -4. Since its output port is open circuited
mer with input voltages V2 and V3 and output Va. Using the voltage
(i2 = 0), V2 may be exactly evaluated using the voltage transfer
transfer equation (3.17) for this building block, we have
equation for this building block, V2 = -4vI. Then, combining,
V3 = 3(-4vd = -12vI. The voltage transfer ratio for the cascade
is the product of the individual voltage transfer ratios. Va = (1 + lOOK) (
20K (60KI130K) 60K + 30K
V2 V3 )
98 Chapter 3 Dependent Sources and Op Amps Section 3.6 Interconnecting Op Amp Building Blocks 99
In summary, to a good approximation a cascade of op amp building blocks has
where K stands for multiplication by 103 and II is the parallel equiv- a voltage transfer equation that may be computed simply by combining the individual
alent operator. voltage transfer equations. For two-ports in cascade, the overall voltage transfer ratio is
1 the product of the constituent two-port voltage transfer ratios. Using this combining rule,
60KI130K = 1/60K + 1/30K = 20K circuits involving multiple building blocks may be designed to accomplish relatively
complex goals and yet be analyzed easily by multiplying voltage transfer ratios. The
Then range and versatility of the building block approach to op amp circuit design, suggested
Va = 2V2 + 4V3 (3.22) by the results of this section, will be fully realized with the introduction of capacitors
and inductors into the building blocks in Chapter 7.
The remaining building block is an inverting amplifier with
-40K
V3 = --VI = -VI
40K EXERCISES
Combining these two results, the overall voltage transfer equation is \
3.6.1. Consider a source with Thevenin equivalent resistance RT and
(3.23) Thevenin equivalent voltage VT. If a load of RL ohms is placed across
the source terminals as in part (a) of the figure, (a) what is the open-circuit
By combining building blocks of different types, we may achieve results beyond load voltage (voltage across RL when RL is infinite)? (b) Find the value
what would be possible using anyone alone. This example shows how a difference, as for resistance RL at which loading has reduced VL to half of its maximum
(open circuit) value. (c) If a voltage follower is inserted between source
opposed to an inverting or noninverting sum, of voltages may be generated by combining
and load as shown in part (b) of the figure, show that loading is eliminated.
inverting and noninverting building blocks. Buffering of a source from a load to eliminate loading is one of the main
The main result of this section, that when op amp building blocks are cascaded
uses of a voltage follower.
the overall voltage transfer equation may be found simply by combining the voltage Answer VT, RT, VL (no buffer) = (RL+ VT)/(RL +RT), but VL
transfer equations of each individual building block, was justified using the ideal voltage (with buffer) = VT. Loading reduces the output voltage by a factor equal
amplifier model for each op amp. The result followed from the lack of dependence to the voltage-divider ratio.
of the output voltage on the output current. The improved op amp model of Fig. 3.9
contains an additional resistor Ra in its output loop. The current produced by its VCVS, Source Load Source Voltage follows Load
which supplies both the output current (i2 in Fig. 3.22) and the current to the feedback
path present in every op amp building block, must all pass through Ra By KVL, the
improved op amp model thus predicts that the output voltage will be reduced as the
output current increases (their sum equals AVin, which does not depend on this current).
The two models seem to disagree as to whether the output voltage will remain constant,
as the ideal voltage amplifier model predicts, or will in fact decrease as the output current
increases as the improved model suggests.
Which model gives the right prediction? As might be anticipated in calling it an
"improved" op amp model, direct measurement verifies that as the output current of any
building block circuit increases, its output voltage declines as predicted by the improved (b)
(a)
model. The decline of load (output) voltage with load current is called loading and is
quite a general phenomenon, not restricted to op amp building blocks. As explored in EXERCISE 3.6.1 Loading with and without a buffer between source and
Exercise 3.6.1, loading is the consequence of voltage division between internal resistance load: (a) direct connection; (b) buffered connection using a voltage fol-
and load resistance. Some degree of loading must occur whenever a source has nonzero lower.
Thevenin equivalent resistance and is connected across a finite load resistance.
Must we then abandon our main results, derived using the ideal voltage amplifier 3.6.2. Design a circuit with the voltage transfer equation Va = 8vI -
model, which has no internal resistance and hence manifests no loading? For sufficiently 2V2 - V3. Use only inverting building blocks and select all resistances to be
low levels of output current the degree of loading, that is, reduction of output voltage, in the range 5 to 500 kQ.
Answer Express Va as Va = -2(-4vI) - 2V2 - V3 Get the term
is negligible and our result holds with excellent accuracy. Current levels may be kept
in parentheses as the output of an inverting amplifier with input VI and
low by keeping the resistances in the building block circuits high, which is why values
voltage transfer ratio -4. Then route this voltage, along with the voltages
of 5 kQ and above are used in the examples and problems.
+
+
EXERCISE 3.6.2
EXERCISE 3.6.3
Va = - iL2c Vc - iL2b Vb
I---~~;--------------------~
In Chapter 2 we considered strictly resistive circuits, those containing only resistors and Rn
+ :--------------------:
independent sources. In the design section of that chapter it was noted that the versatility
: RF1 :
of resistive circuits, their range of useful applications, is limited. The addition of just one
more element, the op amp, greatly expands the range of potential applications, as we shall + +
see below. The examples presented here point to the design of electrocardiogram am-
plifiers for biomedical application, analog-to-digital converters in computer engineering, a
and an averaging circuit for signal processing applications. Many more examples could
be offered as well, but it is hoped that these are persuasive of the range and versatility
of the op amp in practical applications.
The design examples here illustrate uses of op amp circuits constructed using the
straightforward building block cascade approach described in Sections 3.5 and 3.6. Other Inverting Inverting
interconnections of op amp stages will be considered in later chapters, and elements other ,< _ _ _ _ _ _ _amplifier, summer
_ _ _ _ _ _ _ _ _ _ _ _ _ L __________________________
,
I
than op amps and resistors, which will further enhance the variety of electronic circuits
that can be designed. FIGURE 3.26 ECG amplifier.
IL2"
RF2
= -RI2 = 500
Va = (t 2+
k
k=O 255
1
Vk) - 10 (3.25)
we select RI2 = I kQ and RF2 = 500 kQ. Now we may divide the This can be realized with a nine-input noninverting summer as shown
desired gain IL I IL2c = 500 between the two stages any way we wish. in Fig. 3.27. Examining the form of the voltage transfer equation
It is helpful to balance the gains so we set p. I = IL2c = J500. Then for this building block which is given in Fig. 3.21, in order that the
since gains on the eight successive single-bit inputs V7, V6, ... , Vo each be
in the ratio of 2, we must have R6 = 2R7 , Rs = 2R6 = 4R7 , and so
~ Rn 5 x lOs
IL?c = ..;500 = - = --- on. Then with G T = l/R T ,
- R22 R22
GT = G 7 + G6 + ... + Go + G B
we have that R22 = 22.4 kQ. Finally, /11 = RFI/RIII = J500,
so selecting RFJ = 100 kQ, we have fixed the value of the final = G 7 (1 + ~ + ~ + ... + 1~8) + G B
resistance:
255G7
Rill =
R"I
~ = 4.47 kQ
= 128+ GB
..;500 Now to keep the numbers simple let us select G B = 129/128 G 7 so
which completes the design. that
As with almost all designs, this solution is not unique. Indeed,
more attractive solutions can be found, solutions that usc only a sin- GT = 3G 7
gle op amp, have less sensitivity to parameters, arc easier to adjust, Then RT = ~R7' and with the resistive ratios already defined we
or possess a broader range of operating frequencies. But the circuit have, by the voltage transfer equation in Fig. 3.21,
shown in Fig. 3.26 is quite capable of satisfying the simple ECG
amplifier design specifications as stated. Va = (1 + ~:) (~) C~8 + ~~ + ... + ~7 + l~~~B) (3.26)
Our goal is to design a digital-to-analog converter (DAC'). Assume Comparing (3.26) and (3.25), the Vo terms will be identical if
3.2
that we have an 8-bit digital input signal with transistor-to-transistor
(TTL) logic levels +5 V and 0 V, and that the full-scale analog
output should range from -10 V when the digital input is all zero~
(1 + ~:) (~) C~8) = 2~5
(0 V at each input pin, digital value 0000000( 2 ) to 0 V for the highest
Solving the last for Rp yields
input value (+5 V at each input pin, digital value III I I I 11 2 ). Rp = 2.01RA
Since each pin represents a power of 2, the formula hy which
a digital value is converted to its analog value will he Thus in Fig. 3.27 we set RA = 10 kQ, Rp = 20.1 kQ. Since the
Vo terms in (3.25) and (3.26) are now identical and the ratios of
these terms to the VI, ... , V7 terms in the two equations are iden-
(3.24)
tical, all eight of these terms must agree in the two equations. In
Fig. 3.27 we show the bias voltage being produced by voltage divi-
where VB is a bias voltage that sets the zero output value and S i~ a sion of a -15-V source (this is a common power supply voltage).
scale factor. In our case we wish Vo = VI = ... = 117 = () V to illap Finally, each of the conductances Go, Gl,"" G6, and G B have
into Va = -10 V, or been determined only as ratios compared to G 7 . We may choose
any convenient value for G 7 (or R 7 ) and the rest follow. If we se-
-10 = VII
lect R7 = 1 kQ, then R6 = 2 kQ, Rs = 4 kQ, ... , Ro = 128 kQ,
so VB = -10 V. We also want Vo = VI = ... = V7 = +5 V to Illap and RB = 128/129 R7 = 992 Q.
106 Chapter 3 Dependent Sources and Op Amps Section 3.8 Design of Simple Op Amp Circuits 107
R7 IOkO
+
V7 R6 IOkO lOkO
+v6
R5 Rp
O---------------------~------------~--o
Va = (1 + RF)
RA N
~
(VI + V2 + ... + VN) 5V
20kO
Buzzer
Note that for any finite values for RF and R A, the gains on each
input signal will be too large, each by the same factor. An easy
way to remedy this problem is to set RF and RA equal, so that EXERCISE 3.8.1
108 Chapter 3 Dependent Sources and Op Amps Section 3.8 Design of Simple Op Amp Circuits 109
3.8.2. Produce a DAC circuit with two outputs, each the analog-converted
value of one "nibble," the top or bottom 4 bits of the 8-bit input. Assume Negative feedback is necessary for useful op amp behavior in all linear circuit applica-
TTL values and scale each of the two analog outputs to the inverted single- tions, such as the op amp building block circuits: inverting and noninverting amplifiers
ended range 0 V (all 4 bits 0 V) to -75 V (all 4 bits +5 V). and summers, and the voltage follower.
One of many possible solutions The voltage transfer equation is a linear equation relating the input voltage or voltages
of an op amp circuit to its output voltage. The output-to-input voltage ratio is called
80kQ the voltage transfer ratio.
+
vo<J----~I/\llr~
80kQ
40kQ The voltage transfer ratio of two op amp building blocks connected in cascade is the
+0------'1,
VI product of their individual voltage transfer ratios.
The circuit diagrams for physical op amps are complicated, containing many transistors
IOkQ
+0------'1/1,
V3
and other elements, But when properly configured, op amps can be modeled very simply
and are an essential element in modem circuit design.
80kQ
+0--_..11/1
V4
+ PROBLEMS
IOkQ 3.1. Draw the circuit symbols for two distinct controlled 3.6. Find i.
+o-----'lIVlr-
V7 sources, each of which is equivalent to a l-kQ resistor.
3.2. Give examples of an active controlled source and a 2Q
passive controlled source (see Problem 3.1).
EXERCISE 3.8.2 va is the least significant bit, v is the most significant 3.3. For what value of transconductance of the dependent
bit of the input byte. source in Problem 3.8 does a mathematical inconsistency
arise? 42 V 8Q
3.4. Find VI and the power delivered to the 10- Q resistor.
fiGURE P3.6
Dependent sources, those whose source functions depend on the value of a current or
voltage elsewhere in the circuit, are used to model the behavior of active devices such as 30V 3V 3.7. Find VI.
transistors and op amps. An op amp is a device which, under proper conditions, behaves
like a voltage-controlled voltage source with very high gain. Op amps are among the
most versatile of active devices and will be treated as a basic circuit element for the
remainder of this book.
fiGURE P3.4 sin 2tV
Circuits containing dependent sources may be analyzed by treating the sources as if they +
were independent, then substituting the controlling variable into the dependent source 3.5. Find i I . 6Q VI
+ -3 v2 IQ
function.
The ideal voltage amplifier model for the op amp consists in an open-circuit input and _ v2 +
voltage-controlled voltage source output.
2Q
The improved op amp model replaces the open-circuit input with a high but finite 3Q 2Q
(e.g., I MQ) input resistance, and adds a low (e.g., 30 Q) resistance in series in the fiGURE P3.7
output circuit. This model is more accurate but contains more elements.
An op amp is in a negative feedback configuration when an incremental increase in its
output voltage leads to a decrease in its input voltage. 3.8. Which supplies more power, the independent source
I
fiGURE P3.5 or the dependent source?
I ' 110 Chapter 3 Dependent Sources and Op Amps
Problems 111
r~2&
4Q R
this have on the voltage transfer function? On the currents
through the external resistor?
3.23. Compare the power delivered to the 20-krl resistor
in this buffered circuit with that if the voltage follower were
+ 9V
lOY removed (a and b shorted together).
4A lOkQ ~--------------,
1\I\r-.'_Q---i + :b
lOQ
FIGURE 1'3.8 FIGURE 1'3.14
FIGURE 1'3.11 lOV + , ,
1___ _ _ _ _ _ _ _ _ _ __ .J 20kQ
3.15. Find i.
6Q
3.9. Find the Thevenin equivalent circuit.
3.12. We wish to make i = -1 A. Find tW? ~iffere~t
kinds of dependent sources that can do this, specIfymg theIr
FIGURE 1'3.23
2i source functions.
3V 12Q 4Q
i
3.24. If we wish to supply ~ mW of power to the 15-krl
o<~----.----< + - load, RF = ?
+
FIGURE 1'3.15
IOV 3.16. Find Va. Use the ideal op amp model with A
100,000.
0 -_ _ _ _ ...1-_--(- +
~-~---___l+ IV +
14 V 15kQ
+
FIGURE 1'3.9
fiGURE 1'3.12
lOOkQ Va
fiGURE 1'3.24
3.10. What single resistor is this circuit equivalent to?
3.13. Find VI and i I 3.25. Using a voltage divider followed by a noninverting
FIGURE 1'3.16 amplifier, design a circuit with V2 = +~VI. Can we reverse
the order of these two subcircuits and get the same result?
3.17. Repeat Problem 3.16, but use the improved op amp 3.26. Find k in the voltage transfer function V2 = kVI.
+ model with parameters A = 100,000, Ri = 1 Mrl, and
Ro = 30 rl. 50kQ
+ +
+ OV
Design Problems
3.36. An intensive care unit patient monitoring system has
three output voltages: VI (t) measures the puls~ ra~e, V2(t)
the average blood pressure, and V3(t) the respiration ~a.te.
The relationship between these voltages and the quantities
they measure are shown below.
FIGURE P3.27
IS Respiration rate, r
(breaths/min)
3.28. Show how you can create a circuit with V2 = 1000~1 \
(c)
by cascading three identical amplifier stages. Keep all R s FIGURE P3.40
in the range 5 to 500 kQ. FIGURE P3.36c
3.29. Design a circuit with V3 = -2vI - 5V2 Keep all 3.41. Design a circuit satisfying Vs = VI - V2 + V3 - V4.
resistors in the range 5 to 500 kQ. Design a circuit whose output is the alarm voltage Keep all R's in the range 5 to 500 kQ.
3.30. Design a circuit with Vs = -VI - 2V2 - 3 V3 + 4V4 VA = 2b.vI + 3b. v 2 + b.V3, where b.vi is the fractional 3.42. For what range of resistance RI do the values of V2
Keep all resistors in the range 5 to 500 kQ. +3 V deviation of Vi from its nominal value shown. For instance, computed using the ideal voltage amplifier model of Fig 3.7
3.31. Find the voltage transfer function V3 = kl VI + k2 V2 if the patient's pulse rate is 90 bpm, b. VI = (90 - 60) /60 = and the improved model of Fig. 3.9 differ by at least 1%?
+0.5. The inputs to this circuit can be VI, V2, V3, and Use Ri = 1 MQ, Ro = 30 Q.
100kQ 15-V dc sources.
3.37. A photoresistor has a dark resistance of 1 kQ, and 90kQ
its resistance drops 20 Q/lux (a unit of light intensity).
60 Pulse rate, p Design a circuit employing two such photoresistors that will
(beats per min)
output a +5-V signal when the light incident on one, the test
(a) photoresistor, is I lux brighter than the other or reference +
+ photoresistor, and -5 V when the reference photoresistor
FIGURE P3.36a
is 2 lux brighter.
3.38. Design a circuit whose voltage transfer equation is
Vout = 2500VI - 150v2. The voltage gain on any input to
FIGURE P3.42
any single op amp stage should not exceed 8. Use as few
FIGURE P3.31 op amps as you can.
3.43. For what range of resistance RL do the values of V2
3.32. Design a circuit with V3 = 3vI + 2V2 using one More Challenging Problems computed using the ideal voltage amplifier model of Fig. 3.7
noninverting summer. 3.39. Find the Thevenin and Norton equivalents. and the improved model of Fig. 3.9 differ by at least 1%?
3.33. Repeat Problem 3.32 using two inverting stages in- Use Ri = 1 MQ, Ro = 30 Q.
stead of one noninverting. i
<E-- IQ 2Q
3.34. Find the voltage transfer equation (inputs VI, V2; lOOkQ
+
output V3)
40kQ 50kQ
IQ + + 6 cos lOt V
+
100 Pressure
(inches Hg)
(b)
FIGURE P3.39 FIGURE P3.43
FIGURE P3.36b
FIGURE P3.34
i Problems 115
Chapter 3 Dependent Sources and Op Amps
114
117
range of circ.uits than c~u~d be comfortably solved without the level of computational
Chapter Contents support proVIded by a dIgItal computer. Solving, for instance, 10 simultaneous circuit
II 4.7 Duality analysis equa~ions in 10 unknowns definitely loses its charm after the first few hours of
II 4.1 Linearity and Proportionality
Virtual Short Principle for Op hand calculatIOn, and SPICE relieves us of that task.
II 4.2 Superposition II 4.8
Amps
II 4.3 Nodal Analysis
II 4.9 Computer-Aided Circuit
II 4.4 Circuits Containing Voltage Analysis Using SPICE
Sources
II Summary
II 4.5 Mesh Analysis In Chapter 2 we defined a linear resistor as one that satisfied Ohm's law:
II Problems
II 4.6 Circuits Containing Current v = Ri
Sources and we considered circuits that were made up of linear resistors and independent sources.
\
We defined dependent sources in Chapter 3 and analyzed circuits containing both inde-
pend:nt and dependent sources. The dependent sources that we considered all had source
functIOns of the form
y =kx (4.1)
where k is a constant and the variables x and y were circuit variables (voltages or
In Chapter 2 we considered methods of analyzing simple circuits, which we recall are current~). Clearly, ~hm's law is a special case of (4.1). In (4.1) the variable y is
those that may be described by a single equation involving a single unknown current propo~tl?nal to th: varIable x, and the graph of y versus x is a straight line passing through
or voltage. More general circuits must be described by a set of simultaneous equations the ?ngm. For thIS reason, some authors refer to elements that are characterized by (4.1)
involving several unknown circuit variables. as lznear. eler:zents. For our purposes, we shall define a linear element in a more general
In this chapter we meet the challenge of these more general circuits by developing w~y, whIch mcl.udes (4.1) as a special case. If x and y are circuit variables associated
systematic ways of formulating and solving sets of equations that permit complete analysis WIth a two-termmal element, we shall say that the element is linear if multiplying x by a
of any linear circuit. We shall consider two methods, one based primarily on Kirchhoff's constant K results in the multiplication of y by the same constant K. This is called the
current law (the method called nodal analysis) and one on Kirchhoff's voltage law (mesh prop?rti~nality pr~perty and evidently holds for all elements obeying (4.1) since, after
analysis). multlplymg both SIdes by K,
Each of these methods results in a set of equations that contains only a small
subset of all the currents and voltages in the circuit. It should be evident from our (Ky) = k(Kx)
work in previous chapters that a complete analysis of a circuit can be performed by first Thus an element described by (4.1) is linear. In addition, elements with terminal laws of
breaking the circuit, that is, solving for the values of a few key circuit variables. Once the forms
these are known, the rest then follow easily. For example, in a simple circuit consisting
of a single loop, the key variable is the loop current. Once we have broken the
by solving for this current, we may find every voltage in the circuit by applying
y -
_k dx
dt' y= k f x(t)dt (4.2)
individual element laws with known current (and of course, every element current is are also linear, since these laws also imply that
equal to the loop current).
We begin in Sections 4.1 and 4.2 by showing how the principles of (KY)=k(:t KX ), (Ky) = k f (Kx) dt
and superposition can be used to divide a linear circuit problem involving several
The op amp is a multi terminal element and is described by more than one equation.
into component problems each involving a single variable or single source. This
~ow~ver, we shall use the op amp only in a feedback mode, as stated in Chapter 3, and
sometimes be used to make more general circuits simple and thus solvable by
m thIS case the op amp model circuits that result consist of linear elements. In this case
methods introduced previously, which are generally applicable only to simple .
We next develop the methods of nodal analysis and mesh analysis, a pair of tecltlllH:jue~ the. models introduced in Chapter 3, consisting of linear dependent sources and linear
r~slstor~, may be used in place of the op amp. Therefore, we may add the op amp to our
that proves indispensable in the practice of circuit analysis and design. The virtual
hst of lmear elements.
and open principles presented in Section 4.8 facilitate analysis of circuits containing
amps. The chapter concludes with a first look at the computer-based circuit "",,lU'U,uv" . We shall define a linear circuit as any circuit containing nothing but linear elements
and md:pende~t s?urces. As examples, almost all the circuits we have considered thus
and analysis program called SPICE. We will make frequent use of SPICE in the
far are hnear CIrcUlts. Indeed, our attention throughout this book is limited exclusively to
of our study to check our numerical results and to gain hands-on experience with a
(4.3b)
Instead, let us guess a solution, VI = 1 V, and see where this
Thus if XI, ... , Xn were the circuit variable values when y was the source, KXl, assumption leads. If VI = 1 V, surely il = 2 A by Ohm's law,
K Xn will be their values with the net source y scaled by K. and V2 = 1 V since parallel elements have the same voltage drop.
Thus i2 = 1 A, and at the top right node, i3 = i2 + il = 3 A.
We seek i I and i2 in Fig. 4.1. Applying KVL clockwise around the Since i3 flows through a 3-Q resistor, V3 = 9 V. Then, by KVL,
left loop, V4 = V3 + V2 = 10 V. Continuing right to left in this manner,
i4 = 2 A, is = i3 + i4 = 5 A, Vs = 5 V, and finally, if our guess
(4.4a) (VI = 1 V) were right, we would have VgI = V4 + Vs = 15 V.
This is not correct, since actually VgI = 45 V. But after all, this
we see that the source VgI equals a linear combination of i l and i2.
was no more than a guess to get us started. By the proportionality
By KCL at the top right node,
relation, if a 15-V source gives an output VI = 1 V, as we have
20 ~ (4.4b) discovered, our 45-V source will give three times as much, so the
correct answer must be
~il which, like all KCL and KVL equations in a linear circuit, also
45
vgI 40 ig2
equates a linear combination of circuit variables to a net source VI = 15(1) = 3 V
value. Solving the last for i2 and using this to eliminate i2 from
(4.4a), the current il is found to be This method of assuming an answer for the output, working
backward to obtain the corresponding input and finally adjusting the
(4.5a)
FIGURE 4.1 Linear circuit with two sources. assumed output to be consistent with the actual input by means of
and substituting this value for il into (4.4b), the proportionality relation, is particularly easy to apply to the ladder
network, but may be applied to other circuits as well.
. I 2
12 = -6Vg1 + 31g2 (4.5b)
The solution expresses the circuit variables as linear combinations A nonlinear circuit is, of course, one that has at least one element whose terminal
of the sources. By the proportionality principle, scaling each source relation is not of the form (4.1) or (4.2). An example is given in Exercise 4.1.4, for
by 2 should double all circuit variables. Examining (4.5), if VgI and which it is seen that the proportionality property does not apply.
120 Chapter 4 Analysis Methods Section 4.1 Linearity and Proportionality 121
Now suppose that we "kill" the source b; that is, we set its source function Yb to zero
EXERCISES but otherwise leave the circuit unchanged. Then, in this new circuit, we would have an
equation corresponding to the one above:
2Q 3Q 4.1.1. Find iI and V2, with (a) the source values as shown, (b) the source
values divided by 2, and (c) the source values multiplied by -2. Note how (4.7a)
the principle of proportionality applies in (b) and (c).
+ Answer (a) 2 A, 36 V; (b) 1 A, 18 V; (c) -4 A, -72 V
IOQ vz In this equation, the left-hand side would be the same linear combination as before (the
4.1.2. Suppose that VgI = 90 V in Fig. 4.2. Find VI, iI, v2, i2, v3, i3, ai's are unchanged), since only the value of source Yb was changed, not any of the linear
v4, i4, V5, and i5. elements or their placement in the circuit. The values of the circuit variables will be
Answer 6 V; 12 A; 6 V; 6 A; 54 V; 18 A; 60 V; 12 A; 30 V; different due to the presence of just the single source a alone and are designated with
EXERCISE 4.1.1 30 A
an a superscript. xf is thus the response to source a alone. If, instead, we kill the other
4.1.3. Find V and i using the principle of proportionality. source, source a, the circuit equation (4.6) becomes
Answer 8 V; 3 A
(4.7b)
As before, the a/s in (4.7b) will be identical to those in (4.6), and the xf's are the
response to source b alone. Adding (4.7a) and (4.7b) together gives
+
v 4Q
at (xf + xf) + a2(x~ + x~) + ... + an(x~ + x!) = Ya + Yb
Comparing (4.6) with the preceding equation, we arrive at our central result. Thinking
2Q
EXERCISE 4.1.3 of any of the Xi circuit variables as the response of this circuit to the sources, we have
First consider the circuit of Fig. 4.1, which was analyzed in Sec-
tion 4.1. The circuit variables i 1 and i2 were found in (4.5). Let
In this section we consider linear circuits with more than one independent source. The
us now use superposition to find these responses. First we kill the
linearity property makes it possible, as we shall see, to analyze these circuits by adding
current source, resulting in the modified circuit of Fig. 4.3(a), and
together the responses due to each source separately.
determine if and i~. This is the component a problem, and if and
Recall from (4.3) that each circuit equation may be expressed as
i~ are the components of the responses iI and i2 due to source a (the
aIXt + a2x 2 + ... + anXn = Y voltage source). This is a single-loop circuit, and
where the Xi'S are circuit variables and Y the net source value in this KCL or KVL .a VgI
/1 =-
equation. To be specific, suppose that there are exactly two sources, Ya and Yb: 6
b
I] = ] .
31g2 va = 3(6) =4 V
b 2
3+~
12 = 31g2
~----l+
6V
-\------,
The component b problem is shown in Fig. 4.5(b). Killing the
By the superposition principle, each response is the sum of its com- 6-V source has drawn nodes a and c together as shown. Once again,
ponent responses, or 2Q 3Q
after replacing the 2- and 3-Q resistors by their parallel equivalent
a~--~A.A~---*----~A.A~~c
~ Q, vb can be found by voltage division:
.
I] = I]'a +1] b
= ]6Vg] + 31g2
] .
6
6Q
. ab ]
12 = 12 + 12 = -6 V g] + 31g2
2
vb = ~(l8) = 3 V
~+6
These results indeed agree with our previous calculations (4.5).
d The component c problem [Fig. 4.5(c)] has three resistors in
parallel:
fIGURE 4.4 Example with three independent
n b sources. 1
2Q ~ 2Q ~ VC =- 1 1 ] (2) = -2 V
6+2:+3
il~ b ~
'1
By superposition, the value of v is the sum of its components:
4Q 4Q ig2
v = va + vb + VC = 4 + 3- 2= 5 V
6V
Example 4.3 demonstrates that a trade-off exists in the use of superposition. On the
~--~+-~----~
one hand, there are two (or more) component problems to solve instead of just the one
we were given originally. On the other hand, each component problem is simpler than
2Q 3Q
the original, both because it contains only one source and because killing sources reduces a~--~AI\~--'---~I\I,\~~c
the number of nodes and/or loops in the circuit. Indeed, both component problems in
Example 4.3 involve simple circuits (single loop in component a, single pair of nodes
in component b), whereas the original circuit was not simple and could not be analyzed (a)
without solving simultaneous equations in two unknowns [(4.4a) and (4.4b)].
2Q
As a second example, let us find the voltage v in the circuit with
three independent sources shown in Fig. 4.4. Using superposition, 6Q 6Q
we will find the components of v due to each source separately,
referring to the 6-V source as source a, the 18-V source as source
b, and the 2-A source as source c.
The component a problem is shown in Fig. 4.5(a), in which
(b) (c)
the band c sources have been killed. Killing the 18-V source shorts
nodes band d together into a single node. The circuit has been drawn fiGURE 4.5 (a) Component a; (b) component b; (c) component c.
if + 3(6 + if) + 2i b = 0
or if = -3 A, and In this section we develop a general method of circuit analysis in which voltages are the
unknowns to be found. A convenient choice of voltages for many networks is the set of
Vb = 3(6 + if) = 9V node voltages. Since a voltage is defined as existing between two nodes, it is convenient
to select one node in the network to be a reference node and then associate a voltage at
Then, by superposition of component voltages, each of the other nodes. The voltage of each of the nonreference nodes with respect to the
v = va + vb = 6 + 9 = 15 V reference node is defined to be a node voltage. It is common practice to select reference
directions for these voltages so that the plus ends are all at the nonreference nodes and
'a
the minus ends all at the reference node. For a circuit containing N nodes, there will be
~ 12 V N - 1 nonreference nodes and thus N - 1 node voltages. Nodal analysis is a method in
-+
+ which we will break the circuit, that is, solve for a key set of circuit variables, by finding
va 3Q the node voltages themselves. Any other current or voltage will follow easily once the
lQ
2i'{
lQ circuit is broken.
+ The reference node is often chosen to be the node to which the largest number of
branches are connected. Many practical circuits are built on a metallic base or chassis,
and usually a number of elements are connected to the chassis, which becomes a logical
(a) (b)
choice for the reference node. In many cases, such as in electric power systems, the
fiGURE 4.7 (a) Component a; (b) component b. chassis is shorted to the earth itself, becoming part of a single chassis-earth node. For
V~~~+"':f'vj:~
The reference node is thus at ground potential or zero potential, and each other node may VI and V2 are node voltages. The element voltage V IS given by
be considered to be at some potential above or below zero specified by the value of its 2
node voltage. V = VI - V2
VI v2
The equations of nodal analysis are obtained by applying KCL at the nodes. Recall 3 and thus by Ohm's law we have
that each term in a KCL equation is an element current. For a resistor, this current is
proportional to its voltage. This voltage, like any element voltage, is equal to a node
=l- i=-=---
V VI - V2
voltage (if one end of the element is tied to the reference node) or the difference of two R R
fiGURE 4.8 Reference fiGURE 4.10 Single i =
node voltages (if both ends are tied to nonreference nodes). For example, in Fig. 4.8 the or G(VI - V2)
and nonreference nodes. element.
reference node is node 3 with zero or ground potential. The symbol shown attached to h G 1/ R is the conductance. That is, the current from node I to node 2 throu g;
node 3 is the standard symbol for ground, as noted in Chapter 3. The nonreference nodes
no~e. voltag~ at no~e ~ a:~ ~e ~~er~~~:::;: ;:I~~~~W
::
1 and 2 have node voltages VI and V2. Thus the element voltage VI2 with the polarity ; r;:;stor=: the difference of the
shown is
~~v:~~::e :~~~t:~::t~n~rr:~:::~~~::P:C~~~~i;;C~; in te~s of thel no~e v~~:esi
Now returning to the circuit of Fig. 4.9, the sum of the currents eavmg
VI2 = VI - V2
must be zero, or
The other element voltages shown are
il + i2 - igl = 0
VI3 = VI - 0= VI In terms of the node voltages, this equation becomes
V23=V2-0=V2 G1Vl + G2(VI - V2) - igl = 0
These equations may be established by applying KVL around the loops (real or imagined). We could have obtained this equation directly using the proce~ure of the preceding
Evidently, if we know all the node voltages, we may find all the element voltages and paragraph. Applying KCL at node 2 in a similar manner, we obtam
thence all the element currents.
-i2 + i3 + ig2 = 0
The application of KCL at a node, expressing each unknown current in terms of the
node voltages, results in a node equation. Clearly, simplification in writing the resulting or G2(V2 - VI) + G3V2 + ig2 = 0
equations is possible when the reference node is chosen to be a node with a large number
Instead of summing currents leaving the node to zero, we could have used the f~rm
of elements connected to it. As we shall see, however, this is not the only criterion
for selecting the reference node, although it is frequently the overriding one. Since we of KCL that equates the sum of currents Ieavmg
. the node to the sum of currents
h . entenng
ht-hand
the node. Had we done so, the terms igi and ig2 would have appeared on t eng
are going to apply KCL systematically at circuit nodes, the most straightforward case to
side:
consider is that of circuits whose only sources are independent current sources. We begin
with examples of this type. GI VI+ G2(VI - V2) = igl
In the network shown in Fig. 4.9(a), there are three nodes, dashed and numbered G2(V2 - VI) + G3V2 = - ig2
as shown. [This may be easier to see in the redrawn version of Fig. 4.9(b).J Since there
are four elements connected to node 3, we select it as the reference node, identifying it Rearranging these two equations results in
by the ground symbol shown. (G I + G2)VI - G2V2 = i gl (4.8a)
These equations exhibit a symmetry that may be u~ed .to ,:"rite the equat~:ns t~:
h ed form (4.8) directly by inspection of the CirCUit diagram. In (4. ) .
t e re~ang. m of conductances of the elements connected to node 1, while
::=~~i~:~::~:~:~;~~:~:~~ff:e(~~~~~en:,~:.:~:~t;::;~F~,i
3 Thus nod~ 2 plays the role in (4.8b) of n~de 1 in (4.~a) .. That IS, It tt f;o:' ~ea c:rrent
KCL is applied. In each equation the nght-hand Side IS the curre
l:
(a)
(b) sources that enters the corresponding node. ources KCL
fiGURE 4.9 Circuit containing independent current sources. In general, in networks containing only conductance.s and c~~n~: On :he left
applied at the kth node, with node voltage Vb may be wntten as 0 0 .
128 Chapter 4 Analysis Methods
Section 4.3 Nodal Analysis 129
side of the node k equation, the coefficient of the kth-node voltage is the sum of the Gaussian elimination. For the reader who is not familiar with these
conductances connected to node k, and the coefficients of the other node voltages are the methods, a discussion is given in Appendix A. Selecting Cramer's
negatives of the conductances between those nodes and node k. The right side of this rule, first find the determinant of the coefficient matrix, given by
equation consists of the net current flowing into node k due to current sources.
-1
This predictable pattern makes it easy to write down the node equations. Note that
6 (4.11)
the signs, positive on the left-hand side for Vk terms and negative for other node voltage
-2
terms, and positive on the right-hand side for current sources flowing into node k, are
a consequence of the form of KCL chosen. While other forms could be used quite as To determine VI, we replace the first column of the coefficient ma-
correctly, we advocate sticking to the form recommended, with the payoff that the terms trix by the vector of constants on the right-hand side of (4.9)-(4.10),
will always fall in this pattern. It helps to make the pattern of signs fixed and predictable, compute its determinant, and divide by the determinant of the coef-
so we can focus our attention on the larger issues when analyzing a circuit. ficient matrix already found.
II~ ~~ _~I
Nodal analysis consists in writing KCL node equations described above at all non-
reference nodes in the circuit. This yields N - 1 linear equations in a similar number of
unknowns (the node voltages). As discussed in Appendix C, these equations are linearly
independent and thus are guaranteed to possess a unique solution. The node voltages VI = -------- = 1 V
~
may be found by a variety of means, including Gauss elimination, Cramer's rule, and
matrix inversion. V2 is found by replacing the second and V3 the third column of the
coefficient matrix and calculating as above, yielding V2 = 2 V and
V3 = 3 V.
Consider the circuit of Fig. 4.11. The bottom node has been selected
Now that we have broken the circuit by finding the node volt-
as the reference node since so many elements connect to it. The
ages, we may easily find any other voltage or current. For example,
resistors are labeled according to their conductances.
if we want the current i in the 2-S element, it is given by
i = 2(V2 - V3) = 2(2 - 3) = -2 A
Note that the coefficient matrix shown in (4.11) is symmetric [the (i, j) and (j, i)
7A 17A elements are equal]. This follows from the fact that the conductance between nodes
i and j is that between nodes j and i. Symmetry further simplifies writing the node
equations. While symmetry will hold as a general rule for all circuits not containing
dependent sources, symmetry of the coefficient matrix cannot be counted on in that case,
as we shall see in the next example.
fiGURE 4.11 Circuit for Example 4.6.
Consider the circuit of Fig. 4.12, which contains dependent current
Since there are three nonreference nodes, there will be three sources. We will begin by writing the node equations exactly as if
equations in three unknown node voltages. At node VI, we note the sources were independent. At node 1,
that the sum of conductances is 3 + 1 = 4, the negative of the
conductance connecting node V2 to node VI is -1, and the net source
current entering node 1 is 7 - 5 = 2. Thus the first node equation is and at node 2,
(4.12) 4Q 12 Q
[-i
-1
6
0
0
4 -1 0 V3 1
0 -1 -1 6 -1 V4 3
0 -2 0 -1 5 V5 -2
The determinant of the coefficient matrix is (9) G) - (-9) ( - 2) ==
45/2 and the inverse is
\
[2-
Then
2
45 ~ 1J EXERCISE 4.3.2
3A t
5i
~----~~~----~
EXERCISE 4.3.3
ReES
At first glance it may seem that the presence of voltage sources in a circuit complicates
nodal analysis. We can no longer write the KCL node equations, since there is no way
FIGURE 4.12 Circuit containing dependent sources. to express the currents through these circuit elements in terms of their node voltages. As
discussed in Chapter 2, the element law for a voltage source does not relate its current
From Example 4.7 we see that the presence of dependent sources destroys the to its voltage, so we cannot use it to replace a current unknown by a voltage unknown
symmetry in the coefficient matrix [see (4.12)] and that in such circuits the elements in the node equation.
of this matrix may no longer simply be interpreted as sums of conductances, since the However, as we shall see, nodal analysis in the presence of voltage sources proves
dependent sources also contribute. On the other hand, the presence of dependent sources no more complicated, requiring only a small modification to the basic method for writing
has not significantly complicated nodal analysis, requiring only an additional substitution the equations of nodal analysis presented in Section 4.3. In fact, we will come to welcome
step, replacing controlling variables by node voltages. voltage sources, since they reduce the number of simultaneous node equations that must
be solved, yielding one less equation per voltage source.
EXERCISES To illustrate the procedure, let us consider the circuit of Fig. 4.13.
For convenience we have labeled the resistors by their conductances.
4.3.1. Take all resistors in Fig. 4.9 to be 1 n and both current source Note that we have enclosed voltage sources in separate regions indi-
functions to be 1 A. Using nodal analysis, find the node voltages and the cated by dashed lines. Recalling that the generalized form of KCL
three labeled currents. states that all currents entering a closed region must sum to zero
132 Chapter 4 Analysis Methods Section 4.4 Circuits Containing Voltage Sources 133
Supenlode IS Vz 2S ,,-- Supemode B The number of node voltage unknowns is thus reduced by the num-
, ,
A
, ber of independent voltage sources. This is shown in Fig. 4.13(b)
by relabeling each supemode so that one of its node voltages is ex-
pressed as the algebraic sum of the node voltage at the other plus
:5V vS/
,, the known voltage drop across the voltage source.
To complete the formulation of the nodal analysis equations,
IS let us apply KCL to all supernodes not containing the reference node
and to all other nonreference nodes. This has the effect of reducing
' ...... _-----_ ...
the number of node equations by one per voltage source. The pres-
- ence of voltage sources thus reduces the number of equations and the
(a)
number of unknowns in nodal analysis by one per voltage source. For
instance, in Fig. 4. 13(a) there are five nonreference nodes, and if the
2S , voltage sources were all current sources, five equations in five un-
, knowns would be required. Because there are two voltage sources,
however, we may break the circuit by using just three equations in
three unknowns, as we now demonstrate.
:5V V4,+ 4
, We do not require a KCL equation for supemode A because
it contains the reference node. Referring to Fig. 4.13(b), equating
IS currents out of node 2 through resistors to net current into node 2
via current sources in the usual way,
-
(l)(V2 - 5) + (4)(V2 - V3) + (2)[V2 - (V4 + 4)] = 6
(b) while at node 3
FIGURE 4.13 (a) Circuit containing voltage sources; (b) node voltages (4)(V3 - V2) + (3)(V3 - 0) + (2)(V3 - V4) = 0
in supernodes relabeled.
and at supemode B,
(just as they do for ordinary nodes), we will refer to each of these
regions as a supernode.
(2)(V4 - V3) + (2)[(V4 + 4) - V2] + (1)[(v4 + 4) - 0] = 0
Each supemode contains two nodes, one a nonreference node Rearranging these equations in vector-matrix form yields
and another node that may be a second nonreference node (e.g.,
supemode B in Fig. 4.13) or the reference node (e.g., supemode A).
Supemodes containing the reference node have one node voltage [ -2
-~
-4
9 -2]
-2
-2 5
variable, such as VI in Fig. 4. 13(a). But since there is an element
with known voltage drop connecting VI to a known node voltage The coefficient matrix has all the same properties it did in the ab-
(zero since it is the reference node), VI is not in fact an unknown sence of voltage sources: the on-diagonal elements are the sum of all
and may be determined immediately: conductances into the node or supemode at which that equation was
written, the (i, j) element is the negative sum of conductances con-
VI = 5 +0 = 5 V necting the ith equation's node or supemode with the jth equation's,
Similarly, supemodes not containing the reference node have two and the matrix is symmetric. The source vector on the right-hand
node voltage variables [V4 and Vs in Fig. 4.13(a)]. However, since side, however, is no longer simply the net current in due to current
their difference is known, sources, since it now contains terms due to the voltage sources as
well.
Vs - V4 = 4
we may consider just one of these node voltages as an unknown, As a second example, let us find V and i in the circuit of Fig. 4.14.
say V4, and the other known in terms of V4 as Assigning node voltages of 0 V and 0 + 20 = 20 V to one supemode
and V and v+3 to the other, there is one supemode not containing the
Vs = V4 + 4 reference node, and there we must write a KCL equation. Summing
134 Chapter 4 Analysis Methods Section 4.4 Circuits Containing Voltage Sources 135
currents into this region in the usual way gives In writing these equations, we proceeded as if the dependent voltage
I I source were an independent source of value lOv. Next we replace
6 X 103 [(v + 3) - 20] +2x 103 [(v + 3) - 0] +4 X 103 (v - 0) the controlling variable by node voltages. Since v is the element
voltage between nodes with voltages given as lOv and VI,
= 6 X 10-3
v = lOv - VI
Solving the equation yields v = 8 V. Having broken the circuit,
we may now easily find any other circuit variable, such as i. This
or v = bVI. Substituting this into (4.13b), (4. 13 a) becomes
current is 6Vi - 2V2 = 10
i=
I I
[(v+3)-20]= (-9)=_lmA
-12vI + llv2 = -6
6xl~ 6xl~ 2
Solving by Gaussian elimination, we multiply the second equation
Note that while there are three nonreference nodes in this problem, by ~ and add it to the first, yielding
we had to solve only a single equation in one unknown, not three in
three, because of the welcome presence of the two voltage sources. V2 =2V
Substituting this back into the first equation yields VI = ~ V. With
the circuit broken, analysis may be completed by inspection. For
instance, the current i is
i = 2( V2 - VI) = 2(2 - ~) = - ~A
20V
6A 6A
(a)
2S
50 +
V - 4S
(a) (b)
(b)
FIGURE 4.15 (a) Circuit containing dependent voltage sources; (b) re-
FIGURE 4.14 (a) Circuit for Example 4.9; (b) redrawn with supernodes drawn to show supernodes.
and node voltages labeled.
136 Chapter 4 Analysis Methods Section 4.4 Circuits Containing Voltage Sources 137
4.4.4. With x an independent voltage source as in Exercise 4.4.1, suppose We define a mesh current as the current that circulates around a mesh. If an element
that we desire a value of +5 V for v. What should the source function for is located on a single mesh such as Vg1 or R2 in Fig. 4.16, it carries an element current
x be? just equal to this mesh current (mesh current i1 for Vgj, i2 for R4). If an element is located
Answer -3 V on the boundary between two meshes, such as R2 in the same figure, its element current
is the algebraic sum of the two mesh currents circulating through it (i 1 - i2 for element
R2)' Labeling element currents with capital letters and mesh currents with lowercase
letters, both R1 and Vg1 have the same element current II, where
It = i1
The element current h defined positive downward through R3 is
In the nodal analysis of the preceding sections we applied KCL at the nonreference nodes
of the circuit. We now consider a related method known as mesh analysis, in which KVL h = i1 - i2
is applied around certain loops in the circuit. As we shall see, in this case the unknowns
Mesh analysis consists in writing KVL around each mesh in the circuit, using mesh
in the resulting equations will be currents.
currents as the unknowns. The resulting system of equations is guaranteed to be linearly
We restrict attention in this section to planar circuits, circuits that can be drawn on
independent, as shown in Appendix C, and thus possesses an unique solution. As with
a plane surface (like a piece of paper) so that no elements or connecting wires cross. In
nodal equations, the solution may be found by any convenient method such as Gauss
this case the plane is divided by the circuit diagram into distinct areas in the same fashion
elimination, Cramer's rule, or matrix inversion.
that the solid framework in a window separates and outlines each individual windowpane.
The closed boundary of each such windowpane area is called a mesh of the circuit. Thus
a mesh is a special case of a loop, which we consider to be any closed path of elements Consider the circuit of Fig. 4.17. Writing KVL clockwise around
in the circuit passing through no node or element more than once. In other words, a the first mesh,
mesh is a loop that contains no elements within it. R1i1 + R 3(i1 - i2) = Vg1
The circuit of Fig. 4.16 is planar and contains three meshes identified and clockwise around the second mesh,
by the arrows. Mesh 1 contains the elements Rj, R 2, R 3, and Vg1; R2i2 + R 3(i2 - i1) = -Vg2
mesh 2 contains R 2, R 4, Vg2, and Rs; and mesh 3 contains Rs, Vg2,
R 6 , and R 3 . Note that in the second equation we have expressed the element
current through R3 as i2 - iI, so its reference direction arrow points
in the direction in which we are traversing the loop (clockwise).
This way the voltage drop is positive in this KVL equation. Had we
+
R, '0) "" -+ FIGURE 4.17 Circuit with two meshes.
used i 1 - i2 as the element current, as we did in the first equation,
we would have had a minus sign in front of the corresponding term,
- R3 (i 1 - i 2) in the second equation. We will always use implied
R, R~ element current reference directions that point in the direction in
which we are traversing the mesh, thus keeping the signs for all
resistive voltage drop terms positive. This is for convenience, of
course; it is never incorrect to replace a term like + R3 (i2 - i 1) in
FIGURE 4.16 Planar circuit with three meshes. any equation by - R3 (i 1 - i2)'
In the case of nonplanar circuits, we cannot define meshes and mesh analysis There is also a shortcut method of writing mesh equations that is similar to the
cannot be performed. Mesh analysis is thus not as general as nodal analysis, which had shortcut nodal method of Section 4.2. Rearranging the two equations of Example 4.12,
no such topological restriction. A KVL-based technique may still be employed, however,
even for nonplanar circuits. The procedure in this case is similar to mesh analysis, but (R1 + R3)i1 - R3i2 = Vg1 (4.14a)
the equations are not as easily formulated. This technique, generalized loop analysis, is -R3 i 1 + (R2 + R3)i2 = -Vg2 (4.l4b)
summarized briefly in Appendix C. Fortunately, the large majority of electric circuits we
shall have need to analyze in practice are planar, and for those we may choose nodal or we note that in the first equation, corresponding to the first mesh, the coefficient of the
mesh analysis. For nonplanar circuits, the only general method developed in this book first current is the sum of the resistances associated with the first mesh, and the coefficient
is nodal analysis. of the other mesh current is the negative of the resistance common to that mesh and the
\~
around mesh 2, -3 Q is the negative of the net resistance on the
In Fig. 4.16 define i J, i 2 , and i3 as the mesh currents circulating 2$1 border between the meshes, -2v is the clockwise voltage rise due
clockwise in meshes 1, 2, and 3, respectively. Applying the shortcut to voltage sources in mesh 1, and 30 is the rise due to sources in
method to mesh 1, we have it In + In
mesh 2. Replacing the controlling variable v by 1(i2 - i J) via Ohm's
law gives us
(4.15a)
FIGURE 4.18 Example with a dependent
This result may be checked by applying KVL to mesh 1, resulting source.
in
We will solve for i J, which is the desired current i, using Cramer's
rule. The determinant of the coefficient matrix is
[-~;~]
4.5.2. Repeat Exercise 4.5.1 with RJ = 1 0, R2 = 2 0, R3 = 4 0,
= VgJ = 21 V, and Vg2 = O. Check by using equivalent resistance and current
Vg2 division.
Answer 9 A; 6 A
The diagonal elements are the sums of the resistances in the meshes,
4.5.3. Using mesh analysis, find iJ and i2 if element x is a 6-V indepen-
and the off-diagonal elements are the negatives of the resistances EXERCISE 4.5.3 dent voltage source with the positive terminal at the top.
common to the meshes corresponding to the row and column of the Answer 2 A; 1 A
element. That is, - R2 in row 1, column 2 or in row 2, column 1 is
4.5.4. Repeat Exercise 4.5.3 if element x is a dependent voltage source
the negative of the resistance common to meshes 1 and 2, and so on. of 6i J V with the positive terminal at the bottom.
Thus the matrix is symmetric. As was the case with nodal analysis, Answer 5 A; 6 A
this symmetry is not preserved if there are dependent sources present.
unknown mesh currents. Note, however, that each current source bordering only one R] +R2+R3
mesh, such as ig] in Fig. 4.19(a), results in one less unknown mesh current, since the mesh This solution is considerably easier than the three equations in three unknown mesh
current must agree with the given current source function. Thus we relabel the mesh currents we would be left with if the current sources were voltage sources. We welcome
current i3 in Fig. 4.19(b) by its known value, current sources in mesh analysis for this reason.
10 10
ig1 ig1
<E-- <E--
Rl ~ R2 Rl ~ R2 +
20 20
2V 2V
~ R3
~ R3
(a) (b)
(a) (b)
fiGURE 4.20 (a) More complicated circuit; (b) mesh currents rela-
FIGURE 4.19 (a) Circuit with current sources; (b) mesh currents relabeled. beled.
142 Chapter 4 Analysis Methods Section 4.6 Circuits Containing Current Sources 143
Fig. 4.20(a) or (b), since the current sources lie inside its boundary ...
(hence its designation as a supermesh). Writing KVL around mesh 2 EXERCISES
in Fig. 4.20(b) gives
4.6.1. Using mesh analysis, find i.
li2 + 3i y + 1(i2 - il + 9) + 2(i2 - i I) = 0 (4.1Sa)
or (4.20b)
Using (4.20) to eliminate Vx and iy from (4.19) yields There is an interesting symmetry of opposites exhibited by pairs of network equations
which we have encountered. For example, Ohm's law may be stated in v fonn as
-ISi i + 16i2 = -36
2Si l - 23i2 = 56
v = Ri (4.21a)
or in i fonn as
Solving, we find that i I = 2 A and i2 = 0 A. This result may be
confirmed by use of any of our linear algebraic methods (Gauss, i = Gv (4.2Ib)
Cramer, or matrix inversion) or by back substitution of il and i2
into these two equations.
Equation (4.2Ib) may be derived from (4.21a) by solving for i and replacing by G, of
course. But note that either equation may be formed from the other by replacing each
*
With the circuit now broken, any other variable needed may symbol by its paired opposite: i by v, v by i, R by G, G by R.
be easily recovered. The other two mesh currents may be found Similarly, in the case of series resistances, R I , R2 , .. , Rn, the equivalent resistance
from their labels in Fig. 4.20(b) in tenns of il and i 2 . All resistive was shown in Section 2.3 to follow the "series resistance adds" rule:
voltages follow from Ohm's law.
Rs = RI + R2 + ... + Rn
In general, before analyzing any circuit, we should note how many equations are Replacing R by G and series by parallel,
required in nodal analysis and in mesh analysis and choose the method with fewer equa-
tions. The number of nodal equations is the number of nodes (less one, the reference
Gp = G I + G2 + ... + G n
node) minus the number of voltage sources. The number of mesh equations is the number which is the "parallel conductances add" rule of Section 2.4. It is clear that one of these
of meshes minus the number of current sources. It is further assumed that series and equations may be obtained from the other by interchanging resistances and conductances
parallel equivalents have been substituted where possible before perfonning either nodal and the subscripts s and p (i.e., series and parallel).
or mesh analysis. Series elements multiply the number of nodes unnecessarily; similarly, There is a systematic symmetry of opposites, or duality, between resistance and
parallel elements cause meshes to proliferate. conductance, current and voltage, and series and parallel. We acknowledge this by
V2
Series Parallel
v
G Mesh Node
R
Short circuit KVL KCL
Open circuit
Ind. current source v form i form
Ind. voltage source
CCCS Thevenin form Norton form
VCVS
VCCS Voltage divider Current divider
CCVS
(a) (b)
defining these quantities as duals of each other. That is, R is the dual of G, i is the dual FIGURE 4.21 (a) Circuit; (b) dual circuit.
of v, series is the dual of parallel, and vice versa in each case.
are connected to the reference node. The primal (original) circuit
Another simple case of dual equations is
and its dual are shown in Fig. 4.21.
v=O Figure 4.21(a) may be described as Vs in series with Rl and
the parallel combination of R2 and R 3. Replacing the quantities in
and its dual this statement by their duals, we see correctly that Fig. 4.21(b), its
i=O dual circuit, is i g in parallel with Gland the series combination of
G2 and G3.
In the general case, an element described by v = 0 is a short circuit, and one describe.d
by i = 0 is an open circuit. Thus short circuits and open circuits are duals. Table 4.1 IS
Ne~t. consider systematic construction of the dual circuit from the circuit diagram
a table of duals. of ~e on~mal.' or primal, circuit. As illustrated in Example 4.16, meshes in the original
Note that dual quantities are not equivalents; they do not even fit in the same cir-
(pnmal) CircUlt are dual to nodes in the dual circuit. Each mesh current in the primal
cuit. For instance, Rand G are duals, but an 8-Q resistor and 8-S conductance are not
became a node voltage in the dual, thus each mesh in the primal corresponds to a node
equivalent. Similarly, the dual to a Thevenin form (series voltage source and .resistor).is a
in the d.ual. This suggests that we begin constructing the dual circuit diagram by placing
Norton form (parallel current source and conductance), but the~ are not eqUlva~ent smce
a node m the center of each mesh of the primal as shown in Fig. 4.22(a). One additional
they do not in general satisfy VT = RTIN or RN = RT as reqUlred for Thevemn-Norton
no~e is needed in the ~ual circuit, the reference node, for which there is no node voltage
equivalents. vanab.le or ~orr~sponding mesh in the primal circuit. Elements on the outer periphery of
the pnmal CircUlt have only one mesh current passing through them, so they will be dual
Consider the circuit of Fig. 4.21(a). The mesh equations are given
to elements with only one node voltage variable across them. These must be elements
by connected to the reference node in the dual circuit, since all others will have two node
(RI+ R2)il - R2i2 = Vg
(4.22)
voltages across them. Thus we add one last node to our dual circuit, the reference node
placing it somewhere outside the primal circuit as shown in Fig. 4.22(a). '
-R2il + (R2 + R3)i2 = 0
To complete the dual circuit we must add its elements. Note that all terms in the
To obtain the dual of (4.22), we simply replace the R's by G's, the first primal mesh equation will be dual to terms in the first dual node equation. So we
i's by v's, and v by i. The result is attach one element at node 1 of the dual through each of the primal circuit elements
+ G2)VI - G2V2 =
(GI ig
surrounding this node. The element we attach will be the dual element, of course. The
(4.23) completed dual circuit is shown in Fig. 4.22(b). The dual circuit has the same number
-G2VI + (G2 + G3)V2 = 0 of elements as the primal and one more node than the primal has meshes. Note that the
These are the nodal equations of a circuit having two nonreference controlled source, a current-controlled current source (CCCS), has gone over to its dual
node voltages, VI and V2, three conductances, and an independent controlled source, a voltage-controlled voltage source (VCVS). The controlling current
current source i g From our shortcut procedure for nodal equations ic = i3 - i2 is replaced by its dual, the controlling voltage Vc = V3 - V2. Since the
we see that G I and G2 are connected to the first node, G2 is common reference arrows for the primal source and its mesh current pointed in the same direction,
to the two nodes, G2 and G3 are connected to the second node, and the referen~e +/- signs for the dual source agree with the dual variable, V3 (plus at V3).
i g enters the first node. Since G I and i g are not connected to the Duality has many uses in circuit theory. It may be used to generalize a theoretical
second node and G3 is not connected to the first node, these elements result immediately to its dual without further proof. For instance, we could have invoked
4Q
Vl
\
(a)
(a) (b)
FIGURE 4.22 Construction of the dual circuit: (a) node placement; (b) dual
circuit. IQ
duality to write down the current-divider law for conductances in parallel from its dual, the
voltage-divider law for resistances in series. Duality can also be used to produce circuits
whose currents have certain desirable properties from other circuits whose voltages have
those properties, or vice versa. Since nodal analysis on the dual is identical to mesh
analysis on the primal, duality further permits us to use just one general method, say
nodal analysis, whichever way is most numerically efficient. For instance, a circuit with
fewer meshes than node voltages can be dualized and then nodal analysis done with the
same benefits as mesh analysis on the primal. This is particularly useful in the production
of computer-based tools, which by this duality-based approach only need to "know" one
method, either nodal or mesh analysis, both need not be programmed.
(b)
6V 16V
A general approach to the analysis of circuits containing op amps was developed in Chap-
(a) ter 3. Each op amp in the circuit diagram is replaced by a suitable model, for instance,
the ideal voltage amplifier model of Fig. 3.7 or the improved model of Fig. 3.9. Since
these models contain only resistors and dependent sources, the resulting circuit may be
analyzed just as any other linear circuit would be.
There is another approach to analyzing op amp circuits that, where applicable, is
far simpler to carry out. It does not require replacing op amps by models but is applied
to the original circuit diagram directly. For each op amp in the circuit (Fig. 4.24), two
rules are used:
1. The current into the op amp input terminals is zero: iii = ii2 = O.
(b) 2. The voltage across the op amp input terminals is zero: Vin = O.
EXERCISE 4.7.1
.fIGURE 4;24Y~rtLJal.dpenandsho;'tp~inciples.
6V 2Q
+
The first of these rules asserts that the op amp input behaves like an open circuit.
Since the ideal voltage amplifier model of Fig. 3.7 employs an open-circuited input, this
rule, the virtual open principle, is not hard to justify. The second rule asserts that the op
amp input also behaves like a short circuit (no voltage drop across its terminals). This
EXERCISE 4.7.2 rule, the virtual short principle, requires a bit more reasoning. We first give examples of
the use of these rules and then consider the justification of the virtual short principle and
4.7.3. Find a circuit that is self-dual; that is, its dual is identical to itself. the key assumption under which it is valid.
150 Chapter 4 Analysis Methods Section 4.8 Virtual Short Principle for Op Amps 151
In Section 3.5 the inverting amplifier circuit shown in Fig. 4.25 was
analyzed. Replacing the op amp by its ideal voltage amplifier model
with open-loop gain A, it was found that the output voltage V2 is or
given by
(4.25a) To justify the virtual short principle, we reason as follows. Assume, in a given op
+ o--~\I\I'\~A--I
amp circuit, that the op amp output voltage remains finite in the high-op-amp-gain limit
+
In the high-op-amp-gain limit (as A becomes arbitrarily large), we as the op amp gain A becomes arbitrarily large. Since the output voltage is A times the
V2
have the voltage transfer relation for the inverting amplifier (3.8) input, as A becomes arbitrarily large the op amp output voltage will remain finite only if
repeated here: the input voltage becomes arbitrarily small. Passing to the high-op-amp-gain limit, the
RF input must be exactly zero, or we would have the product of a nonzero input times an
V2 = --VI (4.25b) infinite gain producing an infinite output voltage, contradicting the assumption that the
FIGURE 4.25 Inverting amplifier. RA
\ output remain finite.
~e now show how (4.25b) can be derived from the virtual open and Thus, for circuits whose op amp output voltages remain finite in the high-op-amp-
VIrtUal short principles without resort to models or passing to limits. gain limit, the virtual short principle follows. What class of circuits is this? Certainly,
Since the noninverting input terminal in Fig. 4.25 is at ground, by the circuit described by (4.25a) is in this class. Note that as A gets larger and larger, the
the virtual short principle, so is the inverting input terminal. Then op amp output voltage V2 tends not to infinity, but rather, to the finite value (- RF / RA)VI
the node connecting the two resistors is at ground potential, implying specified in (4.25b).
that the full voltage VI is across RA and V2 is across R F . Summing Indeed, any circuit with negative feedback, such as all the op amp building block
the currents into this node gives circuits described in Chapter 3, will be "well behaved" enough to tend to a finite output
VI V2 in the high-op-amp-gain limit. Thus we may use the virtual short and open principles
-RA + -RF = 0 (4.26)
(4.25) to analyze any op amp circuit that employs negative feedback. As has been noted
In this equation the current into the op amp at this node is zero previously, only such circuits are practical linear op amp circuits anyway, since the rest
by the virtual open principle. Rearranging (4.26) slightly yields the will have op amp output voltages that "blow up," exceeding their physical output voltage
voltage transfer relation (4.25b). ratings when plugged in. This violation of safe limits will result in behavior certainly
nonlinear and perhaps rapidly destructive. So our virtual open and short rules will be
Find V2 in Fig. 4.26 . valid for any useful linear op amp circuit.
. Since no current flows into the op amp at the inverting input
terrnmal (the virtual open principle), RF and RA are in series, and The op amp circuits analyzed so far have been of the simple type,
thus they divide the voltage V2, with that is, reducible to a single equation. This example demonstrates
RA how to to solve a general op amp circuit problem. We will use nodal
+ o--...rvlf\r-.---4- VA = v2
RA+RF 5kQ 5kQ
analysis together with the virtual short and open principles. We seek
+
the load current i and the voltage v.
Since no voltage is dropped across the op amp input (the virtual short
V2 The circuit of Fig. 4.27 has two unknown node voltages, la-
principle), this is also the voltage at the noninverting input terminal,
or beled VI and V2. By the virtual short principle, the inverting input
terminal is at ground potential, and we have labeled its node voltage
FIGURE 4.26 Circuit for Example 4.19. V = VA 5V o V. Thus we need two node equations. For circuits not containing
Looking at. the left loop, RI and R2 are in series across VI (again, op amps, we write KCL at each node or supemode containing an
no current mto the op amp due to the virtual open principle). Thus unknown node voltage. Where op amps are present, we modify that
by voltage division of VI, strategy by writing KCL at the inverting input node rather than the
op amp output node. At nodes 1 and the noninverting input node,
R2 FIGURE 4.27 More complicated op amp
V = VI respectively,
RI +R2 circuit.
1 i: I 1
Combining these last three equations, it must be that 10 X 103 + 20 X 103 VI + 10 X 103 (VI - 5) = o(4.27a)
RA R2 1 1 1
V2 = VI
RA + RF RI + R2 10 X 103 (0 - VI) +5x 103 (0 - 5) +5X 103 (0 - V2) = 0(4.27b)
~::~ t~ g~~~nd ~ais ~f d:~~u~~~~u~e~~::~e I~;~~e:~~so:~;u~e::re~~i:~t~~n!n~u::: of repetitive hand "cranking." Fortunately, this is no longer the case. Modem technology
offers the digital computer, a splendid tool capable of relieving us of the tedious and in-
. e a no e equatIOn there. Instead, we choose to write KCL at the invertin efficient use of time inherent in lengthy hand calculations. Over the last four decades the
lllput node, where all currents can be easily determined. Once the circuit is broken th! computer has revolutionized how every kind of engineering is done, whether the design of
op amp output current can be found easily if so desired ' a space station, the improvement of automobile emission control devices, or the analysis
Th . .
. e v~rtual short and open principles rules are designated "virtual" because . of electrical circuits. With computer support for repetitive calculations, larger and more
f::~~~~:h~::!~~ c;~::~~d input;oltabge t~ physical op amps being operated in negati~: realistic problems can be solved and more experience with more complex systems can
enoug to e vIrtually, although not exactly zero This is be gained. Both the student and the engineering practitioner benefit by the opportunity
::~c~~ what we w~uld
expect if the op amp input had high but not infi~ite r~sistance for more advanced, more realistic problem-solving experience afforded by the assistance
amps) ; : ~p g;n A were large b.ut .finite (conditions characteristic of physical o~ of a computer equipped with the right software programs. While many useful computer
. . e ~Irtua op~n and short prlllclpies are excellent approximations and ma be programs are available to aid in the analysis of electrical circuits, we shall settle on one
:~:eto~I~PI:y analYSI~ wherever op ~ps are operated in the linear, negative feed~ack that is widely available on a variety of computer platforms. This software program is
. o.e,. o,:ever, t at use .of the vIrtual short principle applies exactl in the hi h- called SPICE (Simulation Program with Integrated Circuit Emphasis).
op-amp-gaJ~ IIrrut only. If the cIrcuit variables for a specific finite gain A Yd' d ~ We shall make frequent use of this tool. One common use will be to check hand
less convelllent model-substitution method developed in Chapter 3 shouldar:e ::~~o~ed~ results (many of the problems in subsequent chapters ask the student to solve a circuit
problem by hand and then "check using SPICE"). A second will be to solve problems
EXERCISES involving many nodes and/or meshes. A third important use of SPICE will be to explore
the effect of changing the value of a single circuit element, such as a specific resistance
lOkQ within a circuit. This is especially important for "cut-and-try" design, where we may
4.8.1. Use the :irtual short and open (* 11 12) principles to find the try several values of a circuit element before settling on one resulting in the best cir-
voltage transfer ratIO V2/VI for the noninverting amplifier of Fig. 3.19.
+ cuit performance. Typically, the entire numerical analysis needs to be repeated for each
Answer 1 + RF/RA
+ value tried, and the effort involved effectively eliminates hand calculation as a realistic
4.~.2. Use the virtual short and open principles to find the voltage transfer possibility in many design studies. The alternative to repeated trials is to write down an
ratIo V2/VI for the voltage follower of Fig. 3.14.
V2
Answer 1 equation explicitly describing the functional dependence of a selected measure of circuit
:.8/
n
th Use nodal analysis and the virtual short and open principles to
e voltage transfer ratio V2/ V I.
performance on the parameter in question and then somehow solve this equation for the
optimal value. With complex circuits this is seldom a practical alternative to the repetitive
Answer -4/5 process of cut-and-try.
EXERCISE 4.B.3 SPICE will prove helpful in each of these ways, but it remains only a computa-
tional tool. One way in which we cannot use SPICE is to substitute for the intellectually
challenging, sometimes painful process of acquiring a mastery of circuits concepts and
158 Chapter 4 Analysis Methods Section 4.9 Computer-Aided Circuit Analysis Using SPICE 159
:f :[
same current as the desired current through the 2-Q resistor. The
C> C> SPICE input file follows.
0---0+ 0---0+
0---0-
VI
/-lV I 0---0-
VI
t gVI
SPICE input file for the second example
@) @) *
(a) (b) V1 1 0 DC 4
R1 1 2 2
~I:f
~o.cp ~ ,I @)
(e)
ffi
~Ir
~~CP ~
@) \
R2
R3
F1
2
3
5
2
4
*Here comes the dummy voltage source:
VD 5 0 0
7
*Here is the CCCS with current gain 4 :
VD 4
(d) 3
R4 3 4
FIGURE 4.29 Dependent source conventions in SPICE. *And the VCVS with voltage gain 2 :
E1 4 0 2 1 2
.PRINT DC V(2,1) I(V1)
. The ~nal field .in each of these statements is the constant that mUltiplies the con-
trolhng varIable to gIve the dependent source function. These were defined in Chapter 3
END
as voltage gain p, for the VCVS, transconductance g for the VCCS, current gain fJ for
the CCCS, and transresistance r for the CCVS. The resulting SPICE output shows us that the 4-V source current is
-1.4 A and the voltage v is -28.0 V. Then the power dissipated by
This example contains two controlled sources, a CCCS and a VCVS. the 2-Q resistor is (-1.4)(-28.0) = 39.2 W.
Since the controlling current i does not flow through a voltage
so.urc~, we must insert a dummy voltage source in series to per- SPICE has many additional element statements, modes of analysis, and other fea-
mIt thIS current to be used. We wish to determine the voltage v and tures that we consider as the need for them arises in later chapters. For the present,
also the power dissipated by the 2-Q resistor (see Fig. 4.30). SPICE contenting ourselves with dc analysis of linear resistive circuits, we introduce just two
always outputs the power delivered by each independent source in additional capabilities of SPICE: the dc sweep and the subcircuit.
a dc problem without even being asked. Power delivered by or dis- A dc sweep is performed when we wish to step through several values of a dc
sipated by other elements, though, must be computed by hand from source, each time performing a dc analysis and printing the outputs. This mode of
current and voltage. So we will print out v and the current through analysis is useful, for instance, in determining whether an output variable will exceed
the 4-V source, which happens to be in series and thus carries the specified limits over an anticipated range of input (dc source) values. A dc sweep
mode of analysis is triggered by including a DC control statement in the SPICE input
4i file:
~
2Q
.DC SNAME START STOP DELTA
7Q
SNAME is the name of the independent source to be swept (current or voltage), ST ART
4V 4Q and STOP are the initial and final source values desired, and DEL TA is the increment
between successive source values within the sweep. The next example will include a dc
sweep.
A powerful feature of SPICE is the ability to define subcircuits, which can be typed
once in a SPICE input file and then referenced as often as needed without repeating
(a)
its element statements. Each time the subcircuit is needed, it is "called," similar to a
(b)
subroutine call in FORTRAN or a function invocation in C. The subcircuit is defined
FIGURE 4.30 (a) Circuit; (b) prepared for input as SPICE input file. by including a set of lines beginning with a . SUB CKT statement and ending with an
160 Chapter 4 Analysis Methods Section 4.9 Computer-Aided Circuit Analysis Using SPICE 161
. ENDS (end subcircuit) statement. The format is
C0:::t>--9 CD 30Q 0)
SUBCKT
*Su~circuit
ENDS
SUBNAME NA
element
NB NC ...
statements go here
o -
a
I-
a
0 =:r
-
(a) (b)
SUB NAME is a name given to the subcircuit. The remaining fields on the SUB CKT
statement are the numbers of the nodes of the subcircuit at which the subcircuit will be fiGURE 4.32 Subcircuit for Example 4.23.
connected to the main circuit. The subcircuit is called from elsewhere in the SPICE input
file by a statement of the form
proved circuit model of Fig. 3.9 (repeated in Fig. 4.32), defining this
Xyyyyy N1 N2 N3 .. SUBNAME op amp model as a . SUB CKT to avoid entering its element state-
ments twice. The type of analysis we will do is a de sweep of the
Here X indicates a subcircuit call, YYYYY is an arbitrary identifier for this particular independent de source between the limits of -5 and +5 V, print-
subcircuit invocation, N1, N2, N3, ... are the node numbers within the main circuit at ing out the op amp output voltages. Note that same node numbers
which the subroutine is connected, and SUB NAME is the name of the subcircuit to be used. are used independently in the main circuit and the subcircuit. The
SPICE is informed which subcircuit node to connect to which main circuit node SPICE input file used is:
by the order in which nodes appear. The first node on the SUB CKT statement will be
connected to the first node on the X subroutine call statement, the second to the second,
and so on. The node numbers within a subcircuit definition are independent of the
node numbers in the main part of the SPICE input file (with the exception only of Ex4-23: circui t wi.th two op amps
the reference node or common ground, node 0, which always refers to the same node
wherever it may appear). This independent numbering of nodes corresponds to the
*
*Main drcui t:
independent use of "formal" variables in a FORTRAN subroutine or C function and VG1 1 o DC 1
"actual" variables in the main program. Independent node numbering is what permits R1 1 2 10K
us to insert the same subroutine in several different places in the main circuit without R2. 2 3 20K
confusion. X1. o 2 3 OPAMP
*Order on nodes on above statement matches
As a final example, let us consider the circuit containing two op *.order on . subckt stat~ment: + inT - in, out .
amps shown in Fig. 4.31. We will replace each op amp by the im- R3 4 o 1.bK
R.4 4 5 48K
20kQ )(2 3 4 5 OPAMP
*H~l"e comes the op amp subcircuit:
SUBCKT OPAMP L 2 3
~'------l+ CD *NQde 1 is the +in, 2 the -in, and 3 the output node
RIN :Ii 2 1MEG
E.1 4 0 1 2 100K
R04 3 3D
48kQ
16kQ EN.DS
li:Finallytha analysis mode and output statments:
.DCVG1 -5 +5 1
.PRINT DC V(3) V(S)
fiGURE 4.31 Circuit for Example 4.23. .END
162 Chapter 4 Analysis Methods Section 4.9 Computer-Aided Circuit Analysis Using SPICE 163
After running SPICE with this input file, the output contains the Answer
following lines:
Exercise 4.9.1 SPICE input file
R11 0 1 10
R2 1 2 2
VG1 V(3) V(S)
V1 2 0 DC 24
-S.OOE+OO 1.00E+01 400E+01 R3 2 3 3
-4.00E+00 800E+00 3.20E+01 4
I2 0 3 DC
-3.00E+00 6.00E+00 240E+01 ICV1 ) V(3,0)
.PRINT DC
-2.00E+00 400E+00 1.60E+01 .END
-100E+00 2.00E+00 799E+00
O.OOE+OO O.OOE+OO O.OOE+OO
4.9.2. Write a SPICE input file for Fig. 4.6.
1.00E+00 -2.00E+00 -7.99E+00 \ Answer
2.00E+00 -400E+00 -1.60E+01
3.00E+00 -6.00E+00 -240E+01 Exercise 492 SPICE input fi 18
4.00E+00 -8.00E+00 -3.20E+01 R1 1 0 1
S.OOE+OO -1.00E+01 -4.00E+01 V1 2 1 DC 12
R2 2 3 3
H1 3 0 V1 -2
These are the op amp outputs for input voltage VG1 swept from -5 I2 0 2 DC 6
to +5 V by increments of + 1 V. Note that in each row V ( 3) = -2 PRINT DC ICV1 ) V(2,3)
VG1, confirming the observation that the first op amp stage is an .END
inverting amplifier with gain of - R2/ R 1 = -2. Also, V ( 5 ) = 4
V ( 3 ), since V ( 3) is input into a noninverting amplifier with gain The negative of the value I ( V1) will correspond to the desired output
of (l + R4/ R3) = +4. Note also that the column of V( 3) and of variable i in the figure. This sign change could be avoided, as well as
V ( 5) values both scale with VG1, a consequence of the principle the minus sign on the transresistance of the controlled source (see the H
of proportionality. For instance, the values of these variables for statement), by putting a dummy voltage source in series with the 12-V
VG1 = 2 V are double those for VG1 = 1 V. source but with opposite reference direction as the 12-V source.
For all its power, SPICE is just a numerical algorithm and
will have round-off errors. For instance, V ( 5) is exactly four times
V ( 3 ), as expected, for almost all values of V ( 3 ). But the value
V (3) = 2.0000 V yields V ( 5 ) = 7.9999 V, not 8.000 V. SPICE is
not an algorithm endowed with artificial intelligence. It does strictly
numerical calculations, not symbolic ones. With a hand calculator In this chapter tools of great generality for the analysis of linear circuits are introduced:
we may be surprised that division by a large number followed by the principle of superposition, the methods of nodal and mesh analysis, and a computer
multiplication by that same number sometimes will not return the program capable of simulating a very broad range of circuits called SPICE.
original value to the display, so we cannot expect SPICE to give
exact values, just close numerical approximations. Linear circuits satisfy the principle of proportionality: scaling all independent sources
by the same factor scales all responses similarly.
They also satisfy the principle of superposition: the total response is the superposition,
or sum, of responses to each independent source with the others killed.
In nodal analysis the key variables are the node voltages, and KCL is written at all
nodes but the reference node.
In nodal analysis, each voltage source defines a supemode containing only one unknown
EXERCISES node voltage and KCL is applied to the supemode.
In mesh analysis the key variables are the mesh currents, and KVL is written around
4.9.1. Write a SPICE input file for the circuit shown in Exercise 4.1.1. all meshes.
The virtual short and virtual open principles for analyzing op amp circuits in negative
feedback configurations state that the voltage across, and the current into, the input
terminals of the op amp both equal zero.
Nodal analysis is recommended for op amp circuits using the virtual short and open fiGURE P4.9
principles. Do not write nodal KCL equations at op amp output nodes. FIGURE P4.5
\
SPICE requires an input file which details the circuit, establishes the desired mode of 4.6. Find v using the proportionality principle. All resis-
operation and indicates the desired outputs variables and formats. 4.10. Use superposition to find Vgl and i.
tors are 10 Q.
+v-
The examples in this chapter apply the principle of superposition, and the methods of
mesh and nodal analysis, strictly to resistive circuits studied in the time domain. This is
not a limitation of these methods but reflects the limited experience we have gained so
far. Superposition, mesh and nodal analysis will continue to apply, to be our most basic
tools, throughout the book: in the analysis of linear circuits containing storage elements
(Chapters 5 to 7), those studied in the phasor domain (Chapters 8 to 11), the s-domain 6Q
(Chapters 12 to 16), and the Fourier domain (Chapter 17).
FIGURE P4.6
fiGURE P4.7
FIGURE P4.3
15 V
4.8. Find v by superposition. Check using Thevenin-
4.4. If i = 1 A, what must Vg be? Use this to determine Norton transformations.
i if Vg = 1 kV.
FIGURE P4.1
+
4.2. What does the principle of proportionality predict will v 6Q
happen to a linear circuit if all independent sources are fiGURE P4.11
scaled by zero?
4.3. Rl and R2 are linear resistors, but RN L satisfies
i = e- v . Demonstrate that this circuit violates the pro- 4.12. Use superposition to find v. Check by using
portionality principle. fiGURE P4.4 fiGURE P4.8 Thevenin-Norton transformations.
18 V v
+
v
1-----...., ~l_----
15A 2n
50n
FIGURE P4.12
fiGURE P4.19 FIGURE P4.22
4.13. What should we replace the 6-A source function by fiGURE P4.16
in Problem 4.18 in order that VI = O? 4.23. Using nodal analysis, find i.
4.20. Find V and i using nodal analysis.
4.14. Using nodal analysis, find VI and V2.
3kn
+ +
i
L---------~--~~--~~-i+ -~~---------"
20V
fIGURE P4.23
fiGURE P4.14 fiGURE P4.20
fiGURE P4.17
4.24. Repeat Problem 4.17 for the circuit shown.
4.15. U sing nodal analysis, find i.
4.21. Using nodal analysis, find v.
12kn
i 2kn
~ lA
+ v-
fiGURE P4.24
3kn 2n
4.25. Replace the 8-V independent source in Problem 4.24
with an 8i CCVS (same reference direction as shown).
Find i.
fiGURE P4.18 fiGURE P4.21
fiGURE 1'4.15 4.26. Show that by the very definition of node voltages,
KVL around every loop is satisfied automatically.
4.27. Show that by the very definition of mesh currents,
4.16. U sing nodal analysis, find i and i I. 4.19. Repeat Problem 4.17 for the circuit shown. 4.22. Using nodal analysis, find v. KCL at each node is satisfied automatically.
400 1 kO
+
20
50
FIGURE P4.44
40 20 40
fiGURE P4.35
4.45. Repeat Problem 4.44 for this circuit. R = 10 kQ.
fiGURE P4.28
10 10 +
4.40. Solve Problem 4.56 by mesh analysis.
4kO
4.41. Solve Problem 4.23 using nodal analysis, and solve R
again using mesh analysis.
3mA
4.42. Using the virtual short and open principles, find V2
25V in terms of VgI and Vg2' fiGURE P4.45
r-------+---{+ -l---~
~i 4kO
4kO SkO
FIGURE P4.29 4.46. Repeat Problem 4.45, but add a lO-kQ resistor be-
2kO tween a and h.
3kO
+
4.30. Use mesh analysis to find the indicated variables.
SPICE Problems
Solve the mesh equations by Cramer's rule. V2
4.47. Use SPICE to solve Problem 4.53.
fiGURE P4.37 4.48. Use SPICE to solve Problem 4.23.
20 20
FIGURE P4.42 4.49. Use SPICE to solve Problem 4.28. Do a dc sweep
from 0 to 12 V on the 6-V source, with I-V increments.
4.38. Find il and i2 using loop or mesh analysis. 4.50. Use SPICE to solve Problem 4.44. Replace the op
so 4.43. Repeat Problem 4.42 for the circuit shown. amp by the improved op amp model of Fig. 3.9, with Ri =
1 MQ, Ro = 30 Q, and A = 105 .
2i2 V IOkO 4.51. Use SPICE to solve Problem 4.46. Replace the op
r - - - - - < - +>------, amps as directed in Problem 4.50.
FIGURE P4.30 +
4.52. This problem is intended to check how close the
so 400 input current and voltage to an op amp are to zero (as the
V2
4.31. Repeat Problem 4.30 using matrix inversion. virtual open and short principles suggest). In the circuit of
4.32. Repeat Problem 4.30 using Gauss elimination. Problem 4.44, replace each op amp by its improved model
4.33. Repeat Problem 4.30 if the 2-V independent source (Fig. 3.9) with Ri = 1 MQ, Ro = 30 Q, and A = 105 . Let
is replaced by a VCVS (same reference direction) with son 64V VI = 1 V. Use SPICE to determine the currents into each
FIGURE P4.43 op amp and the voltages across these inputs.
source function -2VI.
4.34. Solve Problem 4.3 using mesh analysis.
4.44. Use the virtual short and open principles to find the More Challenging Problems
4.35. Using mesh or loop analysis, find the power deliv-
ered to the 8-Q resistor. voltage transfer ratio V2!VI. 4.53. Use proportionality to find il and i2.
fiGURE P4.38
20 10
12 V
f~
L-_ _ _ ~ ___ ~ _ _ __ L_ _ _ ~
~~ ~----I + -}------,
fiGURE P4.53 60 30
60
SA t FIGURE P4.57
Michael Faraday
1791-1867
180 60 4.58. Use the virtual short and open principles to find the My greatest discovery was
voltage transfer ratio vdvI. Michael Faraday.
Sir Humphry Davy
50kO
fiGURE P4.54
...............................................
4.55. Use superposition to find v. For each component
+ . .. . . . 10 :ch;ij~r~l;J!'Qf. wasbo:rn,near LondoIl." He
was.'flrsta:P:pl'~Ilticed'toa.. 2~.
problem, use the proportionality principle; that is, assume
that the component v = 1 V. . :at(age he realized. his boyhood
dream by bedmring' as~istarttatthe RoyalIn~titutipn to his idol; the great chenrlst
Sir Humphry lJavy:H~ rem~nedatthelnstitutionfor 54 years~ taking over
fiGURE P4.58 Davy's position when D,!-vy retired. Fara.day was perhaps the greatest experi-
120 mentalist who ever lived, with achievements to his credit in nearly all the areas
of physical science under investigation in his time. To describe the phenomena
4.59. Use the virtual short and open principles to find the he investigated, he and a science'-philosopher friend invented new words such
voltage transfer ratio V2/VI. as electrolysis, electrolyte, ion, aIlode, and cathode. To honor him, the unit of
fiGURE P4.55
capacitance is named the farad.
4.56. Use nodal analysis to find v for the circuit shown.
50kO
+ -4A +
v f +
lOkQ Vz
50kO
+
III 5.1 Capacitors III 5.7 DC Steady State la +qC
++++++++++
III
III
5.2
5.3
Energy Storage in Capacitors
Series and Parallel Capacitors
III 5.8 Practical Capacitors and
Inductors
v= Vab
t0
-----------
Test charge
-qC
III 5.4 Inductors III 5.9 Singular Circuits Ib
0
III 5.5 Energy Storage in Inductors III Summary
III Problems FIGURE 5.1 Parallel-plate capacitor.
III 5.6 Series and Parallel Inductors
plate we are leaving. Thus, from the definition of potential difference between two points
in space in Chapter 2 as the net work done in transporting a +1-C test charge between
the points, the plate with the +q charge is at a higher potential than the other one; that is,
Vah = f(q) (5.la)
where apparently the function f is an increasing one, since the larger the separated charge
Up to now we have considered only resistive circuits, that is, circuits containing resistors q, the greater the work done on the test charge.
and sources (and op amps, which can be replaced by models consisting of resistors and For many physical capacitors, the general charge-voltage relationship (S.la) is
sources). The terminal characteristics of these elements are simple algebraic equations linear. A linear capacitor is defined as a two-terminal device whose charge-voltage
that lead to circuit equations that are algebraic. In this chapter we introduce two important relationship is a straight line intersecting the origin, Vah = f (q) = K q, or
dynamic circuit elements, the capacitor and the inductor, whose terminal equations are
q = Cv (5.lb)
integrodifferential equations (involve derivatives and/or integrals) rather than algebraic
equations. These elements are referred to as dynamic because they store energy that can Here v = Vah is the voltage between the plates, and C = 1/K is the slope of the graph
be retrieved at some later time. Another term used, for this reason, is storage elements. plotting q as a function of v. C is known as the capacitance of the device, with the
We first describe the property of capacitance and discuss the mathematical model of dimension of coulombs per volt. The unit of capacitance is known as the farad (F), named
an ideal capacitor. The terminal characteristics and energy relations are then given, for the famous British physicist Michael Faraday (1791-1867). A I-farad capacitor will
followed by equivalents for parallel and series connections of two or more capacitors. separate 1 coulomb of charge for each volt of potential difference between its plates.
We then repeat this procedure for the inductor. The chapter concludes with a discussion It is interesting to note in the example above that the net charge within the capacitor
of practical capacitors and inductors and their equivalent circuits. is always zero. Charges removed from one plate always appear on the other, so the total
charge remains zero. We should also observe that charges leaving one terminal enter
the other. This fact is consistent with the requirement that current entering one terminal
must exit the other in a two-terminal device, a consequence of KCL. Since the current
is defined as the time rate of change of charge, differentiating (S.lb), we find that
A capacitor is a two-terminal device that consists of two conducting bodies that are
separated by a nonconducting material. Such a nonconducting material is known as an dv
insulator or a dielectric. Because of the dielectric, charges cannot move from one con- i=C- (5.2)
dt
ducting body to the other within the device. They must therefore be transported between
the conducting bodies via external circuitry connected to the terminals of the capacitor.
One very simple type, the parallel-plate capacitor, is shown in Fig. 5.1. The conducting which is the current-voltage relation or terminal law for a capacitor.
bodies are flat, rectangular conductors that are separated by the dielectric material. The circuit symbol for the capacitor and the current-voltage convention that satisfies
Starting with a capacitor that is initially uncharged, suppose that by means of some (5.2) are shown in Fig. 5.2. It is apparent that moving a charge of +q in Fig. 5.1 from
external circuit we have transferred q coulombs of charge from one plate to the other. To the lower to the upper plate represents a current flowing into the upper terminal and
determine the potential difference which results, we imagine a + l-C test charge subse- thence onto the upper plate. The movement of this charge causes the upper terminal to
quently being moved between points b and a in Fig. 5.1. Since the test charge is positive become more positive than the lower one by an amount v. Hence the current-voltage
and there is +q charge on the plate toward which we are moving, work must be done to FIGURE 5.2 Circuit convention of Fig. 5.2 is satisfied. Thus, in order that the terminal law (5.2) be satisfied,
move it. Equal work must be done on it due to attractive force of the -q charge on the symbol for a capacitor. the current and voltage reference directions must be related as shown in Fig. 5.2, with
where veto) = q (to) /C is the voltage on C at time to. In this equation, the integral
In (5.2) we see that if v is constant, the current i is zero. Therefore, a capacitor
term represents the voltage that accumulates on the capacitor in the interval from to to t,
acts like an open circuit to a dc voltage. On the other hand, the more rapidly v changes, whereas veto) is that which accumulates from -00 to to. The voltage v( -00) is taken to
the larger the current flowing through its terminals. be zero. Thus an alternative form of (5.3a) is
v(V)
, : : t <
In applying (5.3b), we are obtaining the area associated with a plot
of i from -00 to t. In Fig. 5.3, for example, since v( -00) = and
v= 1 at, t ::: a-I C = 1 F, we have
1, t > a-I
vet) = -
1
lit -00
(0) dr + v( -00) = 0, t:::o
- - - - ' L . - - - - - ' - - - - - - - - - : 7 t(s)
I If this voltage is applied to the terminals of a I-F capacitor (an un- Therefore, v(O) = 0, and
a usually large value chosen for numerical convenience), the resulting
i(A)
(a)
current (in amperes) is
vet) = 1t a dr + v(O) = at,
Suppose that the capacitor is disconnected from any external circuit so that no additional
current can flow, and the charge, voltage, and energy all remain constant. If we subse-
quently connect a resistor across the capacitor, a current flows through it as the charges
EXERCISE 5.1.2
separated on the plates seek one another, until all the stored energy (50 J) has been
dissipated in the form of heat by the resistor and the voltage across the RC combination
5.1.3. A constant current of 10 rnA is charging a 10-/LF capacitor (en-
settles to zero. Systematic analysis of such circuits is found in Chapter 6.
tering its positive voltage terminal). If the capacitor was initially charged
As was pointed out earlier, the voltage on a capacitor is a continuous function of
to 5 V, find the charge and voltage on it after 20 ms.
Answer 0.25 me; 25 V time. Thus, by (5.4), we see that the energy stored in a capacitor is also continuous.
This is not surprising, since otherwise, energy would have to be transported from one
5.1.4. Let the graph of Exercise 5.1.2 be the graph of the current in
~lace to another in zero time. Since power is the rate of energy change per unit time,
milliamperes versus time in milliseconds in a 0.25-/LF capacitor. Find the
voltage at t = -4, -1,6, and 9 ms. mstantaneous change of energy demands infinite power, a physical impossibility.
Answer -90, -190, -56, and 0 V To illustrate continuity of capacitor voltage, let us consider Fig. 5.4, which contains
an ideal switch that is opened at t = 0, as indicated. An ideal switch is defined to
be a two-terminal element that switches between an open circuit and a short circuit
instantaneously. Unless otherwise specified, all switches used in this book are considered
to be ideal.
To discuss the effect of the switching action, we first need to consider two times
near t = O. We shall denote as t = 0- the time just before the switching action and
The terminal voltage across a capacitor is accompanied by a separation of charges between t = 0+ as the time just after the switching action. Theoretically, no time has elapsed
the capacitor plates. These charges have electrical forces acting on them. An electric between 0- and 0+, but the two times represent radically different states of the circuit.
field, a basic quantity in electromagnetic theory, is defined as the position-dependent Vc (0-) is the voltage on the capacitor just before the switch acts, and Vc (0+) is the
force acting on a unit positive charge. Thus the forces acting on the charges within the voltage immediately after switching. Similarly, ic(O-) and ic(O+) are the capacitive
capacitor can be considered to result from an electric field. It is for this reason that the current just before and just after the switch changes state. More formally, vc(O-) is
energy stored or accumulated in a capacitor is said to be stored in the electric field located the limit of vc(t) as t approaches zero from the left [through negative (t < 0) values],
between its plates. Quantitatively, the energy stored in a capacitor, from (1.6) and (5.2), and vc(O+) is the limit as t approaches zero from the right [through positive (t > 0)
is given by values]. Indeed, mathematically, the times 0- and 0+ are convenient fictions without
precise definition (when is the first instant greater than zero?), although the values of
wc(t) = II
-00
vi dT = It
-00
v(c dV ) dT
dT
circuit variables such as Vc (0-) and Vc (0+) are defined precisely by the limits from the
left and right.
178 Chapter 5 Energy-Storage Elements Section 5.2 Energy Storage in Capacitors 179
Suppose that in Fig. 5.4 we have Vs = 6 V and vc(O) = 4 V.
Just prior to the switching action (t = 0-), we have VI (0-) :::
TQRS
Vs - vc(O-) = 2 V. Immediately after the switch is opened, we
In this section we determine the equivalents for series and parallel connections of ca-
have VI (0+) = 0, since no current is flowing in RI (it is in series
pacitors. Recall that two subcircuits are equivalent if they have the same terminal law.
with an open circuit). However, since vc , a capacitive voltage, must
We shall match the terminal law for series-connected capacitors with that for a single
+
be continuous, we have
capacitance, the series equivalent capacitance, and then repeat for the parallel case.
v, C
Let us first consider the series connection of N capacitors, as shown in Fig. 5.5(a).
Vc(O+) = vc(O-) = 4 V
Applying KVL, we find
Thus the voltage on resistor RI has changed abruptly, but that on the V= VI + V2 + ... + VN (5.5)
capacitor has not. The voltage on R2 is the same as the capacitor
fiGURE 5.4 Circuit illustrating continuity of Using the terminal law (5.3a) gives
voltage and thus has not changed abruptly either.
capacitor voltage.
\
vet) = (~ t
CI lto
i dr + VI (to)
Although apparently prohibited by the continuity principle, circuit diagrams can
be drawn for which capacitor voltages seem forced to change abruptly. For example,
consider two capacitors and an ideal switch forming a closed loop. If the capacitors
+ (~
C2
t i dr + V2(tO) + ... + (_1
lto
t i dr + VN(tO)
C lto N
have unequal voltages just as the switch closes, by continuity their values must remain
unequal at t = 0+, while by KVL their voltages must be equal at t = 0+. Apparently,
we must either concede an infinite power flow (by accepting violation of the continuity
= (t~)
C
n=1
t i dr + veto)
lto n
principle) or abandon KVL for these circuits. We shall consider such singular circui!s
where in the last we have used (5.5). This is the desired terminal law for the series-
in more detail in Section 5.9. Singular circuits are similar to the circuits discussed III
connected capacitors. The terminal law for Fig. 5.5(b) is simply
Chapter 2 that contain unequal voltage sources in parallel or current sources in series.
Such circuits lead to mathematical contradictions and cannot be analyzed consistently.
In practice, the contradiction is removed by replacing ideal elements by more accurate vet) = -1 [t i dr + veto)
Cs t
physical models of them, including Thevenin equivalent resistance in the case o~ the
The two terminal laws are identical, hence the circuits are equivalent, if
sources of Chapter 2 and parallel leakage resistance in the present case of capacitors.
In the real world, continuity and the Kirchhoff laws are both valid in lumped physical
circuits.
1 N 1
-=L-
C C s n=1 n
(5.6)
EXERCISES
The equivalent of a chain of series-connected capacitors is a single capacitor whose in-
verse capacitance is the sum of inverses of the series capacitances.
5.2.1. A 0.2-/LF capacitor has a charge of 20 /LC. Find its voltage and
energy. i + VI _
Answer 100 V; 1 mJ ~
5.2.3. !
In Fig. 5.4, let C = F, RI = R2 = 4 Q, and V = 20 V. If the
current in R2 at t = 0- is 2 A directed downward, find at t = 0- and at
t = 0+ (a) the charge on the capacitor, (b) the current in RI directed to the o
right, (c) the current in C directed downward, and (d) dvc/dt.
(a) (b)
Answer (a) 2, 2 C; (b) 3, 0 A; (c) 1, -2 A; (d) 4, -8 VIs
fiGURE 5.5 (a) Series capacitors; (b) equivalent circuit.
180 Chapter 5 Energy-Storage Elements Section 5.3 Series and Parallel Capacitors 181
In the case of just two capacitors in series, this general result simplifies to Determine the equivalent capacitance Ceq of the circuit shown in
Fig. 5.7(a).
I
For two capacitors in series, the equivalent capacitance is the product over the sum of
2 F L l2"F
the two individual capacitances. The product-by-sum equivalence rule was introduced J1, TL_.----'T ~
in (2.15) for the case of two resistors in parallel. As with parallel resistors, this rule is
limited to the special case of exactly two such elements. For three or more capacitors in I L II II
series, the general result (5.6) should be used. "3 J1,F T -3
L ___ -J+------'T "3 J1,F
We next turn to the parallel connection of N capacitors, as shown in Fig. 5.6(a).
KCL at the top node yields
1 L 1 I II II
"3 J1,F TL_"3_J1,_F_T-L-r--1T_"3_J1,F_---,T "3 J1,F
dv dv dv
i = CI - +C2 - + ... +CN - I
dt dt dt
(a)
or i = ( LC
N
n=1
n
) dv
-
dt
fiGURE 5.7 Circuit for Example 5.5.
The corresponding terminal law for Fig. 5.6(b) is Since parallel capacitances add, Fig. 5.7(a) may be immedi-
ately simplified to its equivalent, Fig. 5.7(b). Then, using the series
equivalent rule (5.6), with p,F units,
1 1 1 1
=-+-+-=2
Comparing the last two equations, the terminal laws are identical and the circuits equiv- Ceq 4 1 ~
alent if
or Ceq = ~ p,F.
~;'
-7-
I1t 1t
+
E;' E;' :lc Vl(t) = -
CI to
i(r)dr + VI (to) =
0
(3 cos 2r)dr + 10
f' T
v
(a)
C
'
f' ~'
(b)
at t = 0 gives the correct initial value for VI (t). We may also check
that
i = C 1 -dVI = 1-
dt
d
dt
(3 . )
- sm 2t + 10 = 3 cos 2t A
2
fiGURE 5.6 (a) Parallel capacitors; (b) equivalent circuit. which is consistent with our computed value for i(t).
182 Chapter 5 Energy-Storage Elements Section 5.3 Series and Parallel Capacitors 183
11
IF 2F 5.3.4. Derive an equation for voltage division between two initially un-
charged series capacitors by finding VI and V2.
Answer C2V/(C1 + C2); CI v/(C] + C2)
' 2F
(a) (b)
It is interesting to note that the equivalent capacitance of series and parallel capaci-
tors follows the same rules as equivalent conductance of series and parallel conductances. EXERCISE 5.3.4
That is, for the parallel case the equivalent parameter (conductance G p or capacitance
C p) is the sum of the individual parameters, while in series the equivalent parameter (G s
or C s ) is the inverse of the sum of inverses of the individual parameters. This correspon-
dence follows from the fact that the terminal laws for the conductance and capacitance
are of the same form: the parameter (G or C) multiplies the terminal voltage (or deriva-
tive of terminal voltage) to yield terminal current. This functional correspondence makes In the preceding sections we found that the terminal characteristics of the capacitor are
the foregoing derivations of capacitive equivalents correspond line for line to those for the result of forces that exist between electric charges due to their separation in space.
conductance equivalents presented earlier in Chapter 2. Just as these charges exert position-dependent or electrostatic force upon one another, so
moving charges, or currents, exert electrodynamic force.
The force that is exchanged by two neighboring current-carrying wires was deter-
mined experimentally by the French scientist Andre Marie Ampere (1775-1836) in the
EXERCISES early nineteenth century. This force can be characterized by postulating the existence of
+ a magnetic field. The magnetic field, in tum, can be thought of in terms of a magnetic
5.3.1. Find the maximum and minimum values of capacitance that can flux that forms closed loops about electric currents. The cause of the flux is motion of
be obtained from ten l-{tF capacitors. How should they be connected? charge in the wires, that is, the current. The study of magnetic fields, like that of electric
v
Answer 10 {tF (parallel); 0.1 {tF (series connected) fields mentioned in Section 5.3, comes in a course devoted to electromagnetic theory.
5.3.2. Find the equivalent capacitance. Here we are interested primarily in the role of magnetic fields in linear lumped circuits,
Answe r 20 {tF most prominently in the circuit element called an inductor.
An inductor is a two-terminal device that consists of a coiled conducting wire
wound around a core. A current flowing through the device produces a magnetic flux
that forms closed loops threading its coils, as shown in Fig. 5.9. Suppose that the coil
FIGURE 5.9 Simple contains N turns and that the flux passes through each tum. In this case the total flux
model of an inductor. linked by the N turns of the coil, denoted by A, is
A=N
This total flux is commonly referred to as the flux linkage. The unit of magnetic flux
is the weber (Wb), named for the German physicist Wilhelm Weber (1804-1891). In a
linear inductor, the flux linkage is directly proportional to the current flowing through
the device. For linear inductors, therefore, we may write
EXERCISE 5.3.2
A = Li (5.8)
5.3.3. Derive an equation for current division between two parallel ca- where L, the constant of proportionality, is the inductance in webers per ampere. The
pacitors by finding il and i2. unit of 1 Wb/A is known as the henry (H), named for the American physicist Joseph
EXERCISE 5.3.3 Answer Cli/(CI + C2); C2i/(CI + C2) Henry (1797-1878).
di
a 1----....,
v=L- (5.9)
dt
_ _ _..l-_ _----b======~
~ t (s)
(b)
Clearly, as i increases, a voltage is developed across the terminals of the inductor,
+ the polarity of which is shown in Fig. 5.10. This voltage opposes any additional increase fiGURE 5.11 Current and voltage waveforms in a l-H inductor.
in i, for if this were not the case, that is, if the polarity were reversed, the induced voltage
v would "aid" the current. This physically cannot be true because the current and voltage
is made larger, v changes more rapidly and v increases. It is apparent
would then both increase indefinitely.
that if a-I = (a is infinite) i changes abruptly (in zero time) from
The circuit symbol for the inductor whose terminal law is given in (5.9) and shown
in Fig. 5.10. Just as in the cases of the resistor and the capacitor, the voltage and current to 1 V.
reference directions are assumed to satisfy the passive sign convention. If either reference
fIGURE 5.10 Circuit direction (but not both) is reversed, a negative sign must be introduced into the terminal In general, abrupt changes in the current through an inductor, such as in Exam-
symbol for an inductor. law, yielding v = -L(di/dt). ple 5.7, require that an infinite voltage appear across the terminals of the inductor. As
The terminal law of the inductor, (5.9), shows that if i is constant, the voltage v is observed in the case of the capacitor, infinite current or voltage is a physical impossi-
zero. Therefore, an inductor acts like a short circuit to a dc current. On the other hand, bility. Thus instantaneous changes in the current through an inductor are not possible.
the more rapidly i changes, the greater is the voltage that appears across its terminals. We observe that the current through an inductor is always continuous. This is the other
This is due to the more rapidly time-varying magnetic flux i produced, which is the cause half of the continuity principle, shown previously to apply to capacitive voltage. Induc-
of the induced voltage v across the terminals of the inductor. tive current is also continuous, even though the voltage across an inductor may change
discontinuously, as may capacitive current.
Consider a current that increases linearly from
given by
to 1 V in a-I s, Another interesting parallel to our previous discussion of the capacitor is revealed
by comparing Example 5.7 with Example 5.2. We see that by reversing the roles of i
and v the same terminal pair is produced in the inductor as in the capacitor. This is less
1at,,1, t ~ t ~
i =
t < a-I
surprising when we note that the terminal laws for these two elements, i = C(dv/dt)
and v = L(di/dt), are of the same form with only the current and voltage interchanged.
> a-I
This duality of capacitor and inductor will make the remaining results for the inductor
If this current is applied to the terminals of a I-H inductor, the strongly reminiscent of the previous ones for the capacitor.
resulting voltage is Let us find the current i (t) in terms of the voltage vet) for the inductor. Integrating
(5.9) from time to to t and solving for i(t), we have
0, t<o
v = { a, t > t
~
a-I
~ a-I
0,
Plots of i and v are shown in Fig. 5.11. We see that v is zero when
i is constant and that it is equal to a when i increases linearly. If a
1/ (
L -00
taining the net area under the graph of v from -00 to t, since i (to)
represents the area from -00 to to . Setting to = 0 in Fig. 5.11, for
instance, we can compute the current at t = l/a to be
i(~) =~fol/a la
v(r)dr+O= fo / adr=lA
= 1 i (t)
i(-oo)
Li di = !Li 2 (t) (5.llb)
188 Chapter 5 Energy-Storage Elements Section 5.5 Energy Storage in Inductors 189
EXERCISES
. Comparing the terminal law from Fig. S.13(b), v
equivalent if
Ls(di/dt), the two circuits are
i(t) = (t ~ ) t
n=l n lto
vdr +i(to)
In this section we determine the equivalent inductance for series and parallel connections Comparing the terminal law of Fig. S.14(b) given in (S.lO), they match if
of inductors. Just as we found the series and parallel capacitive equivalents to follow the
rules derived for conductances, we will now see that series and parallel inductors have
equivalents whose values follow the same rules we derived for resistors.
1 N 1
First consider the series connection of N inductors shown in Fig. S.13(a). Summing
voltage drops around the loop and using the terminal law v = L(di/dt) gives -=L-
Lp Ln n=l
(5.13)
di di di
v=L 1-+L2-++LN-
dt dt dt
The equivalent of parallel inductors is a single inductor whose inverse inductance is the
sum of the inverses of the parallel inductances.
i i
~
VI _
i
V2 _ + + i ~ ~
~
~ ... +
+ +
LI L2 + i1t i2t iNt
+ V Lp
V L1 L2 LN
V LN VN V L,
(a) (b)
(a) (b)
FIGURE 5.14 (a) Parallel inductors; (b) equivalent circuit.
FIGURE 5.13 (a) Series inductors; (b) equivalent circuit.
190 Chapter 5 Energy-Storage Elements Section 5.6 Series and Parallel Inductors 191
In the case of two parallel inductors L1 and L 2, (5.l3) may be simplified to 5.6.3. Derive an equation for voltage division between two series induc-
tors by finding V1 and V2.
Lp = L1L2 Answer L1V/(L] +L2); L2V/(L] +L2)
L1 +L2
which is directly analogous to the equivalence of two parallel resistances (2.15). As
mentioned previously in that case, the product-by-sum rule only holds for exactly two
elements. For three or more, the general result (5.l3) should be applied.
v
We wish to find v in Fig. 5.15. We will first find an equivalent
inductor. The two 1-H inductors in series are equivalent to a 2-H
inductor, which is in parallel with the 2-H inductor shown in the
figure. Their parallel equivalent is
EXERCISE 5.6.3
(2)(2) \
~H
3 IH L eq1 = - - = 1 H
2+2 5.6.4. Derive an equation for current division between two parallel in-
ductors with no initial current by finding il and i2
L eq1 is in tum in series with ~ H, yielding a series equivalent of L2i . Lli
~ H. Finally, this is in parallel with the l-H inductor for an overall Answer
Ll +L2' L] +L2
6tA IH . I 2
eqUlva ent Leq:
1 1 1 3 2
-=5'+5'=-+-=1
Leq 3' 2: 5 5
FIGURE 5.15 Circuit for Example 5.10. and Leq = 1 H. Then, for this circuit,
d
v=1-(6t)=6V
dt
EXERCISE 5.6.4
EXERCISES
5.6.1. Find the maximum and minimum values of inductance that can
be obtained using five 20-mH inductors and ten lO-mH inductors.
Answer 200 mH; 800 p,H
5.6.2. Find the equivalent inductance. Inductor values shown are in Before introduction of capacitors and inductors, we focused on resistive circuits, those
millihenries. containing only resistors and sources. In resistive circuits whose independent sources are
Answer 10 mH all dc (constant values), it is significant that all currents and voltages will also be dc. This
follows from the fact that nodal or mesh analysis equations will contain no time-varying
2 4 7 quantities. The solutions of these equations will clearly be constants, that is, dc values.
Substituting these constant node voltages or mesh currents into Ohm's law to find the
remaining circuit variables will produce only more dc currents and voltages.
A circuit is said to be in de steady state if all currents and voltages in that circuit are
3 6
constant. Clearly, only circuits whose independent sources are all dc can possibly attain dc
steady state, for if some source function were not constant, that current or voltage would
not be dc. From Section 5.6, resistive circuits containing only dc independent sources are
always in dc steady state. However, the introduction of derivatives and integrals into the
EXERCISE 5.6.2 analysis equations by inductors and capacitors makes time-varying solutions a distinct
36V 2H 36V
EXERCISES
2i + 4(i + 6) = 36
or i = 2 A. Then, by KVL, EXERCISE 5.7.1
v = 4(i + 6) = 4(8) = 32 V
5.7.2. Assume that this circuit is in de steady state just before t = O.
The assumption of dc steady state in RLC circuits (those with resistors, inductors, Find v(O-) and v(O+). Is the circuit in de steady state just after t = O?
and capacitors), which contain dc sources only, thus leads to self-consistent results. Answer 1 V; 0 V; no
Replacing inductors and capacitors by shorts and opens, we are left with a resistive
circuit to analyze and are thus guaranteed a dc steady-state outcome of the analysis. lQ
Self-consistency, however, only proves that dc steady state is possible with such circuits,
not that it is inevitable. As discussed in Chapter 6, most RLC circuits with dc sources
eventually converge to dc steady state. But some never do; and even for those that do,
the next example shows that they typically do not "start" in dc steady state. IA IF
circuits. Like practical capacitors, they dissipate a small but nonzero amount of power.
This dissipation results from losses associated with the fact that the wire making up the
,inductor coil has some resistance and also from core losses (losses due to currents induced
EXERCISE 5.7.3 in the core by the strong magnetic field present there). The magnetic flux is best focused
in the inductor core by the use of ferromagnetic materials, but this also increases the core
losses. These losses may be modeled by inserting a series resistor into the circuit model
for a practical inductor as shown in Fig. 5.18.
5~.8 eR,j\CTICAlCAPACITORS AND INDUCTORS The unavailability of IC inductors, a consequence of the difficulty of reproducing
coils on flat surfaces and also of the poor magnetic properties of most IC materials, has
Commercially important capacitors are produced either as discrete elements or are de- limited the use of inductors in modem miniaturized circuit design. Much effort has been
posited onto integrated-circuit (IC) substrates during the IC manufacturing process. Dis- FIGURE 5.18 Circuit devoted to finding inductorless designs for IC applications. There are still many electronic
crete capacitors are available in a wide variety of types, values, and voltage ratings. The model for a practical applications, however, where inductors are commonplace, ranging from telephone circuits
capacitor type is generally classified by the kind of dielectric used, and its capacitance inductor. and radio receivers to power supplies and electric motors.
is determined by the type of dielectric and the physical geometry of the device. The
voltage rating, or working voltage, is the maximum voltage that can safely be applied
to the capacitor. Voltages exceeding this value may permanently damage the device by EXERCISES
destroying or breaking down the dielectric.
Simple discrete capacitors may be constructed using two strips of metal foil sep- 5.S.1. Suppose that we wish to construct a 1-F capacitor by putting 10-
arated by a dielectric insulating material. The foil and dielectric are pressed together /LF capacitors in parallel. How many do we need? If the leakage resistance
into sheets that are then rolled or folded into a compact package. Electrical conductors is 10 MQ for each lO-/LF capacitor, what will the total leakage reSIstance
attached to each metal-foil sheet constitute the terminals of the capacitor. and current be if the 1-F capacitor is put across a 100-V de source?
Unlike ideal capacitors, practical capacitors dissipate small but nonzero amounts of Answer 100,000; 100 Q; 1 A
power. This is due primarily to leakage currents that flow within the dielectric material 5.S.2. Suppose that we create a 50-mH inductor by winding 10,000 turns
in the device. Practical dielectrics have a nonzero conductance that allows charge to of thin wire around a I-em-diameter form. If the wire has a resistance of
leak directly from one capacitor plate back to the other (internally, in addition to the 0.01 Q/cm, what power will be dissipated by the inductor if it carries a
nonleakage current moving through the external circuit path). Leakage current may be current of 20 rnA de? What would it be if this practical inductor was an
included in a circuit model for a practical capacitor by placing a resistance in parallel ideal inductor?
Answer 40n mW; 0 mW
with an ideal capacitor, as shown in Fig. 5.17.
Common types of discrete capacitors, designated by their dielectric, include ce-
ramic, Mylar, Teflon, and polystyrene. In addition, larger capacitive values are usually of
the electrolytic type requiring polarized use (one terminal of the capacitor must be kept
at a higher potential than the other). The materials and structures of integrated-circuit
capacitors depend on the IC technology being used, frequently consisting of a thin layer
of an insulating oxide dielectric between layers of doped semiconductor materials acting A circuit in which a switching action takes place that appears to violate the continuity
as the conducting plates. principle by producing discontinuities in capacitor voltages or inductor currents is called
Practical inductors are available only as discrete elements (or preas sembled into a singular circuit. In this section we consider such circuits and show how the apparent
electronic packages containing many discrete elements), not as constituents of integrated violation may be consistently interpreted and resolved.
41:2
FIGURE 5.17 Circuit model for a practical capacitor.
circuits. Like practical capacitors, they dissipate a small but nonzero amount of power.
This dissipation results from losses associated with the fact that the wire making up the
\ inductor coil has some resistance and also from core losses (losses due to currents induced
EXERCISE 5.7.3 in the core by the strong magnetic field present there). The magnetic flux is best focused
in the inductor core by the use of ferromagnetic materials, but this also increases the core
losses. These losses may be modeled by inserting a series resistor into the circuit model
for a practical inductor as shown in Fig. 5.18 .
.5~Q<PRAClICALCAPACITORS AND INDUCTORS The unavailability of IC inductors, a consequence of the difficulty of reproducing
coils on flat surfaces and also of the poor magnetic properties of most IC materials, has
Commercially important capacitors are produced either as discrete elements or are de- limited the use of inductors in modern miniaturized circuit design. Much effort has been
posited onto integrated-circuit (IC) substrates during the IC manufacturing process. Dis- FIGURE 5.18 Circuit devoted to finding inductorless designs for IC applications. There are still many electronic
crete capacitors are available in a wide variety of types, values, and voltage ratings. The model for a practical applications, however, where inductors are commonplace, ranging from telephone circuits
capacitor type is generally classified by the kind of dielectric used, and its capacitance inductor. and radio receivers to power supplies and electric motors.
is determined by the type of dielectric and the physical geometry of the device. The
voltage rating, or working voltage, is the maximum voltage that can safely be applied
to the capacitor. Voltages exceeding this value may permanently damage the device by EXERCISES
destroying or breaking down the dielectric.
Simple discrete capacitors may be constructed using two strips of metal foil sep- 5.8.1. Suppose that we wish to construct a I-F capacitor by putting 10-
arated by a dielectric insulating material. The foil and dielectric are pressed together p,F capacitors in parallel. How many do we need? If the leakage resistance
into sheets that are then rolled or folded into a compact package. Electrical conductors is 10 Mr.l for each lO-p,F capacitor, what will the total leakage resistance
attached to each metal-foil sheet constitute the terminals of the capacitor. and current be if the I-F capacitor is put across a 100-V de source?
Unlike ideal capacitors, practical capacitors dissipate small but nonzero amounts of Answer 100,000; 100 r.l; I A
power. This is due primarily to leakage currents that flow within the dielectric material 5.8.2. Suppose that we create a 50-mH inductor by winding 10,000 turns
in the device. Practical dielectrics have a nonzero conductance that allows charge to of thin wire around a I-em-diameter form. If the wire has a resistance of
leak directly from one capacitor plate back to the other (internally, in addition to the 0.01 Wcm, what power will be dissipated by the inductor if it carries a
nonleakage current moving through the external circuit path). Leakage current may be current of 20 rnA de? What would it be if this practical inductor was an
included in a circuit model for a practical capacitor by placing a resistance in parallel ideal inductor?
Answer 40n mW; 0 mW
with an ideal capacitor, as shown in Fig. 5.17.
Common types of discrete capacitors, designated by their dielectric, include ce-
ramic, Mylar, Teflon, and polystyrene. In addition, larger capacitive values are usually of
the electrolytic type requiring polarized use (one terminal of the capacitor must be kept
at a higher potential than the other). The materials and structures of integrated-circuit
capacitors depend on the IC technology being used, frequently consisting of a thin layer
of an insulating oxide dielectric between layers of doped semiconductor materials acting A circuit in which a switching action takes place that appears to violate the continuity
as the conducting plates. principle by producing discontinuities in capacitor voltages or inductor currents is called
Practical inductors are available only as discrete elements (or preas sembled into a singular circuit. In this section we consider such circuits and show how the apparent
electronic packages containing many discrete elements), not as constituents of integrated violation may be consistently interpreted and resolved.
. ~ .(_t=o~~: + 1 we were to build this circuit physically. What would happen when we threw the switch
open? The power stored in the inductor's magnetic field would not instantaneously drop
to zero, as it would be required to do if the current instantaneously dropped to zero. The
current would seek another path. Indeed, we would probably be greeted with a quite
visible spark arcing across the air gap produced as the switch is opened. A model for
\ the conductivity of the air gap is included in Fig. 5.20(b). If conductance G is small, as
L -_ _ _ _ _ _ _v_2----lT c, = IF expected in practice (air is a poor conductor), then with i(O+) = 1 A by continuity, we
have V (0+) = 1/G, a large value. Indeed, this circuit suggests the design of automobile
breaker point ignition systems, in which the goal is to produce a very high voltage by
(a) (b)
interrupting the current through the ignition coil or inductor with a pair of moving breaker
FIGURE 5.19 (a) Singular circuit with capacitors; (b) improved circuit points. The inevitable arcing predicted by our analysis is what causes these points to
model. corrode and eventually fail.
Recall that our justification of continuity was based on the assertion that infinite G
currents and voltages are not possible in practical circuits. In the realm of pure mathe-
matics, there is nothing wrong with the concept of infinity. It serves us well in imagining
what the function f(x) = l/x approaches as x goes to zero, or what the sum of all t=o _. . ao--~ 1=0
+ v
positive integers up to N goes to as N gets bigger and bigger.
What goes wrong in singular circuits is that they are not useful models for prac-
tical circuits, those that can be approximated in practice in the real world by physical L L
means. They are similar to other impractical circuits we have identified previously, that
is, parallel voltage sources, or series current sources, with different source functions. We i
+-
earlier concluded that the terminal law for an ideal independent source, so useful in the
context of most circuits, clashes irreconcilably with the Kirchhoff laws for these peculiar (a) (b)
circuits. We agreed in that case to resolve the apparent contradictions when they arise by FIGURE 5.20 (a) Singular circuit with inductor; (b) improved circuit model.
use of a better model. In the case of the sources, this was done by adding a finite, nonzero
Thevenin or Norton equivalent resistance to the ideal source model, effectively eliminat-
ing the mathematical conflict. If we stubbornly insist on the original, inadequate source
models in these circuits, we simply have to admit an inherent mathematical contradiction
and refuse to analyze the self-contradictory circuit further. EXERCISES
In the case of singular circuits such as Fig. 5 .19(a), the apparent contradiction
between continuity and the Kirchhoff laws is also removed by using a better circuit
model. One such model takes the nonzero resistance of any practical wire into account 5.9.1. !
Suppose that i = A in Fig. 5.20(b). If the air gap conductance
by adding a series resistance into the loop, as shown in Fig. 5.19(b). Then we may have
G = 10 J1,S between the switch terminals, how much voltage is produced
across the air gap immediately after the switch is thrown open?
continuity satisfied, that is, VI (0+) = 1 and V2 (0+) = 0, and at the same time have KVL Answer 50 kV
satisfied, since the difference of these voltages will now be accounted for as a voltage
5.9.2. If an ideal 100-J1,F capacitor is charged to 100 V and then sud-
drop across the wire VR, where VR(O+) = VI (0+) - V2(0+) = 1 V. Note that if the wi:e
denly shorted with an ideal connecting wire, assuming that KVL is not
resistance R is small, as we might expect in practice, then by Ohm's law the current ill violated, how much energy must be instantaneously shed by the capaci-
the wire, i (0+) = V / R = 1/R, will be quite large. Indeed, in the limit as R gets smaller tor? In the same situation, using a practical model, we include 0.01-0
+
80 JLF FIGURE 1'5.15
'~]:~4
given in (b). Find the current i if i LCO) = -1 A for (a) 0 <
t < 1 sand (b) 1 < t < 2 s.
--I---JL--+--+--I--~~-I---/-----'~ t (ms)
2T T6 ~4 ---+-----1-----~-__7> t (s)
b o--j( I( - (a)
6 ll2
Vz(V)
FIGURE 1'5.16 FIGURE 1'5.21
2>l
TI~
i (0) = 1 A and that i and v together satisfy the
sign convention.
J~l
10 cos lOt rnA. Find (a) the terminal voltage, (b) the power,
(a)
v (V) (c) the stored energy, and (d) the maximum value of the
power being absorbed.
z
V
IF
18 V
+ 1F 5
4Q
2Q 2H
b 2Q + IH
FIGURE 1'5.42 lOA v2
FIGURE 1'5.36
FIGURE 1'5.39
5.43. Show that this circult does not have a dc steady
state. Why?
5.37. The inductances shown are all in mHo Find the 5.40. This circuit is in dc steady state at t = 0-. Find t= 0
equivalent inductance seen at terminals a-b. v(O-), v(O+), i(O-), and i(O+).
FIGURE 1'5.46
12V
I2V c L R
IQ
18 6 3
5.47. Assume that the circuit is in dc steady state at t =
0-. Find i(O-) and i(O+), in terms of Rand C.
FIGURE 1'5.43
3
FIGURE 1'5.37
FIGURE 1'5.40
12 V R c
+
6V v2
5.38. If v(O-) = 72 Y, find i], i2, ie, Vc , and v at t = 0-
and at t = 0+. 5.41. Find the dc steady-state currents i] and i2.
v2
t:rHt:rr--1t:l
10
unknowns should be mesh currents. Then repeat for nodal
equations.
12V
J J J
FIGURE P5.50
Joseph Henry
1797-1878
H. S. Carhart
Response We seek to determine v(t) for t > O. Applying KCL at the top node yields
II 6.3 General First-Order Circuits
without Sources dv v
II 6.9 Design of First-Order Circuits C - + - =0
dt R
II 6.4 Circuits with DC Sources II Summary
fiGURE 6.1 Source-free dv 1
II 6.5 Superposition in First-Order l1li Problems RC circuit. or - + -v = 0 (6.2)
dt RC
Circuits
which is a first-order differential equation. (The order of a differential equation is defined
\ as the order of the highest-order unknown derivative contained in the equation.)
Several methods are available for solving first-order differential equations. One
straightforward method is to rearrange the terms in the equation so as to separate the
variables v and t. These variables may be separated by rewriting (6.2) as
dv 1
In Chapter 4 resistive circuits were analyzed, and in Chapter 5 the two energy storage -=--dt
v RC
elements, capacitors and inductors, were introduced. In this chapter we analyze the class Then, taking the indefinite integral of each side, we have
of circuits that result when we add to resistive circuits a single storage element. Such a
circuit is called a first-order circuit because, as we shall see, the storage element results 1 fdt
dv = _ _
f
in a first-order differential equation characterizing the circuit. v RC
We concern ourselves first with simple RC and RL circuits that contain no inde- t
or In(v) = - - +k
pendent sources. Lacking such sources, any circuit response can only be the result of RC
initial capacitive or inductive energy stored within the circuit and is thus dictated by the with k the constant of integration. Exponentiating both sides, we have
nature of the circuit itself. For this reason the response is known as the natural response
of the circuit. The natural response is characterized by a single time constant or rate of v = K e-(t/RC) (6.4)
exponential decay. where e is the base of the natural logarithms, e = 2.71828 ... , and K = e+k.
Following our study of circuits without sources, we shall consider the total resp?~se For this solution to be valid in the interval of interest t > 0, continuity requires
of first-order circuits containing sources. We shall find that it consists of two addItIve that v(O) must match the specified initial condition v(O) = Vo. Thus while vet) in (6.4)
parts, a natural response, identical in form to the response in. the absence of s~urces, and will satisfy the differential equation for any K, for only a single value of K will this
ajorced response, governed by the form of the forcing functlOns (source fun~tlOns of the solution satisfy both differential equation and initial condition. This is found by forcing
independent sources). The case in which all independent sources are dc, that IS, constant, v(O) in (6.4) to the required value of Vo:
is considered in detail. Step and pulse sources and their responses are next discussed.
The chapter concludes with a look at how SPICE can be used to determine responses for v(O) = Ke o = K = Vo
these circuits. whence
v(t) = Voe-(t/RC)
We begin our study of first-order circuits by considering the simple one-loop circui~ co~ Note that all other circuit variables follow immediately from (6.5), although for the
taining just a capacitor and a resistor, as shown in Fig. 6.1. Suppose that the capac~t~r.ls moment we will keep our focus on the capacitive voltage vet).
charged to a voltage of Vo volts at an initial time, which we shall take as t = O. By Imt~al A graph of the circuit response vet) is shown in Fig. 6.2. The voltage begins at Vo,
time we mean the beginning of the time period we are interested in, not the time of the CIr- as required, and then decays exponentially to zero with time. The rate at which it decays
cuit's construction (the circuit must have a history extending beyond our initial time, since to zero is given by the RC product of the circuit. Since this response is governed by
the value of Vo volts must have been established by external circuitry before this time). the circuit elements themselves and not by some independent source "forcing" a different
208 Chapter 6 First-Order Circuits Section 6.1 Simple RC and RL Circuits without Sources 209
v(t) Then, since v(t) is also the voltage across the resistor,
o v 2 (t)
PR(t) = - - = 0.02Se-2000t W
fiGURE 6.2 Voltage v(t) in the circuit of Fig. 6.1. R
Thus the maximum power of 2S mW is dissipated right at the initial
behavior during t :::: 0, this response is called the natural response of the circuit. time t = O. This is confirmed by the graph in Fig. 6.2, which shows
natural response is synonymous with the response in the absence of independent that the sharpest reduction in capacitive voltage, and thus stored
At any time t, the energy stored in the capacitor weCt) is given by ~Cv2(t). power, occurs at the initial time.
resistor stores no energy; thus the total energy stored in the circuit at time t is, by
weCt) = ~c [Voe-(t/RC)]2 = ~CVie-2t/RC
Example 6.1 used elements with values commonly found in low-power circuits:
Comparing with the initial stored energy given by (6.1) gives resistance in kilohms and capacitance on the scale of microfarads. This is a departure
from most other examples, which have used elements on the scale of 1 Q and 1 F. While
weCt) = weCO)e- 2t / RC
convenient for calculation, this scale of R and C values is not often seen in practice.
Note that the stored energy is exponentially decaying to zero from its initial value Resistors of the scale of 1 Q lead to high current and power dissipation at common
Where does this lost energy go? We know that the power dissipated by a resistor is voltage levels of I.S to ISO V, and capacitors on the scale of 1 F would be massively
large and expensive.
v~ [Voe-(t/RC)]2
For circuits with elements in the far more common kQ-JLF ranges, it is often
PR(t) = Ii = R
convenient to use a scaled metric system to write equations. Carrying cumbersome
Integrating from the initial time 0 to time t, we have the total energy dissipated by powers of 10 through most calculations can be avoided. Suppose that we use the scaled
resistor from time 0 to t: units listed in Table 6.1. The first three are basic SI quantities, the rest are derived from
them. Since R = E / I (we are using E to avoid confusing voltages or electrical potential
WR(t) = VJ
R
t e- 2r /RC dr:
10 difference, with the unit of volts), then VirnA = kQ and the scaled resistance unit is
the kilohm. Similarly, the scaled unit of capacitance is mA-msN or JLF. In this system
or WR(t) = ~CVJ(1 -
e- 2t /RC ) = weCO) (1 - e- 2t / RC ) inductance is measured as V-ms/rnA = V-s/A, which is the basic SI unit of henries (H).
2 The scaled SI unit system V-mA-ms-kQ....JLF-H is consistent without the need to carry
Comparing the last to (6.6), we see where the missing energy has gone. powers of 10. Had we used this scaled system in Example 6.1 with R = 1 kQ, C =
1 JLF, and Vo = 1 V, applying (6.S) we would have written
weCO) - weCt) = WR(t)
At each instant of time, the decrease in energy stored in the capacitor since t = 0 v(t) = Se-t/(l)(l) = Se- t V
equals the total energy dissipated by the resistor up to that time. As time increase~,
energy remains stored in the circuit; and in the limit as t gets very large, all the
stored energy Wc (0) has been dissipated as heat in the resistor and no stored
remains. As this occurs, v(t) and all other voltages and currents in the circuit go to
as well. Clearly, it is the internal energy stored at the initial time that is driving Quantity Standard SI Unit Scaled SI Unit
"natural" response.
Basic: Voltage (E) V V
Current (I) A rnA
Consider the circuit of Fig. 6.1 with C = 1 JLF, R = 1 kQ, Time (t) s ms
initial voltage of v(O) = S V across the capacitor. We wish to Derived: Resistance (ElI) Q kQ
the current i(t) and the maximum power dissipated by the Capacitance (I - tiE) F J.LF
From (6.S), the capacitive voltage is Inductance (V - t II) H H
Power (E - l ) W mW
v(t) = Voe- t/ RC = Se- IOOOt V
210 Chapter 6 First-Order Circuits Section 6.1 Simple RC and RL Circuits without Sources 211
and the time constant is r = 1, with units of milliseconds. Continuing, we find that This is called the characteristic equation for this differential equation, and its solution s
specifies the exponent in the solution of the differential equation, in this case
v(t) Se- t t
i (t) = R = -1- = Se- rnA R
s= - -
L
PR(t) = v 2(t) = (Se- t )2 = 2Se-2t mW
R 1 We conclude that a valid natural solution must be of the somewhat more specific form
Comparing Example 6.1, the scaled system results in more compact equations. The reader i(t) = Ke-(R/L)t
is encouraged to use this system on problems that fit these scaled units. Many such will
be found in the exercises and in the end-of-chapter problems in the remainder of this Any i (t) of this form satisfies the differential equation (6.7). The only other test a so-
book. lution must pass is that it satisfy the given initial condition i (0) = 10. Applying this
We next turn to the simple RL circuit shown in Fig. 6.3. As with the previous requirement,
RC circuit, there are no external sources and the response will be driven by initial stored
energy, in this case proportional to the square of the initial current i (0) = 10 through the
\ i (0) = 10 = K eO = K
inductor. The energy stored at the initial time is, by (S.I1), The only trial natural solution satisfying both the initial condition and the differential
equation is the one with specified constants sand K, or
L R wdO) = !Llg
Applying KVL yields
i (t) = 10e-(R/L)t (6.10)
L di + Ri = 0
FIGURE 6.3 Source-free dt
RL circuit. di R.
or -+-1 =0 (6.7) which is the desired solution. We may verify this solution by substituting i (t) given by
dt L
(6.10) into the original differential equation we set out to solve (6.7) and confirming that
This equation is of the same form as that of (6.2) for the RC circuit. We may therefore it matches the required initial condition as well. It passes these tests.
solve it by the same method, separation of variables. Let us review our reasoning. We guessed at an exponential form for the solution
Instead, let us observe that since the present equation is of the same form as (6.2), and then found values for its parameters sand K that made our guess demonstrably
we expect its solution will be of the same form as well. We know that the form of this correct. There is no doubt the result is the desired solution, since it solves the differential
previous solution, given in (6.S), is a constant times an exponential function of time. equation and has the right initial value, and we ask no more of a solution than these two
Guided by this experience, let us introduce a trial natural solution, requirements.
i(t) = Ke st (6.8)
To demonstrate the method of characteristic equations in the case
where K and s are constants to be determined. If the trial natural solution truly solves
that the initial time is not zero, consider the circuit of Fig. 6.3 with
the differential equation, we must be able to substitute it successfully into this equa-
R = 2 Q, L = 1 H, and initial current i (to) = 3 A at time to = 10 s.
tion, or
By (6.7) we have
di
- +2i = 0
dt
from which we see that
The characteristic equation is s + 2 = 0 or s = -2. Then the trial
(s + ~) (Ke st ) = 0 natural solution is
i(t) = Ke- 2t
If our trial natural solution is to work, this equation must be valid for all t 2: O.
However, the factor K est cannot be zero for t 2: 0, since that would require K = 0; and evaluating at to = lOin order to match initial conditions,
hence i (t) = 0 for all t 2: 0 in contradiction to the initial value i (0) = 10 . We
conclude that if the trial natural solution is valid, it must be that the other factor i (to) = 3 = K e- 2 (10)
is zero:
which yields K = 3e+ 20 . Thus the desired current is
R
s+- =0 (6.9)
i(t) = 3e 20 e- 2t A
L
212 Chapter 6 First-Order Circuits Section 6.1 Simple RC and RL Circuits without Sources 213
i(t) = 3e- 2 (t-lO) A v (V) i (A)
or
A plot of this current is shown in Fig. 6.4.
i (A)
e-l~ ----------
3
- i r - - - - ' - - - - - - - - - - - 7 t (s) -+____---L-_ _ _ _ _..:;,. t (s)
r=RC r=LIR
(a) (b)
-f---..l-------~ t(s)
10
FIGURE 6.5 (a) RC circuit response; (b) RL circuit response.
FIGURE 6.4 i(t) in Example 6.2 for t ~ 10.
Consider the graph shown in Fig. 6.5(a), which illustrates a natural response (the
capacitive voltage) of the simple RC circuit discussed in Section 6.1. The time required
Examining (6.5) and (6.10), we see that the capacitive voltage and inductive cur-
for a natural response to decay by a factor of 1 j e is defined as the time constant of the
rent in these single-loop, source-free circuits are both decaying exponentials. Since the
circuit, which we shall denote by r. For the given RC circuit, the natural response is
derivative of an exponential is also an exponential, the current through the capacitor
i = C(dvjdt) and the voltage across the inductor v = L(dijdt) in these circuits are also vet) = Voe- t / RC (6.11)
decaying exponentials with the same exponents. The significance of the constants 1j R C By definition of the time constant r, we require that r satisfy
and RjL in these exponentials will be taken up in Section 6.2.
v(t+r) 1
vet) e
or vet + r) = e- 1v(t)
EXERCISES
Substituting for the voltage v, this equation becomes
6.1.1. In Fig. 6.1, let to = 0, Vo = 10 V, R = 1 kQ, C = 1 p,F. Find v Voe-(t+T)/RC = e- 1Voe- t / RC
and i at t = 1 ms, t = 2 ms and t = 5 ms. or, after canceling common factors, we have
Answer 3.68 V, 3.68 rnA; 1.35 V, 1.35 rnA; 0.067 V, 0.067 rnA
+
6.1.2. If v(O) = +10 V, at what time will vet) equal +1 V?
Answer 6.9 s r =RC (6.12)
12 Q 4Q 1F v
6.1.3. If wet) is the energy stored in the capacitor in Exercise 6.1.2, find
a formula for wet), t > O. At what time does half the original stored energy
remain? The units of rare Q - F = (VjA)(CjV) = CjA = s. In terms of the time constant r,
Answer 50e- 2t / 3 J; 1.04 s the natural response is
EXERCISE 6.1.2 6.1.4. In Fig. 6.3, R = 1 kQ. If the voltage across the inductor, defined vet) = Voe- t / T (6.13)
to satisfy the passive sign convention together with i, is vL(t) = lOe- 200t
for t > 0 s, find L and the initial current i (0). The time constant for the RL circuit is computed similarly from Fig. 6.5(b), with
Answer 5 H; 10 rnA i (t) = Ioe-(R/L)t
r = ReqC = (10)(1) = 10 s
Therefore, by (6.13) the voltage is
216 Chapter 6 First-Order Circuits S(,< li()n h.:! Tinw COIlSLlllls 217
lQ a a
a
+
-<.> lOQ
+
v:::J=::: IF
4Q Vab I4F 20Q Req
12 F
b b
b b
(a) (b)
(a) (b)
fiGURE 6.8 (a) RC circuit; (b) equivalent circuit.
fiGURE 6.7 (a) Circuit for Example 6.3; (b) equivalent for t:::: O.
Since Figs. 6.8(a) and (b) are equivalent circuits, the voltage Vab(t)
EXERCISES must be the same. This voltage has time constant 12 s in (b), so it
must also in (a). Indeed, all currents and voltages in both Figs. 6.8(a)
6.2.1. In an RC circuit, determine (a) T for R = 2 kQ and C = 10 JJ-F, and (b) decay exponentially to zero with "[ = 12 s.
(b) C for R = 10 kQ and t = 20 JJ-S, and (c) R for v(t) on a 2-JJ-F capacitor
to halve every 20 ms. From Example 6.4 we see that RC circuits containing a single equivalent Req and
Answer (a) 20 ms; (b) 2 nF; (c) 14.43 kQ
Ceq can be analyzed by reduction to these equivalents, with "[ = ReqCeq . Clearly, the
6.2.2. An RC circuit consists of a 20-kQ resistor and a 0.05-JJ-F capaci- same statement applies to RL circuits with single equivalent inductance and resistance,
tor. It is desired to decrease the current in the network by a factor of 5 with-
with "[ = Leq/Req. Note that it may not always be possible to use series and parallel
out changing the capacitor voltage. Find the necessary values of R and C.
Answer 100 kQ; 0.01 JJ-F
rules to find these equivalents, however. The next example shows the general approach
in this case.
6.2.3. In a single-loop RL circuit, the current is determined to be 2 rnA
at t = 10 ms and 100 JJ-A at t = 46 ms. Find the time constant T and the
initial current at time t = o. We wish to find the time constant of the circuit of Fig. 6.9(a). The
Answer 12 ms; 4.6 rnA circuit to the right of the inductor is purely resistive, and we seek
its equivalent resistance Req. Since none of the resistors is in series
or parallel, we cannot use series-parallel equivalents. To find Req,
we will force i to flow into the circuit, as shown in Fig. 6.9(b), and
determine the resulting voltage v. Since the terminal law for the
;WITI-IOVT SOURCES circuit is v = Req i, the ratio between v and i is the desired Req.
Figure 6.9(b) is a three-mesh circuit with one current source,
The simple first-order RC and RL circuits examined in Section 6.2 were seen to have so two mesh equations are required. The mesh equations are
decaying responses governed by a single time constant. Generalizing, we must expect
the same behavior from any circuit that is equivalent, whether it is simple (solvable from
a single equation) or not.
R = (20)(5) = 4 Q
eq 20 + 5
(a) (b)
The time constant "[ in Fig. 6.8(b) is
fiGURE 6.9 (a) RL circuit; (b) circuit to determine Req.
"[ = ReqCeq = (4)(3) = 12 s
(6.16)
In summary, a sourceless circuit containing a single storage element, or the equiva-
By KVL around the outer loop, lent of a single storage element, will have all responses decay to zero with time constant
T = RC in the RC case or T = L/R in the RL case. Note, however, that circuits with
two or more storage elements may well not have a single equivalent storage element and
= -h(7i + 12i) thus will not be broken through the solution of a single first-order differential equation.
19
These are not first-order circuits, and their responses cannot be characterized by a single
= 16 1 time constant. Examples include any circuit with both capacitors and inductors and the
RC circuit of Problem 6.16. Such circuits are taken up in Chapter 7.
Thus Req = ~ Q and
L 0.038
T = - = - - = 0.032 s = 32 ms
Req 19/16 EXERCISES
Op amp circuits will also exhibit first-order behavior when they contain one storage 6.3.1. The circuit is in steady state at t = 0- and the switch is moved
element or when all storage elements can be combined into a single equivalent. They from position 1 to position 2 at t = O. Find the response vet) for t > O.
can be analyzed efficiently using nodal analysis and the virtual short- and open-circuit What is the time constant r?
principles as discussed in Section 4.8. Answer 16e-4t V; s !
4Q
Suppose that at t = 0, the output in Fig. 6.10 is V2 (0) = Vo and that
vg(t) = 0 for t > O. We wish to find V2(t) for t > O. By the virtual
short principle, node 1 is at ground potential. Summing currents into
+
this node gives 1 6Q v 8Q
i6 F
Vg - 0 + V2 - 0 + C d (V2 - 0) = 0
RA RF dt
or, with Vg = 0 for t > 0,
EXERCISE 6.3.1
dV2 1
-+--V2=0
dt RFC 6.3.2. Find i for t > 0 if the circuit is in steady state at t = 0-.
Answer le-
4
2t A
c t=o 12Q
it
4kQ
~kQ
3 3kQ
12 Q
2Q
6Q
~--=--18 v
+ 1
+ i2F
lkQ
EXERCISE 6.3.2
220 Chapter 6 First-Order Circuits Section 6.3 General First-Order Circuits without Sources 221
1=0
+
v - 6.3.4. (a) Find the equivalent resistance of the circuit below a-b. (b) Use
the result to find vet) for t > -1 if v( -1) = 20 V.
lOmF Answer (a) 10 Q; (b) 20e- 1O (t+l)
a b
R
5Q 3Q
i
----';>-
2i Examining (6.17) we see that the left-hand side is a linear combination of v and
lQ its derivative, while the right-hand side is just a constant. Thus vet) must be some time
function with the property that a linear combination of v and its derivative is constant
for all t > O. What kinds of functions have this property? Certainly, v(t) itself equal
to a constant is one such class of functions, for the derivative of a constant is zero;
so every linear combination of a constant and its derivative is, as required, also just a
EXERCISE 6.3.4
constant.
Guided by this observation, let us use a trial forced solution of the form
Vj(t) = A
where A is a constant to be determined. As we did with the trial natural solution employed
earlier, we will substitute it into the differential equation that it must satisfy to test its
validity. Replacing v by Vj(t) = A in (6.17) yields
In the preceding sections we have considered source-free circuits whose responses have
d A 10
been the result of initial energies stored in capacitors and inductors. All independent -A+-=-
current or voltage sources were removed or switched out of the circuits prior to finding the dt RC C
natural responses. It was shown that these responses, when arising in circuits containing or A = loR
a single equivalent capacitor or inductor, decay to zero exponentially over time.
Thus we have found a (constant) time function Vj(t) that satisfies the differential equation:
In this section we examine circuits that in addition to having initial stored energies
are driven by constant (dc) independent current or voltage sources. For these circuits we Vj(t) = loR
shall obtain solutions that are the result of both initial stored energy and that continually
supplied by the sources. We shall find that the responses in this case consist of two parts, Any such time function satisfying the forced differential equation is called a forced
one of which is, like the sources themselves, a constant. solution. We have found a forced solution to (6.17).
Let us begin by considering the circuit of Fig. 6.11. The network consists of the There is a second requirement that any proposed solution to our original problem
parallel connection of a constant current source and a resistor that is switched at time must satisfy, however. It is not enough to satisfy the differential equation, as Vj(t) has
t = 0 across a capacitor having a voltage v(O-) = Vo V. For t > 0, the switch is closed been shown to do. A valid solution must also satisfy the initial condition v(O-) =
and a nodal equation at the upper node is given by v(O+) = Vo V (we have used continuity of capacitive voltages to project Vo across
the switch time t = 0). Evaluating our forced solution Vj(t) at t = 0+, we have
dv v Vj(O+) = loR. So the forced solution agrees with the given initial condition if and only
C-+-=Io
dt R if Vo = loR. Examining Fig. 6.11, Vo and 10 are clearly the result of quite separate
dv 1 10 causes; Vo is an initial condition established before t = 0, and 10 is an independent
or -+-v=- (6.17)
dt RC C source function. Thus, while Vo = loR may be satisfied for particular choices of initial
This first-order differential equation is identical to that for the source-free RC case (6.2), conditions and source functions, there is no reason in general for this equation to be
except for the addition of a forcing term on the right-hand side. A forcing term in a true.
differential equation is a term f(t) independent of the unknown time functions. Thus, if our forced solution Vj (t) happens to satisfy the given initial condition, we
As in the unforced case, this first-order differential equation may be solved either have found the solution; it is vet) = Vj(t). In most cases, Vj(t) will not satisfy the initial
by separation of variables or by a trial solution approach. We choose the latter, since this conditions. We are apparently stuck, for there are no additional free parameters in our
method extends to higher-order circuits where separation of variables cannot be used. forced solution Vj(t) to use to satisfy the initial condition.
222 Chapter 6 First-Order Circuits Section 6.4 Circuits with DC Sources 223
Let Vn (t) denote any solution to the unforced version of our differential equation; Summarizing this method, the following steps solve the forced differential equation
that is, Vn (t) satisfies with constant forcing term and given initial conditions:
dV n 1
- + - v =0 (6.18) l. Using a trial forced solution equal to the unknown constant A, substitute into the
dt RC n
forced differential equation to find the forced solution.
The unforced version is formed by setting the forcing term to zero. From previous work 2. Using a trial natural solution equal to K est, substitute into the unforced differential
we know that the trial natural solution for Vn is of the form equation, solving the characteristic equation for s.
3. The total solution is the sum of natural and forced solutions. Evaluate the sum at
the initial time and set equal to the required initial value to find K.
with characteristic equation
We wish to find i(t) for t > 0 in the circuit of Fig. 6.12. Before
1
s+-=O t = 0, the circuit is as shown in Fig. 6. 13(a). Assuming that this
RC circuit has had sufficient time to go to dc steady state prior to t = 0,
or s = -lIRC, so the inductor acts like a short circuit and
Vn = K e-(l/RC)t (6.19) 12 16
iL = i + i1 = - + - = 10 A
2 4
We have thus far discovered a vJ that solves the forced differential equation (6.17), and a
16 V at t = 0-. At t = 0 the switch opens, and for t > 0 the circuit is
Vn that, for any value of its free parameter K, satisfies the associated unforced differential 12V
redrawn as in Fig. 6. 13(b). In this circuit, by KVL,
equation.
Consider finally a sum afforced and natural solutions, v = Vn + vJ' Substituting
this sum for v on the left side of (6.17),
-di + 21. =
12 (6.21)
dt
fiGURE 6.12 Circuit for Example 6.7. and i (0+) = i (0-) = 10 A by continuity of inductive currents. The
ddt (vn + vJ) + _l_(v n + vJ) = (dV n + _l_ vn ) + (dVJ + _l_VJ) trial forced solution is i J = A, and inserting this into the forced
RC dt RC dt RC
differential equation (6.21) gives
O 10
= +C 0+2A = 12
where the first term in parentheses on the right evaluates to zero since Vn satisfies the or A = 6. The unforced version of this equation is
unforced differential equation, and the second evaluates to laiC since vJ satisfies the din .
forced version. Thus we see that v = Vn + vJ satisfies the forced differential equation - +21 n =0
dt
regardless of the value of the free parameter K in the natural solution Vn. Returning to the
with characteristic equation
requirement that not only the forced differential equation, but also the initial condition,
be satisfied, s+2=0
vet) = Vn(t) + vJ(t) = Ke- t/ RC + RIo (6.20) or s = -2. Putting these results together yields
Evaluating at t = 0, the initial condition is matched if i(t) = in(t) + iJ(t) = Ke- 2t +6
V(O+) = Va = K(l) + RIo
which requires that
K = Va - RIo
12 V 12 V IH
Inserting this into (6.20), we have found our desired solution to the forced differential
equation (6.17) with given initial condition:
vet) = (Va - Rlo)e-(l/RC)t + RIo
(a) (b)
This solution may be tested by substitution into (6.17) and noting that evaluation at t = 0
does yield the specified initial value Va. fiGURE 6.13 (a) Circuit before t = 0; (b) circuit after t = O.
224 Chapter 6 First-Order Circuits Section 6.4 Circuits with DC Sources 225
and matching initial conditions, we have The presence of the op amp has not altered the fact that a single
storage element will yield a single first-order differential equation
i(O+) = 10 = K(1) + 6
(6.22). As recommended in Chapter 4, we used nodal analysis with
We conclude that K = 4 and the total solution is op amp circuits and wrote our node equation at an op amp input
node rather than the output node.
i(t) = 4e- Zt + 6 A
This example, shown in Fig. 6.14, demonstrates application of these Our final example, with circuit shown in Fig. 6.15, illustrates a gen-
ideas to op amp circuits. The circuit is in dc steady state just before eral first-order circuit containing several independent and dependent
the switch acts at t = 0, and we wish to find vz(t) for all times t. sources. In this circuit we seek the current i (t) through the inductor
With the op amp input shorted to ground for t < 0, v] and Vz are zero for t > 0, where i (0-) = 2 A.
for t < O. For t > 0, the voltage at node 1 is 10 V, since the voltage Such problems can be approached several different ways. A
50 JLF at the noninverting input is 10 V by the virtual open principle, and basic approach would be to perform mesh or nodal analysis on the
v] equals this value by the virtual short principle. The node equation circuit. This would lead to coupled first-order differential equa-
at the inverting input node is then tions, which could then be reduced to a single first-order differential
equation by substitution. The reduction is not always straightfor-
10kQ
20kQ
(25: 103) (10 - 0) + (50 x 10-6 ) :t (10 - V2) ward, and here we recommend other routes to securing this single
first-order differential equation. In this example we employ the strat-
egy of simplification by Thevenin-Norton transformation. Another
+
+ (20: 103) (10 - V2) =0 method, based on superposition, will be developed in Section 6.5.
lOV We will find the Thevenin transform of the entire circuit less
25kQ
dV2
its storage element. Then, replacing all but the storage element by
or - +V2 = 18 (6.22) the two-element equivalent, we will be left with a simple one-loop
dt
circuit to analyze. To do so, we will find the open-circuit voltage
The trial forced solution is vZf = A, and plugging it into (6.22), VaG and short-circuit current iSG defined in Fig. 6.16 at the inductor
A = 18. The characteristic equation of the unforced version of terminals abo VaG is found from Fig. 6.16(a). By KVL,
fiGURE 6.14 Circuit for Example 6.8. (6.22) is s + 1 = 0, or
Ii, + 2i, + 2U, + 10) = 80
Then the total solution is or i, = 12. Then around the left loop, again by KVL,
At t = 0+, there is no current flow through the 1O-kQ resistor, so so VaG = 32 V. To find iSG in Fig. 6.16(b), we write the two required
the voltage at the noninverting input is 10 V. Since no voltage drop mesh analysis equations,
occurs across the op amp input terminals (the virtual short principle),
v,(0+) = 10 V also. Continuity of the capacitive voltage requires Ii] = -3v + 80
that 2i2 + 2Uz + 10) = 3v
V2(0+) - v,(O+) = V2(0-) - v](O-) =0- 0 =0
lQ 2Q
Thus V2 (0+) = 10 V also, and evaluating the constant K, we obtain
vz(O+) = K(I) + 18 = 10 V
or K = -8. Thus 80V 10A
226 Chapter 6 First-Order Circuits Section 6.4 Circuits with DC Sources 227
IQ 2Q steady state discussed in Section 5.7. In Example 6.9, for the response variable i(t) given
in (6.25), the transient response is _8e- 4t A and the dc steady-state response is 10 A.
EXERCISES
fiGURE 6.16 (a) Circuit for Voc; (b) circuit for isc .
In this equation, x(t) is the circuit variable (current or voltage) and f(t) the forcing
fiGURE 6.17 Circuit equivalent of Fig. 6.15. function. If we kill all independent sources in the circuit, we will have the unforced
equation, that is, f(t) replaced by 0, but otherwise the equation is unchanged. Let Xic(t)
The solutions that we have encountered so far in this chapter are often referred to solve the problem with independent sources killed; that is,
in other more descriptive terms. Two such terms that are widely used are the transient
dXic 1
response and the steady-state response. The transient response is the transitory portion of --
dt
+ -Xic
r
=0,
the total response, which approaches zero as time increases. The steady-state response, on
the other hand, is that part of the total response that remains after the transient response has Here Xic(t) is the response due to the initial conditions. Now suppose we return the
become zero. In the case of dc sources, the steady-state response is constant and is the dc sources into the circuit, but instead, kill (zero out) the initial condition. The response in
228 Chapter 6 First-Order Circuits Section 6.5 Superposition in First-Order Circuits 229
this case we label xs(t), the response due to the independent sources, satisfying Denote the component due to the 60-V source as VIs. Killing .
the other source, the node equation at the top center node is
dx s 1
-d
t
+ -Xs
r
= f(t), 1 1 2dvls
'3 (Vis - 60) + '1 (Vis - 0) + ~ =0
It is easy to see that the superposition of these two responses x(t) = Xic(t) +xs(t) satisfies
the original forced differential equation, since
dVls 2
or --
dt
+ -VI
3 s
= 10 (6.27)
This is the solution for t > 0, with v(t) = 60 V the solution for
FIGURE 6.19 (a) Equivalent for t < 0; (b) for t > O. t < O.
230 Chapter 6 First-Order Circuits Section 6.5 Superposition in First-Order Circuits 231
The mesh equations for Fig. 6.22(a) are
sov lOA
Solving the first of these equations for i I and substituting into the
second yields
FIGURE 6.20 Circuit for Example 6.11.
d
dt i2 + 4i2 = 20, i2(0+) = 0
We shall redo Example 6.9 using superposition. The circuit diagram
Example 6.11 shown in Fig. 6.15 is repeated for convenience as Fig. 6.20. First with zero initial inductive current, since this is a forced response
\
we will compute iie> the component of i due to the initial conditions with initial conditions killed. The solution to the above is
with all independent sources killed. In this case the resistors may
be replaced by an equivalent of value Req = ~, and we have a
single-loop circuit as shown in Fig. 6.21. By KVL,
The third and final component, i 2f (t), is computed with the voltage
source killed and the current source returned to the circuit, as shown
(6.31) in Fig. 6.22(b) (the leftmost two parallel branches have been drawn
+ in reverse order for convenience). The mesh equations are
Using Ohm's law, v = ~iiC' and dividing by ~, the above is
5"4 (di
dtl ) + (il - i2) - 3(i2 - il) = 0
di ic .
FIGURE 6.21 Circuit for component he -+4z ic =0
dt 2i2 + 2(i2 + 10) + (i2 - il) = 0
The given initial value is i (0-) = i (0+) = 2 A, leading to the
component Solving the second mesh equation for i2 and substituting into the
first gives us
(6.32)
dil
- +4i l = -20
Next we need the forced components ifl due to the independent dt '
voltage source and i f2 due to the current source. To determine
once again using the zero initial inductor current to derive initial
i fl we kill the initial conditions and all other independent sources,
conditions. This solution is
in this case the current source, as shown in Fig. 6.22(a). Here
the branch containing the inductor has been drawn on the right for
convenience.
From Fig. 6.22(b) we see that
1Q 2Q
If
+
fiGURE 6.22 (a) Circuit for component ifl; (b) circuit for if2' which agrees with our result from Example 6.9.
232 Chapter 6 First-Order Circuits Section 6.5 Superposition in First-Order Circuits 233
.. u(t)
EXERCISES 1[1 _ _ _ _ _ __
6.5.1. Use superposition to find i for t > 10 s. Assume that the circuit'
is in dc steady state at time t = 10- s.
Answer 5 + 10[1 - e- 500 (t-1O)] A
FIGURE 6.23
r
o
Graph of the unit step function u(t).
function is the function that is equal to zero for all negative values of its argument and
lOY 500 Q 10mA
that is equal to one for all positive values of its argument. If we denote the unit step
\ function by the symbol u(t), a mathematical description is
EXERCISE 6.5.1
u(t) = { ~: t < 0
t > 0
(6.35)
6.5.2. Repeat Exercise 6.5.1 using the method of Thevenin transform.
6.5.3. Use superposition to find V3 in terms of VI and V2.
dVI RF
Answer V3 = -RFC- - -V2
From its graph shown in Fig. 6.23 we see that at t = 0, u(t) changes abruptly from
dt RA
o to 1. Nowhere else does its value change except at this "singular" moment in its
behavior. Some authors define u(O) to be 1; others prefer u(O) = !;
but we are leaving
c u(t) undefined at t = O.
A voltage that steps from 0 to V volts at t = 0 may be represented by the product
Vu(t). Clearly, this voltage is 0 for t < 0 and V volts for t > O. A voltage step source
+ + of V volts is shown in Fig. 6.24(a). A circuit that is equivalent to this source is shown
in Fig. 6.24(b). A short circuit exists for t < 0, and the voltage is zero. For t > 0, a
voltage V appears at the terminals of the equivalent circuit.
Equivalent circuits for a current step source of I amperes are shown in Fig. 6.25.
An open circuit exists for t < 0, and the current is zero. For t > 0, the switching
action causes a terminal current of I amperes to flow at the terminals of the equivalent
circuit.
EXERCISE 6.5.3
Unit step sources of the type shown in Figs. 6.24 and 6.25 will be frequently
used in circuits involving sources that instantaneously switch in or out. Note in these
In the preceding sections we analyzed circuits in which independent sources have been +
suddenly inserted into the networks. At the instant these sources are applied, the voltages
Vu(t)
or currents at the points of application change abruptly. Forcing functions whose values
change in this manner are called singularity functions, since they have one time instant
at which they exhibit singular or unusual behavior.
There are many singularity functions that are useful in circuit analysis (and we
defer their full description to Chapter 13). One of the most important is the unit step (a) (b)
function, so named by the English engineer Oliver Heaviside (1850-1925). The unit step
FIGURE 6.24 (a) V volt voltage step source; (b) equivalent circuit.
21-------1
Consider the circuit of Fig. 6.28(a). For all t < 0 the voltage source
has a value of 10 V, as shown in Fig. 6.28(b), leading to a dc steady-
o to
state current of i (0-) = 2 A. After t = 0 the source may be replaced
by a short circuit [since u( -t) = 0 for t > 0], leading to the circuit
FIGURE 6.26 Time-shifted unit step. of Fig. 6.28(c). From our previous analysis of the simple sourceless
RL circuit, we know that (6.10)
Consider the voltage vet) formed as a linear combination of a unit
i (t) = Ioe- t / T A
step with a second, time-shifted unit step:
With T = L/R = 4 and 10 determined by the initial condition,
vet) = K1U(t) + K2U(t - to)
i(O+) = i(O-) = 2 = 10 (1)
Suppose first that the second unit step is right-shifted (delayed); in
other words, to > O. We will determine the value of vet) by use of whence the total solution
(6.35) and (6.36). For t < 0, neither unit step has "switched on" yet i(t) = 2e- t / 4 A
236 Chapter 6 First-Order Circuits Section 6.6 Unit Step Function 237
5Q Answer u(3 - t) + u(7 - t) + 2u(t - 10)
i(A)
lOu(-t) V
+
(a)
5u(4-t)A 6Q
5Q
EXERCISE 6.6.2
5Q
6.6.3. Find vet) for t > 4. Assume that the circuit is in de steady state
just before the unit step jump time.
EXERCISE 6.6.3 Answer 30e-(t-4)/3 V
lOY
(b) (e)
fiGURE 6.28 (a) Circuit with time-reversed step source; (b) equivalent
for t < 0; (c) equivalent for t > O.
EXERCISES
- The step response is defined as the response of a circuit having only one independent
source, which is a unit step function. The response and step source can each be either a
current or a voltage. Since the unit step source presents zero source function until time
t = 0-, the dc steady-state response by the circuit at t = 0- must be zero. There can
6.6.1. Express VI (t) as a sum of scaled and time-shifted unit steps. How be no initial energy stored in the storage elements at the singular time when the source
is V2(t) related to VI (t)? Use these two to express V2(t) as a sum of scaled switches on because there has been no source for t < 0 to energize the storage elements.
and time-shifted unit steps. Thus the step response is the response to a unit step input with no initial energy stored
Answer u(t) +u(t -1) +u(t - 2) - u(t - 3) - u(t -4) - u(t - 5); in the circuit.
V2(t) = 2VI (t-l); 2[u(t-1)+u(t-2)+u(t-3)-u(t-4)-u(t-5)-u(t-6)]
As an example, let us find the step response vet) in the simple RC
circuit of Fig. 6.29(a) having an input of vg(t) = u(t) V. The equiv-
6 alent circuit is shown in Fig. 6.29(b). For t < 0, vg(t) = 0 and
so the dc steady state at t = 0- is v(O-) = 0 V. By continu-
ity of capacitive voltages, v(O+) = 0 V as well. Applying KCL
4 for t > 0 yields
3 dv 1 1
(6.38)
2 2 dt + RC v = RC
The characteristic equation is
-..J,.-...l.---L--'----'--'--3 t (s) t (s) 1
2 3 5 2 3 4 5 6 s+-=O
RC
(a) (b)
leading to the natural response
EXERCISE 6.6.1
vn(t) = Ke- t / RC (6.39)
6.6.2. Express i(t) as a sum of two time-reversed and one normal unit The trial forced solution is vI = A, which when substituted into
step. (6.38) yields A = 1. Combining this with the natural solution (6.39),
238 Chapter 6 First-Order Circuits Section 6.7 Step and Pulse Responses 239
t=O R
R As a second example, let us find V2(t) in the circuit of Fig. 6.30,
k-~l
consisting of a resistor, a capacitor, and an op amp, when the input
is a unit step voltage source. The nodal equation at the inverting
+ terminal of the op amp is, upon division by C,
Vg = u(t) V C v IV
dV2 1
dt = -RC VI
C
Integrating both sides of this equation between the limits of 0+ and
(a) (b)
t, we have
V (V)
1 ------------ While our current interest is centered on the specific case VI (t) =
Vj = u(t) V
Vz u(t), note that by (6.42) the general effect of this circuit is to create
an output voltage V2(t) that is proportional to the integral of the input
,, VI (t) plus a constant. For this reason the circuit is called an inte-
,
grator. Some authors prefer the term inverting integrator because
Step ~esponse 1- e- tiRC
FIGURE 6.30 Circuit for Example 6.15. of the negative sign in the voltage transfer equation (6.42).
1 L - - . . . . : L - . . : . - - - - - - - - - - - - - - ; 7 t(s)
r=RC In the case of a unit step input VI (t) = u(t), all currents and
(c) voltages in the circuit are zero at 0- and, by continuity, vc(O+) =
o V as well. The virtual short principle tells us that the inverting
FIGURE 6.29 (a) RC circuit with a voltage step input; (b) equivalent terminal is at ground potential; thus Vc = V2 and the initial condition
circuit; (c) step response of RC circuit. term in (6.42) is V2(0+) = 0 V. For t > 0, the integrand VI (t) =
u(t) = 1; thus
the total solution is
vet) = K e- t / RC +1 V2(t) = __1_
RC
r
10+
dr + 0 = ( __ 1_)
RC
t
Finally, matching the initial condition v(O+) = 0 gives K = -1.
The result is The step response of our integrator circuit is 0 for t < 0 and the
above for t > 0, or, using multiplication by u(t) to make this equa-
vet) = { ~'_ e- t / RC ,
t < 0
t>O
(6.40) tion concise, for all t
1
where we have used the fact that before the singular time all cur- V2(t) = --tu(t)
rents and voltages were zero. This can be written somewhat more RC
concisely as The function tu(t) is called the unit ramp singularity function and is
vet) = (1 - e-t/RC)u(t) (6.41) shown in Fig. 6.31(a). Our step response V2(t) is a negative-going
ramp function, scaled by -lIRC and is shown in Fig. 6.31(b).
avoiding the "laundry list" form of (6.40). The step response is
shown in Fig. 6.29(c). tu(t) 1
Once again it is the time constant that tells us the rate at which the transient response ~"'.-~ :> t
decays to zero. Each period of duration of one time constant, the transient reduces to e- 1
times its value at the beginning of that period. Circuits with short time constants reach
their steady-state value more quickly.
Note by (6.41) that multiplication of any other time function by a unit step forces the
(a) (b)
product to be zero prior to the singular time and to be equal to the other time function
afterward. We will make frequent use of this convenient property of the unit step to FIGURE 6.31 (a) Unit ramp function; (b) step response of the circuit
simplify other equations arising in subsequent chapters. of Fig. 6.30.
240 Chapter 6 First-Order Circuits Section 6.7 Step and Pulse Responses 241
i(A)
Closely related to the step response is the pulse response. We saw in Section 6.6
how a linear combination of two unit steps could be used to produce a rectangular pulse, 61------- .. -..
6 ---------------- -----------:.,-
' . .
for instance Fig. 6.27(a). We now consider the response of a first-order circuit to a pulse
input. As before, we assume no other independent sources are present in the circuit.
.. .' .. -
'
3.7ge-(t-l)
Prior to the leftmost jump, or leading edge of the pulse, there are no currents or
voltages for the same reason mentioned for the step response: the dc steady-state response
to a zero-value source must be zero. Nothing has turned on yet. During the pulse, the
response must be identical to a step response, since the pulse and equal-amplitude step --:-o--'------..L
1----..:::;;.--- t (8) -~o'-.lL-------'-l----.........:::,.. t (8)
inputs will only differ after the second jump time, which has not yet occurred. Finally, (a) (b)
after this second jump time, or trailing edge of the pulse, the source is back to zero. We
need only compute the response of a sourceless circuit with the proper initial condition FIGURE 6.33 (a) Pulse current source; (b) pulse response.
(applied at the trailing edge of the pulse). A = 6. Matching the zero initial condition,
i(O+) = K(1) +6
To illustrate the pulse response of a first-order circuit, consider
Fig. 6.32(a), where the source is the current pulse or K = -6. We have thus far that i (t) = 0 for t < 0, and
i(t) = 6(1 - e- t ) A, 0< t < 1 (6.45)
ig(t) = 6(u(t) - u(t - 1)) A
This portion of the response is seen in Fig. 6.33(b). The dashed
This source is zero before t = 0, supplies 6 A in the interval 0 < line shows that, if the source did not switch back off at t = 1, the
t < 1, and then shuts down again. It is sketched in Fig. 6.33(a). response would approach the value of 6, characteristic of the step
Replacing the source and 5-Q resistor by its Thevenin equivalent, response for (6.44). This is only reasonable, since if the source did
we have the single-loop circuit of Fig. 6.32(b). KVL around the not switch back off it would be a step source.
loop and division by 5 yield Returning to (6.43), for t > 1 we have ig(t) = 0 or
di di
- +i = ig(t) (6.43) -+i=O t>1 (6.46)
dt dt '
Before t = 0 all currents and voltages are zero, since there is no Thus in the interval t > 1 we need only solve an unforced first-order
source of energy to stimulate nonzero responses. Thus i (0-) = circuit. The time constant is r = 1, and the solution is
i(O+) = 0 A. For 0 < t < 1, ig(t) = 6, and for this interval of time i(t) = Ke- t , t> 1 (6.47)
we have
Since (6.45) is valid for t > 1, its initial condition must be at t = 1+
di (6.44)
dt + i = 6, i (0+) =0 just after the trailing edge of the pulse. We have already determined
the current just before this time, since by (6.45)
The characteristic equation is s + 1 = 0, and the natural response is
i(I-) = 6(1- e- 1) = 3.79 A
K e- t The trial forced solution is A, and plugging in we see that
By continuity of inductive currents, i(I+) = 3.79 as well. To sim-
plify the initial condition matching, we will rewrite the natural so-
lution (6.47) as
5H
(6.48)
These two forms are equivalent, with the new unknown K 1 replacing
5Q
K(K = K1e 1). Then, by (6.48) evaluated at time 1+, Kl = 3.79
and
i(t) = 3.7ge-(t-l) A, t> 1
(a) (b)
which completes the full time history of i(t). A graph of this pulse
FIGURE 6.32 (a) Circuit for Example 6.16; (b) equivalent circuit. response is shown in Fig. 6.33(b).
6.7.3. Use the answer from Exercise 6.7.2 to find the response i to Vg ==
21Ou(t) V. This example illustrates the use of SPICE both to determine initial
Answer By linearity, 10(1 - e-7t )u(t) A conditions and the subsequent transient analysis. We seek the current
idt) in Fig. 6.34. Since the voltage source is 12u(-t) V, its value
for t < 0 is 12 V. The initial condition i L (0+) = i dO-) may be
found by replacing the inductor with a short circuit and finding the
dc value for iL with the 12-V source in place.
RESPONSE
SPICE was introduced earlier and used to analyze resistive circuits with dc sources. The
program, of course, does much more than this. Here we discuss its use in obtaining
transient responses.
+
The basic format of a SPICE input file and the element statements for resistors,
dc sources, and linear controlled sources were introduced in Section 4.9. The format for
38 mH vL CD <--~\,"I\~--t
storage element statements (capacitors and inductors) is
246 Chapter 6 First-Order Circuits Section 6.8 SPICE and the Transient Response 247
20 ttF
leA)
or
50kr2
v (V)
---+---+!---+-----l-\+------.J~ t (8) CD
+ +
TR TF
Vg v2
statements
248 Chapter 6 First-Order Circuits Section 6.8 SPICE and the Transient Response 249
**** 10/10/94 09:26:24 ********* Evaluation PSpice (January 1991) ************
J(A) or V(V) J(A) or V(V)
EXAMPLE 6.18: STEP RESPONSE
V2 Vz ----------------:::~~;~----------
\,-
TRANSIENT ANALYSIS TEMPERATURE = 27.000 DEG C
******************************************************************************
VI I---.J.. - - - - - - -:- - - - - - - - - - - - - - - - - -.
,
TIME V(3) ' -_ _....L...-_L-_--'-_ _ _ _ _~ t (8)
(*)---------- -2.0000E+00 -1.5000E+00 -1.0000E+00 -5.0000E-Ol O.OOOOE+OO Tl T2 T3
'-------'-------~ t(8)
---------------------------
O.OOOE+OO -7.956E-06 * TSPEC = PWL(Tl VI (T2 V2 ... TSPEC = EXP(VI V2 (Tl TCI T2 TC2
1.000E-Ol -1.894E-Ol (a) (b)
2.000E-01 -3.594E-01
3.000E-Ol -5.140E-01
4.000E-Ol -6.548E-01 FIGURE 6.39 (a) Piecewise linear source; (b) exponential source.
5.000E-01 -7.844E-Ol
6.000E-Ol -8.996E-Ol \
7.000E-Ol -1.006E+00 J (A) or V (V)
8.000E-Ol -1.100E+00
,
9.000E-Ol -1.187E+00 VO+VA - - - - - - - - -:',
1.OOOE+OO -1.264E+00
1.100E+00 -1.335E+OO
1.200E+00 -1.398E+00
1.300E+00 -1.456E+00
1.400E+00 -1.507E+00
1.500E+00 -1.555E+00
1.600E+OO -1.597E+00
1.700E+00 -1.636E+00
1.800E+00 -1.670E+00 VOI---'"
1.900E+OO -1.702E+00
2.000E+00 -1.730E+00
2.100E+00 -1.756E+OO
2.200E+00 -1.779E+00
2.300E+00 -1.801E+00
2.400E+00 -1.820E+OO I---~-__:;~--------------?>- t(8)
2.500E+00 -1.837E+00
2.600E+00 -1.852E+00 VO - VA ---------:--
2.700E+00 -1.867E+00 , ,
2.800E+00 -1.879E+00
2.900E+00 -1.891E+00
3.000E+OO -1.901E+OO TSPEC = SIN(VO VA(F TD TH)
3.100E+OO -1.911E+00
3.200E+00 -1.919E+OO FIGURE 6.40 Damped sinusoidal source.
3.300E+00 -1.927E+OO
3.400E+OO -1.934E+OO
3.500E+OO -1.940E+OO at the maximum available screen resolution. The control statement
3.600E+OO -1.946E+OO
3.700E+OO -1.951E+OO
3.800E+OO
3.900E+OO
-1.956E+OO
-1.960E+OO
.PROBE
4.000E+OO -1.964E+OO
4.100E+OO -1.967E+OO
4.200E+OO -1.970E+OO when added to the input file results in the creation of an auxiliary data file that supports
4.300E+OO -1.973E+OO the operation of PROBE, a graphics postprocessor software program bundled with PSpice.
4.400E+OO -1.976E+OO
4.500E+OO -1.978E+OO Upon execution, PROBE permits the circuit variables to be plotted and scales for the axes
4.600E+OO -1.980E+OO
4.700E+OO -1.982E+OO * and other display characteristics to be selected and changed interactively. A PROBE plot
4.800E+OO -1.984E+OO *
4.900E+OO -1.985E+OO * of the same pulse response. PLOTted in Fig. 6.41 is shown in Fig. 6.42.*
5.000E+00 -1.987E+OO * Finally, a circuit with nonzero or multiple switch times can be analyzed with SPICE
5.100E+OO -1.988E+OO *
5.200E+OO -1.989E+00 * by shifting the switch time (or times) to zero. For instance, suppose that we have a cir-
5.300E+OO -1.990E+OO *
5.400E+OO -1.991E+OO * cuit in dc steady state at t = 0-, at which time a switch acts, followed by a second
5.500E+OO -1.992E+OO *
5.600E+OO -1.993E+OO * switch action at t = to > O. To determine the conditions at t = 0-, we first do a dc
5.700E+OO -1.993E+OO * analysis of the circuit with the switches in their t < 0 positions. The resulting values
5.800E+OO -1.994E+OO *
5.900E+OO -1.995E+OO * are used as initial conditions on the storage element statements for a second SPICE run,
6.000E+OO -1.995E+OO *
6.100E+OO -1.996E+OO * this time a transient analysis of the circuit with its switches in their 0 < t < to positions.
6.200E+OO -1.996E+OO *
FIGURE 6.38 Step response for Example 6.18. 'Users of SPICE version 3, DESIGN CENTER, and other SPICE variants or supersets may also have
access to high-resolution graphics. Consult your user's guide.
******************************************************************************
0.5V
TIME V(3)
1*)---------- -l.SOOOE+OO
O.OOOE+OO 1.680E-08
-l.OOOOE+OO -S.OOOOE-Ol O.OOOOE+OO
------------
*
S.OOOOE-Ol
!-----Dt----'---D
!~ 0 ~ 0
O.OV
~ _ _ _r;-
1.000E-Ol -1.890E-01
2.000E-01 -3.603E-01
3.000E-01 -S.113E-01
4.000E-01 -6.S70E-01
S.OOOE-Ol -7.822E-01
6.000E-01 -9.014E-01
7.000E-01 -1.004E+00 \ -O.5V
8.000E-01 -l.lOlE+OO 1
9.000E-01 -1.186E+00
1.000E+00 -1.266E+00
1.100E+00 -1.149E+00
1.200E+00 -1.041E+00
1.300E+00 -9.419E-01
1.400E+00 -8.S41E-01
1.SOOE+00 -7.710E-01
-1.0V II
1.600E+00 -6.986E-01 I
1.700E+OO -6.307E-01
1.800E+OO -S.71SE-01
1.900E+OO -S.lS9E-01
2.000E+OO -4.674E-01 -1.5V ' -_ _ _ _ _-'-_ _ _ _ _- L_ _ _ _ _- L_ _ _ _ _--'-_ _ _ _ __
2.100E+OO -4.220E-01
2.200E+OO -3.823E-01 O.Os 1.0s 2.0s 3.0s 4.0s 5.0s
2.300E+00 -3.4S2E-01
2.400E+OO -3.127E-01
Time
o V(1) V(3)
2.S00E+OO -2.823E-01
2.600E+OO -2.SS8E-01
2.700E+OO -2.309E-01 FIGURE 6.42 Pulse response for Example 6.18 (PROBE output).
2.800E+OO -2.093E-01
2.900E+OO -1.889E-01
3.000E+OO -1.712E-01
3.100E+OO -1.S4SE-01 We terminate this analysis at the time to, outputting the values of the capacitive voltages
3.200E+00 -1.400E-01
3.300E+OO -1.264E-01 and inductive currents. These values are input as initial conditions for a final run, using a
3.400E+OO -1.14SE-01
3.S00E+00 -1.034E-01 SPICE input file reflecting the circuit configuration that results by setting the switches in
3.600E+OO -9.367E-02 their t > to positions. Since SPICE always begins a transient analysis at t = 0, we will
3.700E+OO -8.4S7E-02
3.800E+OO -7.662E-02
3.900E+OO -6.917E-02
remember to add to to every time listed in the final output. An example with multiple
4.000E+OO -6.268E-02
4.100E+OO -S.6S8E-02
switch times is given in Exercise 6.8.3.
4.200E+00 -S.127E-02
4.300E+OO -4.628E-02
4.400E+OO -4.193E-02
4.S00E+OO -3.786E-02
4.600E+OO -3.430E-02
4.700E+00 -3.097E-02
EXERCISES
4.800E+OO -2.806E-02
4.900E+OO -2.S33E-02
S.OOOE+OO -2.29SE-02 6.8.1. Using SPICE, find (a) h(O-) and (b) iL (30 JLs).
S.lOOE+OO -2.072E-02
S.200E+OO -1.877E-02 Answer (a) -2.475 rnA; (b) +0.285 rnA
S.300E+00 -1.69SE-02
S.400E+00 .-1.S36E-02
S.SOOE+OO -1.386E-02
S.600E+OO -1.2S6E-02
S.700E+OO -1.134E-02
S.800E+OO -1.027E-02
S.900E+OO -9.27SE-03
6.000E+OO -8.404E-03
6.100E+OO -7.S87E-03
6.200E+OO -6.874E-03 v=6V
i=5mA
EXERCISE 6.8.1
252 Chapter 6 First-Order Circuits Section 6.8 SPICE and the Transient Response 253
6.8.2. ~epeat Exercise 6.8.1 for the case in which the rightmost switch Exercise 6.8.4
closes at time t = 20 f.LS rather than t = O.
VS 1 0 DC 5
Answer (a) -2.475 rnA; (b) +0.218 rnA
XR1 1 2 RSUBCKT
6.8.3: Plot the response v(t) for 0 < t < 10 ms if Vg is a pulse of XU 2 3 LSUBCKT
amplitude 10 V and duration 2 ms, and the initial value of v is +2 V. 4 RSUBCKT
XR2 3
XR3 4 5 RSUBCKT
2kQ 4.7kQ XL2 6 5 LSUBCKT
RL 6 0 2K
.SUBCKT RSUBCKT 1 3
Vg
R1 1 2 5K
5vrnA 1kQ
R2 2 3 10K
R3 2 0 4K
.ENDS
(a) .SUBCKT LSUBCKT 1 2
v (V) U 1 2 10M 1C=0
L2 1 0 20M 1C=0
12
.ENDS
.END
2
First-order circuits have two parameters that may be assigned. These may be specified as
- - - - - - I - -2L --------:::,.. t (ms) the resistance R and inductance L, resistance R and capacitance C, or the time constant
r and the gain of the circuit. First-order circuits may be either passive, designed using
(b) only RLC elements, or active, incorporating op amps or other active devices. What they
have in common is that they are described by first-order differential equations.
EXERCISE 6.B.3
In Design Example 6.1 we demonstrate one common use of first-order circuits: to
block or trap dc. Often, a dc bias voltage is produced by a source circuit and it is desired
to prevent this dc value from appearing across the terminals of the load. In this case a first-
6.8.4. Write a SPICE input file for this circuit using two subcircuits
order circuit can be interposed between source and load, serving as a buffer between them.
Assume that all initial currents are zero. Orru't the TRAN or DC statement.
and output control statements.
Answer Design a first-order buffer circuit B which blocks any dc source volt-
age component from appearing across the load RL [see Fig. 6.43(a)].
In addition, the half-amplitude pulse width of the step response
5 kQ 10 kQ 10 mH 5 kQ across the load should be no more than 1 ms for any load R L .
The i-v law for a capacitor is i = C dv/dt. If the voltage
v(t) is a constant (dc), then dv/dt = 0 and no dc current will
flow through the capacitor. Thus we can place any capacitor C in
2kQ series with the source voltage to block dc. If the buffer circuit B
in Fig. 6.43(b) was only the capacitor, in other words if the resistor
labeled R B in this figure were omitted, the time constant of the
resulting RC circuit would be r = RL C and could be arbitrarily long
EXERCISE 6.B.4 for large RL regardless of our choice for C. So we put a resistor RB
--4---------------'7 t(s)
+
(a)
(b) + +
O.OOOE+OO 4.163E+00
--------- --------- - - - *- -
- - --
2.000E-01 4.107E+00 *
A power company is supplying electrical power to a load region 4.000E-01 4.070E+00
6.000E-01 4.047E+00
modeled by the resistive network shown in Fig. 6.46. If an L-henry S.OOOE-Ol 4.032E+OO
inductance is introduced between the pair of nodes hand h' shown, 1.000E+00 4.022E+OO
1.200E+00 4.016E+OO
11
find a formula for the time constant of the resulting R L load circuit 1.400E+00 4.012E+OO
a 1.600E+OO 4.009E+OO
as seen from the source terminals a-a'. 1.S00E+00 4.00SE+OO
This design problem is a bit different from the previous ones, 2.000E+00 4.007E+00
2.200E+OO 4.006E+00
in that we must design many circuits at the same time, those with all 2.400E+00 4.00SE+00
v(t) possible time constants r. Also, we have rigid structural constraints 2.600E+00 4.005E+00
2.S00E+00 4.00SE+OO
on our design; the form of the circuit layout is given. There is no 3.000E+00 4.005E+00
single format for design problems, they come in many forms and 3.200E+00 4.005E+OO
3.400E+00 4.005E+OO
a' with a great variety of constraints. 3.600E+00 4.005E+OO *
Since there is only one inductor, this is a first-order circuit with 3.S00E+00 4.005E+OO *
@ 31
time constant LIReq , where Req is the equivalent resistance. Rather
4.000E+OO
4.200E+OO
4.005E+OO
4.005E+OO
*
*
4.400E+OO 4.005E+OO *
than compute the time constant using exact analytic methods, which 4.600E+00 4.005E+OO
FIGURE 6.46 Circuit for Design Example 6.3. *
would be time consuming since the resistors are neither in series nor 4.S00E+00 4.005E+00 *
5.000E+00 4.005E+00 *
parallel, we will use SPICE and numerical approximation. Suppose
that we set L = 1 H. If we determine r = rl for L = 1 H, we would
know Req = l/rl and hence the desired formula r(L) = LIReq. A JOB CONCLUDED
SPICE input file for this design strategy is shown below.
TOTAL JOB TIME .33
*
I 2 1 DC 1
The resulting plot, shown as Fig. 6.47(a), shows that the steady-state
R1 1 0 voltage response is 4.005 V. Since the initial voltage was 4.163 V,
75
R2 2 0 the size of the transient response is 4.163 - 4.005 = 0.158 V. The
8
R3 1 3 time constant r is the time at which the transient is reduced to e- I
1
R4 2 5 of its initial value, or
3
R5 3 0 2 v12Cr) = 4.005 + e- I (0.158) = 4.063
R6 5 0 0.25
R7 3 4 Figure 6.47(a) has too coarse time sampling to make out the time
3
R8 4 5 at which v(l, 2) takes on this value with acceptable accuracy. Ex-
4
L 4 0 panding the scale using the edited transient mode control statement
1 IC=O
.TRAN 0.2 5 UIC
PLOT TRAN V(1,2) .TRAN 0.01 0.5 UIC
END
results in the plot shown in Fig. 6.47(b), from which we see that
Design Example 6.3 was the first in which SPICE was used to save labor. We
~roceeded by evaluating a candidate circuit, editing the input file, and evaluating again
First-order circuits are those characterized by a single first-order differential equation.
IIIa cyclic manner until a design goal was reached. Sometimes it is the circuit element
They can be identified by the presence of a single equivalent storage element (inductor
values or locations that are changed; other times, output parameters such as the transient
or capacitor). The behavior of first-order circuits may be studied by writing and solving
sampling interval and duration in this example. In either case the tedium of repetitive,
their differential equations in the time domain, which is the focus of this chapter.
260 Chapter 6 First-Order Circuits
Summary 261
An unforced first-order differential equation can be solved by separation of variables or o 6.4. If the initial stored energy in the capacitor at time 0 6.8. In terms of r, how much time does it take a response
by the characteristic equation method. The solution is a real exponential time function
characterized by a single time constant.
;m t = 0- is 0.18 fLJ, how much energy will remain stored at I!illl in an unforced RC or RL circuit to decay by a factor of 2?
time t = 20 ms? At time t = 200 ms? By a factor of 1O? By a factor of 1000?
The total solution to a forced first-order differential equation is the sum of a forced 6.9. Sketch wet) versus t for the circuit of Problem 6.7.
solution and an unforced (natural) solution. The natural solution can be found as
~O
What is the time constant of this waveform? Why does it
above, with forcing terms removed from the differential equation. The forced solution differ from r = RC?
can be found by use of a trial forced solution.
The multiplier of the natural solution is determined by requiring that the total solution
match the initial condition.
O.04fLF 1 . 500kQ
6.10. From Fig. 6.6, the initial slope of a waveform with
time constant linearly extrapolated hits v(t) = 0 at t = To
Where does a similar construction from the point vCr) hit
the horizontal axis?
If more than one independent source is present, superposition may be used to determine
the overall forced response. For each source, all other sources are killed, and the 06.11. Design a source-free circuit with r = 1 fLS. If we
resulting forced responses superposed. I!illl have
w(O) = 1 J in this circuit, how much energy will it
FIGURE P6.4 dissipate by time t = 1 fLS? t = 5 fLS?
The unit step function u(t) is defined as zero for t < 0, one for t > O. It models the
function of a switch which turns on at t = O. The unit step response of a circuit is a 6.12. A sourceless RL circuit has 4-rnA current at time
common way to characterize its behavior. 6.5. The energy w = ~ Li 2 stored in the inductor satisfies t = 2 ms and 1 rnA at time t = 4 ms. What was the initial
an enforced first-order differential equation current at time t = O?
Inductors and capacitors with arbitrary initial conditions are supported by SPICE. The 6.13. Show that this circuit is characterized by a second-
.TRAN control statement is used to generate a transient analysis. dw
-+aw=O order differential equation, so it is not a first-order cir-
dt cuit.
Unlike resistive circuits, first-order circuits contain responses that do not simply Find a.
mimic the waveshapes of the sources exciting them. These are the simplest circuits that
can create something new, something not already inherent in the circuit inputs, namely
real exponential time functions. Circuits with more than one storage element can produce
a much broader range of new responses and are the subject of the next chapter.
L R
PROBLEMS
6.1. Find vet) and i (t) for t > 0 if v(O) = 2 V.
FIGURE P6.13
FIGURE P6.5
+
v lmA
6.6. Design a sourceless RL circuit with a time constant
6.14. Repeat Problem 6.13 for the case in which CI is
of 10 ms using only 2-mH inductors and l-kQ resistors.
replaced by a second inductor L2.
Use the minimum number of circuit elements possible.
+ 6.15. Find the time constant r.
6.7. What is the time constant r? Express vet) and i (t)
v*k F :s.lOQ FIGURE P6.2 in terms of r if v(O) = +100 V.
6.3. Determine VI(t) for t > 0 if VI(O-) = -12 V. IQ
i
--?>-
FIGURE P6.1
L
SIMQ
15 Q
lOkQ ,AA
vv
lOQ
,AA
+ + vv
+ IH 2Q v +
20kQ
vz :;:~fsp 24Q> v
FIGURE P6.25
FIGURE P6.19 i
~
~R
vv
\
9Q 72Q
1t (iCr) + 2) dr + 3i(t) = 6
2kQ
+ for t > 0 by converting it to a differential equation.
10 kQ lOkQ 6.28. Find v for t > O. Assume dc steady state at t = O.
40V 9Q v
12V
+
FiGURE P6.20
t=O
FIGURE P6.23
IkQ 2kQ
6.21. Find all currents and voltages for t > 0 in this cir-
FIGURE P6.17 6.24. Find v for t > 0 if the circuit is in steady state at
cuit. Assume dc steady state at t = 0-.
t =0-.
FIGURE P6.28
6.18. Find i (t) for t > 10 s if v(10-) = 2 V. Repeat for
v(1O-) = 0 V. 15 Q
6.29. Find v(t) for t > O. Assume dc steady state at time
t = O.
IV +
4Q 2Q 2Q a b 4Q
6Q v 72Q
=
.----'\I\I\.----<J~+>--------'t
O\I\I\~
2i IQ 3Q
24Q 12V lp
8
1\ 16V
FIGURE P6.24
2Q
1
8" F v
3kQ
lOY 1mB
+
FIGURE 1>6.42
+
3V 6.43. Repeat Problem 6.42 for the pulse response if
FIGURE P6.34
fiGURE 1"6.30 vg(t) = u(t) - u(t - 1).
3V
6.44. Find v for t > 0 if Vg = 3u(t) V.
6.31. Find v for t > 10. Assume dc steady state at t = 6.35. Find v for t > 0 using superposition. Assume the
circuit is in dc steady state at t = 0-. 24Q
10-.
fiGURE P6.38
6Q
t= 10 Vg U---~If\IV--+------I+
6.39. Express vet) in terms of step functions. ~--+---oV
1 +
241~
3A
12V v
10
fiGURE P6.44
FIGURE 1"6.31
FIGURE P6.35
~.45. Find for t > 0 the current downward in the capacitor
6.32. Find i for t > O. Assume dc steady state at t = 0-. If v(O) = 0, and (a) Vg = 4 V, (b) Vg = 2e- 2t V, and
4 (c) Vg = 2 cos 2t V.
The controlled source has transresistance of 3 n. 6.36. Repeat Problem 6.35 using the method of Thevenin
transformation of all but the storage element. 2
1SQ
6.37. Find i for t > O. Assume dc steady state at t = 0-.
I? t
-4 -3 -2 8
lOQ
12V 12 V
12 V
FIGURE P6.39 +
v
fiGURE 1"6.32 6.40. Sketch the voltage vet) = -3u( -t) + 3u(t + 2) +
2u(t) - 2u(t - 4).
6.41. Express
ramp functions.
too v(T)dT, vet) from 6.39, in terms of
6.33. Repeat Problem 6.32 for the case in which the switch FIGURE P6.45
FIGURE P6.37
closes, rather than opens, at time t = O.
Problems 2G7
2GG Chapter 6 First-Order Circuits
6.46. Find v for all time if Vg = 2u(t) V. 6.49. Find V2 if the capacitive voltage at time t = 0- is Design Problems 6.58. Find all currents and voltages in this circuit. Assume
O. Check using SPICE. 6.53. Interference from a car's igmtIon system induces i(O-) = O.
4000 identical equally spaced pulse voltage waveforms per
second across the antenna wire when it is open circuited. i
The rise and fall times are TR = TF = 1 tLS (see Fig. 6.36) ----72mH
and full-amplitude width is PW = 8 tLS. The pulse has base-
10MQ
v line VI = 0 and amplitude V2 = 1 mY. If the radio is a
Vg +
+ 100-kQ resistive load as seen by the antenna, determine an
+
ideal inductor that can be placed in series with the radio in-
IV I kQ V2
put which reduces the peak-to-peak interference across the
weFI
I
radio input to 0.1 mV 5%. This is called an RF sup-
pressor. Model the antenna as having a 100-kQ Thevenin
equivalent resistance.
- FIGURE P6.49 6.54. A certain loudspeaker\s modeled as an 8-Q resistor
FIGURE P6.46 in series with a 2-mH inductor. Design a circuit to connect FIGURE P6.58
6.50. Find V2 and v for t > to if veto) = 4 V. Check using the loudspeaker to an independent voltage source Vg so that
6.47. Find v for all t if (a) Vg = 2u(t) V, and (b) Vg = SPICE. the time constant of the overall circuit is 1 ms and the power
2[u(t) - u(t - 1)] V. dissipated by the loudspeaker when Vg = 10 V dc is 1 W. 6.59. Find v(t), t > 0 if the circuit is in dc steady state at
+ v _ t = O.
6.55. Design an RC circuit whose zero initial condition
response to the current input ls(t) = u(t) - 2u(t - 1) has
a zero crossing of the capacitive voltage at time t = 2 s.
2kQ
IQ
More Challenging Problems
+
o
....
.... 6.56. Find the time constant T. +
2Q
>----11---0 v
.... v
V2
+ 1f1F
R FIGURE P6.59
FIGURE P6.50
6.60. Find the voltage transfer equation for V2(t) in terms
fiGURE P6.47 6.51. Solve Problem 6.18 using SPICE. of VI (t).
6.52. Use SPICE to determine the output of the circuit to FIGURE P6.56
SPICE Problems
a unit step input. Plot the response from t = 0 to t = IH
6.48. Find v and for t > O. Assume dc steady state
VI
20 ms. Use subcircuits to describe the op amps and the
f
at t = 0-. Check usin SPICE. The controlled source has
integrators.
6.57. Find VI, V2, and V3, for t > O. Assume the circuit
is in dc steady state at time t = 0-.
transconductance g = 8" S.
40Q +
75 Q V2
t= 0
r~F
+
I rnF 1 kQ lkQ lmA
+ +
5V lOQ vI
40 FIGURE P6.60
+ v2 _
8 rnF 6.61. Repeat Problem 6.29 for the case in which the switch
moves a --+ b at t = 0 and then moves back to a at time
FIGURE P6.48 FIGURE P6.52 FIGURE P6.57 t = 3 s.
+
+
V2
FIGURE P6.62
Samuel F. B. Morse
1791-1872
Samuel F. B. Morse
!01:):.;:u;lst~~vll;.~.a~il~g~:s~t!ts'ithe; son.,ofa'~m.sterarld.
1\.dad~thyofArts in '.
was' c4;,n~idetedlto t:>emoaeratelV successful. In 18Z6he
t:le(;ante the;frrstp.t'esid~ritoftlie Nationa1:AcademyOf Design:
But the previous yearhiswit'e~a~died,in rS26 his fatlier died, and iri1828
his motlier,(!ied. The (ol1ow~ngyelifth~ ~str~ssed Morse ;wenpo Europe to
recover and study further: In 1832, ",hileret1.lrninghome on boardtliepassenger
ship Sully, he met an eccentric inventor and becaIlle intrigued with developing
a telegraph, theprindple of which had already been considered by Henry. By
IS36 Morse had a working model, and. in 1837 he acquired a partner, Alfred
Vail, who financed the project. Their efforts were rewarded with a patent and
the financing by Congress of a telegraph in 1844, over which Morse--on May
24, 1844-sent his now-famous message, "What hath God wrought!"
, -4i1
di2
+ -dt + 4i2 = 0 (7.1b)
811 2H
From the second of these we have
= -I (di
11. 2
- +412 . )
(7.2)
IH
4 dt
which differentiated results in
di1 = ~ (d 2i2 +4di2) (7.3)
The circuit analysis equations for linear circuits with energy storage elements may be dt 4 dt 2 dt
expressed as linear differential equations, because the element law.s are such that each FIGURE 7.1 Second-order circuit with two Substituting (7.2) and (7.3) into (7.la) to eliminate iI, we have, after
term in the mesh or nodal equations is a derivative, integral, or multiple of the unknowns inductors.
multiplying through by 2, the describing equation for i 2 :
and the source variables. Evidently, a single differentiation will remove any integrals
that it may contain, so in general the mesh or nodal analysis equations for a given d 2i2 di2
-dt 2 + 10- + 16i2 = 2vg (7.4)
circuit may be considered to be differential equations. The describing equation, a si~gle dt
equation whose only unknown is a selected output current or voltage, may be obtamed This is a second-order differential equation (one in which the highest
by manipulation of these analysis equations. order of differentiation of the unknown is 2). For this reason we
The circuits containing storage elements that we have considered so far were first- refer to Fig. 7.1 as a second-order circuit and note that, typically,
R
order circuits. That is, they were described by first-order differential equations. This is second-order circuits contain two storage elements.
always the case when there is only one storage element in the circuit, only one remains The two storage elements in a second-order circuit may be of
after series-parallel simplification, or when a switching action divides the circuit into two the same type, as in Fig. 7.1, or one inductor and one capacitor, as
L in Fig. 7.2. This is the important series RLC circuit, which we return
or more independent subcircuits each having at most one storage element. .
In this chapter we consider second-order circuits, which, as we shall se~, cont~n to throughout the chapter. By KVL,
two storage elements and have describing equations that are second-order differential
equations. We show that the total response is the sum of a natural and forced response, Ri + I
C
It
to
i(r) dr
d'
+ vc(to) + L~
dt
= Vg
as with first-order circuits. Examination of the characteristic equation shows the natural
FIGURE 7.2 Series RLC circuit. Differentiating and dividing by L yields
response of a RLC circuit is not limited to real exponential decays, as with RC or RL
circuits, but may also include sinusoids, damped sinusoids, and t-multiplied forms. The d 2i R di I dVg
forced response is determined from a trial form as with first -order ~ircuits, and th~ t?~al dt + L dt + LC i = dt
2
response is computed by requiring that two initial conditions, denved from the imtIal which is a second-order differential equation; hence the series RLC
energies stored in the two storage elements, be satisfied. . circuit is also a second-order circuit.
In general, nth-order circuits, containing n storage elements, are descnbed by nth- There are exceptions to the rule that two-storage-element cir-
order differential equations. The results for first- and second-order circuits may readily be cuits have second-order describing equations. If two or more storage
extended to the general case using the differential-equations-based methods of Chapter.s 5 elements of the same kind (inductors or capacitors) can be replaced
and 6 but we shall not do so here. There is a more efficient method for analyzmg by a single equivalent, they count a~ a single storage element in
these higher-order circuits based on Laplace transform analysis of the circuit, which is determining the describing equation orCter. In other cases, the stor-
introduced in Chapter 12. age elements may not interact. For example, consider the circuit of
272 Chapter 7 Second-Order Circuits Section 7.1 Circuits with Two Storage Elements 273
Fig. 7.3, which has two capacitors. Nodal analysis equations for this sources. As an example, for the circuit of Fig. 7.1, the describing equation is (7.4).
circuit are Comparing this equation with (7.6), we see that aj = 10, ao = 16, f(t) = 2v g, and x = i2.
dVj
dt +Vj = Vg
From Chapter 6 we know that the total response satisfying (7.5) is
x(t) = xn(t) + xf(t)
d V2
dt + 2V2 = 2vg where xn(t) is the natural response, obtained by setting f(t) = 0, and xf(t) is the forced
response, which satisfies the forced differential equation (7.5).
These are uncoupled first-order differential equations; that is, neither
unknown appears in both equations. Each may be solved using the Let ~s see if this same procedure can be applied to the second-order equation (7.6).
FIGURE 7.3 First-order circuit with two storage By a solutIOn to (7.6), we shall mean any function x that satisfies (7.6) identically. That
elements. methods of Chapter 6. Since the describing equations are first order,
this is not a second-order circuit even though two storage elements is, when x. i~ substitu~ed into (7.6), the left member becomes identically equal to f(t)
for all t WIthIn a specIfied solution region, usually t > to for some initial time to.
irreducible to a single equivalent are in the circuit. Replacement of
If Xn is the natural response, it must satisfy the unforced (or homogeneous) differ-
the ideal voltage source by a practical one with nonzero Thevenin
ential equation
equivalent resistance in this circuit, however, results in a second-
d2Xn dX n
order circuit, as shown in Exercise 7.1.2. dt 2 + aj dt + aOXn = 0 (7.7)
7.1.1. Find the equation satisfied by the mesh current i2 Adding (7.7) and (7.8) and rearranging the terms, we may write
2Q 3Q Answer d 2i2/dt 2 + 7(di2/dt) + 6i2 = dvg/dt
d2 d
7.1.2. Place a l-Q resistor in series with the ideal voltage source Vg in dt 2 (xn + xf) + al dt (xn + xf) + ao(xn + xf) = f(t) (7.9)
Fig. 7.3, and find the describing equations for Vj and V2
IH d 2vj 11 dVj 4 2 dVg 4 Com?aring (7.6) and (7.9), we see that the sum of Xn and xf is indeed a solution, as it
Answer dt 2 + 5 d t + SVj = Sdt + SVg; was In the first-order case. That is, x satisfying (7.6) is made up of two components, a
d 2v2 11 dV2 4 4 dV2 4 natural response Xn satisfying the homogeneous equation (7.7) and a forced response x
dt2 + 5 d t + SV2 = Sdt + SVg satisf~ing the o:iginal equation (7.8) or (7.6). As we shall see, the natural response wih
EXERCISE 7.1.1 7.1.3. Find Vj and V2 in Fig. 7.3 for t > 0 if Vg is a +6-V ideal dc contaIn two arbItrary constants and, as in the first-order case, we will choose a trial form
source, Vj(O-) = 1 V, and V2(O-) = 4 V. forced solution so that it will end up having no arbitrary constants. We consider methods
Answer 6 - 5e- t V; 6 - 2e- 2t V of finding the natural and forced responses in the next two sections.
. If the indepen~ent sources are such that f(t) = 0 in (7.6), the forced response
IS zero and the solutIOn of the differential equation is simply the natural response. A
reader who has had a course in differential equations will note that the natural response
and the forced response may also be called, respectively, the complementary solution
and the particular solution. The natural or complementary solution contains the arbitrary
constants needed to match initial conditions, as will be discussed in Section 7.5, and the
In Chapter 6 we considered first-order circuits and saw that their describing equations particular solution, as its name implies, contains no free constants.
were first-order differential equations of the general form
dx
dt +ax = f(t) EXERCISES
In Section 7.1 we defined second-order circuits as those with describing equations
were second-order differential equations, given generally by
7.2.1. Show that Xj = Kje- 2t and X2 = K2e- 3t are each solutions of
d2x dx
d 2x dx dt 2 + 5 dt + 6x = 0
dt 2 + al dt + aox = f(t)
regardless of the values of the constants Kj and K2.
In (7.5) and (7.6) the a's are real constants. x may be either a voltage or a current, 7.2.2. Show that Xj +X2 = Kje- 2t + K2e- 3t also satisifies the equation
and f(t), the forcing function, is a known time function determined by the incieplend.ent of Exercise 7.2.1, regardless of the values of Kj and K2.
(7.16)
The natural response Xn, a comonent of the total solution x = Xn + xf, must satisfy the is a solution of (7.10) for any pair of values Kl and K2 as long as SI and S2 satisfy the
characteristic equation. To verify this, we need only substitute this expression for x into
unforced equation, which we repeat from (7.7):
(7.10). This results in
d 2x dx (7.10)
- +al- +aox = 0 d2 d
dt 2 dt dt 2 (Xnl + Xn2) + al dt (Xnl + Xn2) + aO(xnl + Xn 2)
The trial form for the natural solution
d 2xnl dXnl ) (d 2xn2 dXn2 )
xn(t) = Ke st
(7.11) = ( df2 + al dt + aOXnl + df2 + al dt + aOXn2
worked well previously for the first-order case. Let us see if ~t w.ill help with the sec~nd =0+0=0
order case. Any natural solution must satisfy (7.10) so substitutmg Xn from (7.11) gIves
Since (7.16) includes the individual solutions (7.15a) and (7.1Sb) as special cases, it is
Ks 2est + Ksal est + Kaoest = 0 called the general solution of the homogeneous equation if SI and S2 are distinct (i.e., not
equal) roots of the characteristic equation (7.12). Note that if we put K2 = 0 in (7.16)
or we have Xnh and Kl = 0 results in X n 2.
If the trial form works, this equation must be true for each t or ~lse (7 .10) wo~ld fail
to hold. Equation (7.12) will hold if either K = 0 or the factor m p.arentheses ~s ;~~~ As an example, the homogeneous equation corresponding to (7.4) in
(est cannot be zero for any time t). If K = 0 we have the null solutIOn, xn(t) - Example 7.1 is repeated here,
each t. This corresponds to the special case of no stored energy. M~re generally, we d 2i2 di2
will not have Xn = 0 at the initial time, so we cannot have K = O. ThIS leaves only the dt 2 + lOTt + 16i2 = 0 (7.17)
possibility that the other factor in (7.12) is zero, or and the characteristic equation is
S2 + al S + aos = 0 s2 + lOs + 16 = 0
. 't Note
This is the characteristic equation introduced in Chapter 6 for first-order ClfCUl s. . ... The roots are s = -2 and s = -8, so the general solution of the
that it can easily be derived from (7.10) simply by replacing derivatives by corr~sp~ndI~g homogeneous equation is given by
powers of s. That is, the second derivative of x is replaced by S2, the first denvatlVe Y (7.18)
S and the zeroth derivative of x, x itself, is replaced by sO = 1. fi
' I . . the rst- The reader may verify by direct substitution that (7.18) satisfies
Since (7.13) is a quadratic equation, we have not on~ so utIOn, as m 'c for-
order case, but in general two solutions, say SI and s2, gIven by the quadrati (7.17) regardless of the value of the unspecified constants Kl and K2.
Rep = ~~
For instance, a circuit with characteristic equation s2 + 2s + 16 has undamped natural For R > Rep the damping ratio ~ < 1, which is described below
frequency Wa = +.JT6 = 4 radls and damping ratio ~ = 2/(2)(4) = ~. The characteristic as the underdamped case, and for R < Rep we have ~ > 1, the
exponents for this circuit are overdamped case.
The circuit of Fig. 7.4 is called a parallel RLC circuit. The nodal Overdamped Case G > 1)
equation for this circuit is
- + - 1t
With ~ > 1 the characteristic exponents Sl and S2 in (7.20) are real and distinct, since
v 1 dv the discriminant ~2 - 1 > O. In this case each s has only a real part,
vCr) dr + iLCta) + C-
dt = ig
R L ~ Sl = ai, S2 = a2
Differentiating and dividing both sides by C, the describing equation and the natural response is given by a sum of two real exponentials,
for v is
d2v 1 dv 1 1 dig (7.21)
-+ --+-v=--
dt
2 RC dt LC C dt
(7.22)
In this case the general form for the natural solution given in (7.16) does not contain two
free constants, since both terms may be collapsed into one. If we are to be able to meet
We need to use some properties of complex quantities to clarify this form and initial conditions imposed by two storage elements, we must further generalize the trial
recognize its structure. For a real circuit, one with purely real element laws and sources, form for the natural solution. Consider the trial form
Xn will be real as well. This requires KI and K2 to be complex conjugates, since otherwise
(7.26)
the imaginary part of Xn in (7.22) would not be zero. Since the sum of a complex number
and its conjugate is twice its real part, Thus far we have used h(t) = 1. To see if another h(t) may also work, we substitute
(7.25) into the unforced differential equation (7.10). Using the product rule to compute
(7.23) derivatives, after gathering like terms, we obtain
where K2 = Ki is the complex conjugate of K I . The real part of a product of complex
numbers is the product of their real parts minus the product of imaginary parts. Using (7.27)
the Euler identity,
Evaluating this equation for the present repeated root case s = a, the first term in brackets
e jwt = cos wt + j sin wt (7.24)
is zero since a satisfies the characteristic equation, and the second term is zero since
we may expand the real part of the product in (7.23) as
+ als + ao = (s - a)2 = s2 - 2as + a 2
s
Re(Kle+ jwt ) = !BI coswt - !B2 sinwt
shows that al = -2a, so 2s + al = 2a - 2a = O. Thus (7.26) is satisfied if and only if
where for convenience we have defined
d2h
-=0
dt 2
Graphs for these four distinct responses are shown in Fig. 7.5.
and the damping ratio for the series RLC case is
R
~s = 2JC/L
(a) (b)
The critical series resistance Res is found by solving for R with
~ = 1:
282 Chapter 7 Second-Order Circuits Section 7.3 Natu ral Response 283
In the parallel RLC circuit of Fig. 7.4 and Example 7.3, let R = ~ n, 7.3.2. In the series RLC circuit of Fig. 7.2, let Vg = 0, R = ~ Q,
L = 2 H, and C = to
F with no source, ig = O. If v(O-) = 4 V L = ~ H, and C = ! F. Find the natural solution in.
and there is no current through the inductor at time t = 0-, find Answer in = Kle- 2t + K2e-18t
v(t) for t > O. 7.3.3. Write nodal equations for the circuit shown. Solve the VI nodal
IF
4 equation for V2, and use this to eliminate V2 from the differentiated V2 nodal
With no independent source, there will only be a natural re-
sponse. The describing equation, from Example 7.3, is equation, yielding the describing equation for VI. For what value of R > 0
is the circuit critically damped?
d 2v 1 dv 1 1 dig Answer d2vI/dt2+(R+l)(dvl/dt)+(R+4)VI = dvg/dt-Rvg;
dt 2 + RC dt + LC v = edt R=5Q
d 2v dv EXERCISE 7.3.3 7.3.4. Find the natural response iln for the circuit shown.
or -+6-+5v=0 Answer iln = Kle-(2+V2)t + K2e-(2-V2)t
dt 2 dt
The characteristic equation is IH IH
S2 + 6s + 5 = 0
or Sl = -1 and S2 = -5. This is an overdamped natural response,
1[l 2H
v = K1e- t + K 2e- 5t (7.34)
The capacitive voltage is continuous, so v(O+) = v(O-) = 4 V. By
KCL,
EXERCISE 7.3.4
3 1 dv
-V+iL + - - = 0
5 10 dt
where i L is defined with its reference arrow pointing downward in
Fig. 7.4. Evaluating this equation at t = 0+ yields
3
-(4)+0+-- 1 dv I =0
5 10 dt 0+
The forced response xf of the general second-order circuit must satisfy the original, or
where we have used continuity of inductive current to evaluate
forced, differential equation (7.8), repeated here:
iL(O+). Then at t = 0+, v = 4 V and dv/dt = -24 Vis. Ap-
plying these conditions to (7.34), we have d 2xf dXf
-d 2 + a l - +aOxf = f(t) (7.35)
v(0+)=4=K 1 +K2 t dt
There are a number of methods for finding xf' but for our purposes we use the pro-
-dv I = -24 = -KI - 5K2 cedure of guessing a trial form solution that has worked well for us thus far. Ex-
dt 0+ amining (7.35), we note that for this equation to be true, a linear combination of xf
Adding gives and its derivatives must equal f(t). Consider a trial solution in which xf is guessed
-20 = -4K2 to be a linear combination of f(t) and its derivatives. Since a linear combination of
or K2 = 5. Then back-substituting, K I = -1 and the solution is linear combinations is still a linear combination, this trial form might well work. In
other words, the differentiations of the linear combination of f(t) and its derivatives
v = 5e- 5t - e- t V required on the left side of (7.35) may result in the cancellations required to leave
only f(t) itself, which is what is needed to satisfy the equation. The adequacy of
any trial form will, in the final analysis, be judged by its ability to satisfy the differential
EXERCISES equation.
7.3.1. What unforced second-order differential equation yields charac- As an example, let us consider the de source Vg = 16 V in the circuit
teristic exponents Sl = -10 and S2 = -3? What are wo and n
Repeat for of Fig. 7.1. Then, by (7.4),
Sl = -2 + j5 and S2 = -2 - j5.
Answer d 2x/dt 2 + 13(dx/dt) + 30x = 0; Wo = 5.48 rad/s, ~ = d 2i2 di2
(7.36)
1.19; d 2x/dt 2 + 4(dx/dt) + 29x = 0; Wo = 5.39 rad/s; ~ = 0.37 dt 2 + lOdt + 16i2 = 32
or A = 2. The forced solution is then The new trial form thus works if we set A = 6, or
if = if I + i12 = 2 + 2e-7t vf = 6te- t (7.43)
which can be verified by direct substitution into the forced differen- The existence of an A for which the new, t-multiplied trial forced solution works (satisfies
tial equation. the forced equation) is sufficient to justify our guess. Equation (7.43) is the forced
solution, and the total solution is
In the circuits and describing equations studied so far, we have used trial forms v= Vn + vf = Kle- t + K 2 e- 2t + 6te- t
for the natural and forced solutions that have been chosen completely independently of
Generalizing, when the trial forced solution matches a term in the natural solution, it is
one another. The form for the natural solution depended only on the left side of the
necessary to t-multiply the trial forced solution. If there is still a match, t-multiply again,
describing equation (through the coefficients of the characteristic equation), while the
form for the forced solution depended only on the right side [f(t) itself]. There is one and so on, until there is no match.
Finally, if the trial forced solution matches a t-multipliedform in the natural solution,
condition under which the natural and forced trial solution forms cannot, however, be
determined independently. it may be necessary to t-multiply the trial solution further, as demonstrated in Example 7.9
Consider the following equation, for which the trial forced solution happens to and in Exercise 7.7.3.
match a term in the natural solution.
d 2v dv
- + 3 - + 2v = 6e- t (7.40) d 2v
dt 2 dt dt 2 = f(t)
The characteristic equation is
The characteristic equation is s2 = 0, or Sl = S2 = O. This is a
S2 + 3s + 2 = (s + 1)(s + 2) = 0 repeated root case, and by (7.28) the natural solution is
and natural solution Vn = Kle st + K2test = KI + K2t
(7.41) First consider the case where the forcing term is f(t) = 2. The trial
forced solution is vf = A. This matches one of the terms (KI) of the
From Table 7.1, for f(t) = 6e- t the trial forced solution [consisting of the general linear
natural solution, so we t-multiply vf, yielding vf = At. This still
combination of f (t) and all its derivatives] is
matches a term in the natural solution (K2t), so, recognizing that At
(7.42) would still satisfy the natural, not the forced, differential equation,
EXERCISES
IH
d (dVI
2- - + VI - 12) + 3 (dVI
- + VI - 12) - VI =0
or d; 0 = 2 + 6 -
d'l 5(1) =3 dt dt dt
or, after division by 2,
Using these values, (7.46) may be rewritten as
d 2vI 5 dVI 18
KI + K2 =-1 (7.47a) dt2 + 2dt + VI =
(7.50)
Differentiating the second mesh equation, we obtain Using BI = 12, this implies that B2 = -12, and the solution (7.57) is
di2 di] i l = 12e- t (cos t - sin t) A
2di - di + 2i2 = 0 (7.56)
To get the describing equation for ii, solve (7.55a) for i2 and sub-
stitute into (7.56): EXERCISES
d
2-
dt
(1 di] .) di] di] )
--+ZI - - + 2 --+i] =0
2 dt dt 2 dt
(1 7.5.1. Find x(t) for t > 0, where
or IH
d~
dt
+ 2x + 1t0
x dt = J(t)
Vg = u(t) V _ IH
di
I-+Ri+4
dt
1t 0+
i(r)dr+ve(O+) = 1 (7.58)
or
v
e
= 1 - 4(te- 21 )
ve(t) = 1 - e- 21 -
- -(te- 21 )
dt
2te- 21 V (7.64)
or, after differentiating,
Note that the step response consists of a transient natural response
d2i di of the form (7.59), plus a dc steady-state response of 1 V. In dc
- +R- +4i =0
dt 2 dt steady state, the capacitor is an open circuit and 1 V is the value of
fiGURE 7.10 Circuit for Example 7.12. This is an unforced second-order describing equation for the series the unit step source voltage across the open circuit. This critically
current i, which was taken up in Example 7.4 for several different damped step response is shown in Fig. 7.11.
values of R. The critical resistance for this series RLC circuit is, by We next determine an overdamped unit step response Ve with
(7.29), Res = 2J47I = 4 Q. For this value of R, by (7.30), R = 5 Q. From (7.31),
d',
-.!:..
dt 0+
= 1 = K2
vr
Overdamped step response (R
sin Vgt
1
= 1 - e- I/8 cos gf + ..1255 A Date/Time run: 05/11193 11 :26:00 Temperature: 27.0
LOV
VT
two terms, one of which has a longer time constant (T = 1 s com- Critically damped step response (R = 4 ohms)
!
pared to s for the critically damped case) and thus decays to zero
Date/Time run: 05/11/93 10:25:41
LOV
Temperature: 27.0
more slowly than the critically damped case. The transient in the un-
derdamped response is oscillatory, with the amplitude of these oscil-
lations decaying to zero with an even longer time constant, T = 8 s.
Thus, although the underdamped response first reaches the value of
1 V much faster than the other responses, it "overshoots" this value, 0.0 V 1oC'-~_------;:L-- _ _ _ __ L_ _ _ _ _LI_ _ _ _ _LI_ _ _ _- - '
and its actual convergence to steady state is even slower than the 2s 4s 6s 8s lOs
overdamped case. FIGURE 7.12 PROBE output for three PSpice runs.
EXERCISE 7.6.3
17.0 nF
1.050E+OO -4.128E-02 *
Differentiating and dividing by L yields 1.100E+OO -1.570E-02 *
1.150E+OO 9.871E-03 *
IF d 2i R di 1. 1 d 1. 200E+OO 2.922E-02 *
dt2 + L dt + LC I = L dt vs(t) (7.68) 1. 250E+OO 3.871E-02 *
1. 300E+OO 3.673E-02 *
fiGURE 7.14 Circuit to compute a differential 1.350E+OO 2.347E-02
This equation is of the desired form. Setting L = C = 1 and R = P 1.400E+OO
*
equation. 2.987E-03 *
and identifying the mesh current i(t) as the unknown y(t) in the 1.450E+OO -1.881E-02 *
1.500E+OO -3.664E-02 *
differential equation (7.67) that we are solving by the simple analog 1.550E+OO -4.511E-02 *
computer circuit of Fig. 7.14, we need only match forcing terms. 1.600E+OO -4.140E-02 *
1.650E+OO -2.678E-02
Since the right-hand sides of (7.67) and (7.68) match for L = 1 with 1.700E+OO -4.017E-03
*
1.750E+OO 2.220E-02
vs(t) = 4sin lOt 1.800E+OO
1.850E+OO
4.545E-02
6.089E-02
*
*
this is the desired voltage source in the circuit. The SPICE input 1.900E+OO 6.552E-02 *
1.950E+OO 5.780E-02
file is 2.000E+OO 4.019E-02 *
- - - - - - - - - - - - - - - - -
Design Example 7.2 FIGURE 7.15 SPICE plot for Design Example 7.2.
*VS 1 0 S1N(O 05 1.59) The output is shown in Fig. 7.15. Note that the forced response,
1.59) = (xoff xampl freq)
*R (0 05
1 2 .1
which is a sinusoid at w = 10 rad/s (l.59 Hz), is superposed with
the natural response, an underdamped sinusoid with much lower fre-
L 2 3 1 1C=O quency.
C 3 0 1 1C=O
.TRAN .05 2 U1C Design an active analog computer circuit that solves the equation
.PLOT TRAN 1(VS)
.END d2 d
dt 2 y(t) + dt y(t) = x(t)
0
8'"' I 1.15",
+
1-"'("
dy + -y + -1
c-
dt R L
f [-x(t)]dt = 0
EXERCISE 7.7.1.
IF
20mF
1$1
+
+
+
+
x(t) -x(l)
yet)
EXERCISE 7.7.2.
fiGURE 7.16 Circuit to compute another differential equation.
There are many other uses to which second-order circuits may be put. We did not
consider filters, perhaps the most important use of all, as this topic is considered in detail
in Chapter 14, and general methods for second-order filter design are discussed at length
in Chapter 18. The design examples and exercises here, though, begin our exploration Second-order circuits are those characterized by a single second-order differential equa-
of second-order circuit design. tion. They can be identified by the presence of two equivalent storage elements (inductors
and/or capacitors). Their behavior may be studied by writing and solving their describing
equations in the time domain using the techniques introduced in Chapter 6 for first-order
circuits. The natural behavior of second-order circuits is, however, much more diverse
and interesting, ranging from pairs of real exponentials and t-multiplied exponentials to
sinusoids both damped and undamped.
DESIGN EXERCISES
The natural response of a second-order circuit is K I e P1 t + K2e P2t , where PI and P2 are
7.7.1. Design a circuit whose step response includes real exponentials the distinct roots of the characteristic equation, or if PI is a repeated root, the second
with time constants T = 10 ms and T = 1 ms. term is replaced by the t-multiplied form K2teP,t.
20 IF
FIGURE P7.7
1k0 4kO 2kO
20 I kO 2kO 2j.tF 7.8. Consider the series RLC circuit shown in which the
resistance is time varying. (a) Write the describing equation
for i. (b) If in and if are natural and forced solutions, is
their sum a solution?
FIGURE P7.13
(a) R(t)
fiGURE P7.2
10
7.14. Find the natural response i (t), t > 0 of the circuit
of Problem 7.1(a) if i(O) = 3 A (ig = 0).
L 7.15. Find i for t > 0 if il (0) = 9 A and i(O) = 3 A.
r::) 7.3. Write a second-order differential equation satisfied
by V2 in the diagram for Problem 7.2.
lH 7.4. Repeat Problem 7.3 if the capacitor is replaced by a
ig t 4
IH
10
!-H inductor.
7.5. Insert a l-Q resistor in series with Vg in Fig. 7.3 FIGURE P7.B
thereby making the source a practical rather than an ideal
r7:J one. Show that in this case V2 satisfies the second-order
equation,
7.9. For a second-order differential equation 10 40 20
d 2x dx
20 dt 2 + al dt + aox = f(t)
(b)
d 2v2 dVg dV2 let Xnl and Xn2 be natural solutions and Xfl, Xf2 be two
5 -2- + 11- +4V2 = 4 - +4V2
fiGURE 1'7.1 dt dt dt distinct forced solutions. Show that any linear combination FIGURE 1'7.15
I
r
7.16.Find i for t > 0 if the circuit is in steady state at 7.23. Sketch a circuit whose describing equation is d;tV + 2Q IH 1=0 2 Q
~
t =0-. 2 dv
dt + V -- 2 sin 2t .
r
Hl +
3Q v -=-SV
iH 3Q 3Q
16V -=- 4Q Hl
FIGURE P7.16 7.28. Find i for t > 0 if the circuit is in steady state at
7.31. Find i, t > 0, if there is no initial stored energy and
t = 0-. \
7.17. Find i for t > 0 if i(O) = 4 A and v(O) = 8 V. FIGURE P7.24 (a) R = 2 n, J-t = 2; (b) R = 2 n, J-t = 1; and (c) R = 1 n,
J-t=2.
4H t= 0
7.25. Find the forced solution vf(t).
1=0 4Q 12 Q
+ lp
lp v SQ
v
2Q 4
4 R 4Q
+ vz _ FIGURE P7.31
FIGURE P7.25
7.29. Find i for t > 0 if the circuit is in steady state when
lp the switch is opened at t = O.
8 7.26. Find the forced solution if(t)
SQ 7.32. Find i for t > 0 if i(O) = 2 A and (a) v(O) = 6 V.
1=0
2Q lp + 4Q
8 3Q
+
26 cos 3t V lp v
6
IQ 5H
FIGURE 1>7.19
cos 4tV
fiGURE 1'7.33
fiGURE 1>7.38
4!1 S!1
7.34. Solve Problem 7.33 if the switch closes at t = 10 s
instead of t = 0 s. 7.39.
fiGURE 1>7.46
7.35.
2H 4!1 7.47. Find the mesh currents for t > 0 assuming all initial
fiGURE 1'7.43 conditions are zero at t = O.
2!1
4!1 7.44. Find the unit step response. Check using SPICE.
t=O
16V
100k!1
2!1
FIGURE 1'7.39 +
12V
FIGURE 1'7.35
7.40. Find the total solution vet) if R is set to the critical
resistance (critical damping).
7.36. +
2!1
8V IH FIGURE 1'7.48
2H
7.49. Write two different second-order differential equa-
IF
tions that both have the same forced solution XI! = X2! =
2t + 1 but that have different forms for their natural solu-
fiGURE 1'7.37 FIGURE 1'7.41 FIGURE 1'7.45 tions. Solve both equations assuming zero initial conditions.
20 !! R
FIGURE P7.51
d2y dy
dt 2 - 2 dt + 2y = 0
FIGURE 1>7.50
Specify what variable in your circuit y(t) is, and how the Charles Proteus Stei nmetz
initial conditions y(O), dy Idt It=o are determined. 1865-1923
Design Problems 7.53. An electric motor is modelled as a 22-Q resistor in
7.51. Find a value for R such that the natural responses of parallel with a l/2-H inductor. Design a passive circuit I have found the equa-
this circuit all consist of damped sinusoids K eat cos(wt +8) to connect the motor to an ideal voltage source so that the tion that will enable us to
resulting circuit is critically damped at its natural frequency transmit electricity through
with w = 3000 rad/s. Use SPICE to help search for the
of 60 Hz. alternating current over
solution.
thousands of miles. I have
reduced it to a simple prob-
lem in algebra.
315
314 Chapter 7 Second-Order Circuits
Chapter Contents for analyzing ac circuits with no more difficulty than dc circuits, called phasor analysis,
is one of the greatest intellectual achievements in the field of engineering analysis. Mod-
II 8.1 Properties of Sinusoids II 8.7 Kirchhoff's Laws and em engineering, electrical and otherwise, would be unthinkable without the power and
II 8.2 RLC Circuit Example Impedance Equivalents versatility of phasor analysis, the principal subject of this chapter.
II 8.8 Phasor Circuits
II 8.3 Complex Sources
III II Summary
8.4 Phasors
III Problems
II 8.5 Current-Voltage Laws for
Phasors We devote this section to a review of some of the properties of sinusoidal functions. Let
II 8.6 Impedance and Admittance us begin with the sine wave,
which is sketched in Fig. 8.1. The amplitude of the sinusoid is Vm , which is the maximum
value that the function attains. The radian frequency, or angular frequency, is w, measured
in radians per second (rad/s).
In Chapters 6 and 7 we analyzed circuits containing storage elements and have seen The sinusoid is a periodic function, defined generally by the property that there is
that the complete response is the sum of a natural and a forced response. The natural a smallest number T such that for all t,
response is obtained after killing all independent sources and therefore is not a function
of these sources, which are also known as excitations (unlike dependent sources, they
excite responses without requiring coupling with other sources). The forced response,
on the other hand, depends directly on the functional form of excitation applied to the
circuit. In the case of a dc source, the forced response is a dc (constant) response, an where T is the period. That is, the function goes through a complete cycle, or period,
exponential input evokes an exponential forced response, and so on. which is then repeated every T seconds. In the case of the sinusoid, the period is
Perhaps the single most important type of excitation is the sinusoid. Sinusoids are
2]"(
found everywhere in nature: for example, in the motion of a pendulum, the propagation T = - (8.3)
of light or sound waves through space, and the vibration of strings or steel beams. As W
we have seen, any undamped second-order circuit produces a sinusoidal natural response, as may be seen from (8.1) and (8.2). Thus in 1 s the function goes through liT cycles,
and any underdamped second-order circuit will have a decaying sinusoid as its natural or periods. Its frequency f is then
response. 1 W
f=-=- (8.4)
In electrical engineering and technology, sinusoidal time functions are found at the T 2]"(
core of many, perhaps most, important applications. The carrier signals generated for cycles per second, or hertz (abbreviated Hz). The latter term, named for the German
communication systems are sinusoids, and the sinusoid is dominant in the electric power
industry. Indeed, as we shall see later in the study of Fourier series, almost every useful vet)
signal in electrical engineering can be represented as a sum of sinusoidal components.
Because of their importance, circuits with sinusoidal excitation, or ac circuits, are
considered in detail in this and several subsequent chapters. Since for all linear circuits
the natural response is independent of the excitation and can be found by the methods of
earlier chapters, we concentrate on finding the forced response to sinusoidal excitation.
This response is important in itself since for stable circuits it is the ac steady-state response
that remains after the time required for the transitory natural response has passed.
Being interested only in the forced response, we shall not limit ourselves, as we
did in Chapters 6 and 7, to first- and second-order circuits. As we shall see, higher-order
RLC circuits may be handled, insofar as the forced response is concerned, in much the
same way as resistive circuits were handled in Chapter 2. The discovery of this technique FIGURE 8.1 Sinusoidal function.
316 Chapter 8 Sinusoidal Sources and Phasors Section 8.1 Properties of Sinusoids 317
by a - {J. An equivalent expression is that Vz lags VI by a - (J. In Fig. 8.2, Vm sin wt
physicist Heinrich R. Hertz (1857-1894), is now the standard unit for frequency. The
lags Vm sin(wt+) by radians. Positive lagging phase implies a right shift in the graph
relation between frequency and radian frequency is seen by (8.4) to be
of the function.
As an example, consider
w=2n! (8.5) VI = 4 sin(2t + 30)
Vz = 6 sin(2t - 12)
A more general sinusoidal expression is given by Then VI leads Vz (or Vz lags VI) by 30 - (-12) = 42. The graph of VI is left shifted by
42 relative to Vz.
Thus far we have considered sine functions rather than cosine functions in defining
vet) = Vm sin(wt + ) (8.6) sinusoids. It does not matter which form we use since
, cos( wt - ~) = sinwt (8.7a)
where is the phase angle, or simply the phase. To be consistent, since wt is in radians,
should be expressed in radians. However, degrees are a very familiar measure for sin(wt + ~) = cos wt (S.7b)
angle. Therefore, we may write
The only difference between sines and cosines is thus the phase angle. For example, we
v = Vm sin( 2t + ~) may write (S.6) as
Then, by (S.Sa),
- sinwt = sin(wt + 180)
we have
Vz = 2sin(2t + ISO + 180)
= 2cos(2t + ISO + lS0 - 90)
where
A = Mcose (S.lOb)
B = Msine (S.lOc)
The decomposition of a sinusoid into sine and cosine components is called the quadrature As an example of a circuit with sinusoidal excitation, let us find the forced mesh current
representation of the sinusoid.
i in the series RLC circuit shown in Fig. S.4. By KVL,
Note that we must be clear on what is meant by (S.9b), since some mathematics
books take this expression as the principal value of the arctangent and place e in a specific
quadrant. We mean that the terminal side of the angle e is in the quadrant where the
d'
2--':
dt
+ 2i + 10 1t0+
i(r)dr + vc(O+) = 15cos2t (8.11a)
1
A lO F
fiGURE 8.3 Triangle useful in adding two sinusoids. fiGURE 8.4 Series RLC circuit with sinusoidal excitation.
320 Chapter 8 Sinusoidal Sources and Phasors Section 8.2 RLC Circuit Example 321
Differentiating and dividing by 2, +
d 2i di
- + - +5i = -15sin2t (8.11b)
dt 2 dt 4V2cos4t A
From Table 7.1, the trial forced solution is
i = A cos 2t + B sin 2t
Substituting this into (8.11) gives EXERCISE 8.2.1
(-4A cos 2t -4B sin2t) + (-2A sin 2t + 2B cos 2t) + 5(A cos 2t + B sin2t) = -15 sin2t 8.2.2. Repeat Exercise 8.2.1 if the inductor is removed from the circuit.
Answer 2cos(4t - 45) V
and matching coefficients of the cos and sin terms,
A +2B = 0 (8.12a)
-2A+B=-15 (S.12b)
Doubling the first of these equations and adding to the second yields 5B -15 or An alternative method for treating circuits with sinusoidal sources, which will be our focus
B = -3, and substituting into the first, A = +6. The forced response is of interest for this and the next several chapters, involves replacing the given sources by
complex sources, those whose source functions have real and imaginary parts. Currents
i = 6 cos 2t - 3 sin 2t A (8.13) and voltages in circuits excited by complex sources will themselves be complex valued.
These two quadrature terms may be combined into one using the conversion formulas Since we rely heavily on complex numbers and their manipulation, readers unfa-
(8.10): miliar with complex numbers or who seek to refresh their understanding should consult
Appendix B. For convenience, we review several key definitions and properties of com-
M=J6 2 +3 2 =3Vs plex numbers before going on.
Each complex number is a point in the complex plane. The complex number A is
e= tan -1 ( -63) = -26.6 written in rectangular form as
IAI = Ja + b
2 2 (S.l7a)
EXERCISES -1 b
a=tan - (8.17b)
a
8.2.1. Find the forced response vf.
Answer 2 cos(4t + 45) V The relation between rectangular and polar forms is shown in Fig. 8.5.
322 Chapter 8 Sinusoidal Sources and Phasors Section 8.3 Complex Sources 323
1m
The behavior of the complex exponential function e jwt is central to our studies. By
A =a + jb the polar form of the Euler identity (8.1Sb),
"'------.J'---_ _ _-'-_--.:~ Re Examining this expression, the magnitude of the complex exponential is always unity,
a while its angle increases uniformly at the rate of w radians per second. Thus the complex
fiGURE 8.5 Rectangular and polar representations.
exponential e jwt traces out unit circles in the complex plane, beginning on the positive
real axis at time t = 0 and moving counterclockwise, completing one full circle (or period)
every T = 2Jr /w seconds. The projection of this point onto the horizontal axis is its real
The important Euler identity is given by part (coswt) and that onto the vertical axis is its imaginary part (sinwt).
The general scaled and phase-shifted complex exponential Vme(jwt+) shown in
Fig. 8.6(b) is similar, except that at t = 0 its initial phase is radians and it traces out
circles of radius Vm . By horizontal projection,
Re[Vmej(wt+)] = Vm cos(wt + )
or, using (8.18a) and (8.17), in its polar form,
and by vertical projection,
Im[Vmej(wt+)] = Vm sin(wt + )
which can be verified by inspection of the rectangular form of the Euler identity.
We next tum to the subject of main interest, that is, the application of complex
numbers to electric circuits. Consider the linear circuit C of Fig. 8.7, which has indepen-
The latter form of the Euler identity may be used to generate a useful alternative way to dent source vg We wish to determine the forced response to the excitation vg Assume
write complex numbers in polar form. Since
that all other independent sources in C have been killed, along with all initial conditions.
A= IAI~ = IAI(l~) Now let fl (t) and h(t) be any two real functions of time. Suppose that when we use
the source Vg = II (t) we measure the forced response il (t), and when Vg = h(t) is
then by (8.18b) we have the exponential polar form for a complex number: used, the forced response is i2(t), as shown in Fig. 8.7(a) and (b). These responses to the
real inputs II and h are real. By the proportionality principle, scaling the excitation h
A = IAle ja (S.19)
by the constant j will scale the response i2 by the same factor. Then, by superposition,
Rectangular, polar, and exponential polar forms of the same complex number A will each the response to the sum II + j h will be the corresponding sum of separate responses
prove convenient in different contexts, and it is essential to be able to move easily among
them. 1m 1m
w rad/s
The three forms each describe the same complex number A, the FIGURE 8.6 (a) Complex exponential e jwt ; (b) general complex exponen-
same point A in the complex plane. tial Vme(jwt+,p).
324 Chapter 8 Sinusoidal Sources and Phasors Section 8.3 Complex Sources 325
Substituting the trial form i = Ae j2t , we have
(-4 + j2 + 5)Ae j2t = 15je j2t
Solving for A yields
15 .
(a) A = _1_ = 3-./5/26.6
1 + j2
The response i to this complex source is Ae j2t , or
i = (3-./5/26.6 0 )e j2t = 3-./5e j (2t+26.6) A
where the exponential polar form (S.19) has been used in the last
expression.
(b) Since the real part of 15e j2t is 15 cos 2t, the response to the
source Vg = 15 cos 2t V must be
Re[3-./5e j (2t+26.6 0
)] = 3-./5 cos(2t + 26.6) A
which agrees with our previous calculation. Moreover, since the
imaginary part of 15e j2t is 15 sin 2t, had we instead applied the
sinusoidal source Vg = 15 sin 2t, the response would have been
(e)
Im[3-./5e j (2t+26.6 0
)] = 3-./5 sin(2t + 26.6) A
FIGURE 8.7 (a) Response to f,; (b) response to 6; (c) response to the
linear combination f1 + jf2. As another example, let us find the forced response if of
il + ji 2 , as shown in Fig. S.7(c). Here we have used the fact that proportionality and d 2i di
superposition work equally well for complex quantities as for real ones. -2 + 2- + Si = 12.J2cos(2t + 15)
dt dt
Figure S.7 reveals an important property of linear systems. Vg = fl + jh is an First we replace the real excitation by the complex excitation,
arbitrary complex source, and i = i 1 + j i2 its forced response. Since i 1 is also the
separate response to fI, which is the real part of the excitation v g , the response to the VI = 12.J2e j (2t+I5)
real part of a complex source is the real part of the response. Similarly, noting that i2 where for convenience the phase is written in degrees. The complex
is the response to 12, the imaginary part of vg , the response to the imaginary part of a response i 1 satisfies
complex source is the imaginary part of the response.
This association is particularly significant when the complex source takes the form d 2i I + 2dil + Si = 12.J2 e j (2t+15)
of a complex exponential Vg = Vmej(wt+</J). By Table 7.1 we know that the forced dt2 dt 1
response to this source will be a complex exponential of the same frequency w, which we and it must have the general form
denote as i = Imej(wt+O). Once the response to the complex exponential source Vmej(wt+) il = Ae j2t
has been found, we may immediately write down the response to the sinusoidal sources
Vm cos(wt + ) as its real part and the response to Vm sin(wt + ) as its imaginary part. Therefore, we must have
(-4 + j4 + S)Ae j2t = 12.J2ej2tejJ50
Let us revisit the RLC circuit that was the subject of Section S.2.
Replacing the sinusoidal source 15 cos 2t V by the complex source or A =
12.J2 e jl5 12.J2 /15
= 3 /-30
15e j2t V, (S.l1a) is 4 + j4 4.J2/45
d'
2--':
dt
+ 2i + 10 1t
0+
i(r)dr + vc(O+) = l5e j2t
which gives
il = (3/-30 )e j2t
Differentiating and dividing by 2 as in (S.ll b) gives The response to the original excitation is the real part of the response
d 2i di . to this complex excitation:
- 2 +- +5i = l5je J2t
dt dt if = Re (il) = 3 cos(2t - 30)
326 Chapter 8 Sinusoidal Sources and Phasors Section 8.3 Complex Sources 327
In summary, given a circuit excited by a sinusoid, the forced response may be found We are concerned with the forced response of a circuit to sinusoidal excitation at
by replacing the sinusoid by a complex exponential whose real (or imaginary) part is the frequency w. Each sinusoidal source may be expressed as a cosine
given sinusoid. The describing equation will be easier to solve, since the trial forced
solution will only have one undermined constant, rather than two, requiring the solution Vg(t) = Vm cos(wt + )
of simultaneous equations for the coefficients. Once solved, we may easily relate the Suppose that we replace each such source by a complex exponential source given by
complex excitation response to our desired sinusoidal response by taking the real part of
VgI (t) = Vmej(wt Hl
the response if the original sinusoidal source is the real part of the complex exponential
source, or the imaginary part if it is the imaginary part of the complex source. Comparing these two expressions, we see that the complex source has the same frequency
w, and that the original source is the real part of the complex source we have chosen to
replace it.
Consider the forced response in the new circuit. Since it is excited by complex
EXERCISES exponential sources of frequency w, all currents and voltages will also be complex expo-
\ nentials of frequency w. This follows from the form of the trial forced solution Ae jwt .
8.3.1. (a) From the time-domain equations, find the forced response v if Then each current will be of the form
Vg = lO(e jSt ) V. (b) Using the result in part (a), find the forced response v
if Vg = lOcos 8t V.
Answer (a) 2e j (St-53.IO) V; (b) 2cos(8t - 53.1) V
(S.20a)
1Of.!
+
5f.! v
(S.20b)
where I and V are complex numbers. We define I and Vas phasors, that is, the complex
EXERCISE 8.3.1 numbers that multiply e jwt in the expressions for currents and voltages. To distinguish
them from other quantities, phasors are printed in boldface. The units for phasors are
I 8.3.2. Find the forced response v in Exercise 8.3.1 if Vg = 10 sin 8t V. the same as the currents and voltages they are associated with; thus in (S.20) I inherits
16 F [Suggestion: sin 8t = Im(e jSt ).] the units of iI, usually amperes, and V has the same units as the voltage VI, usually
Answer 2sin(8t - 53.1) V volts.
8.3.3. Using the method of complex excitation, find the forced response Phasors are thus defined in terms of the response to complex excitations, but the
i if Vg = 20 cos 2t V. importance of phasors lies in their direct link to sinusoidal responses. In the sinusoidal
EXERCISE 8.3.3 Answer 2 cos(2t + 36.9) A circuit with which we began this discussion, all forced responses are sinusoids at fre-
8.3.4. Repeat Exercise 8.3.3 if Vg = 16cos4t V. quency w. Let one such response be vet). After substituting complex exponential sources,
Answer 2cos4t A the same response variable will be VI (t) = Ve jwt , where V = IVILft is its voltage phasor.
But we recall from the last section that the response to the real part of a source is just
the real part of the response. Since the sinusoidal source Vm cos(wt + ) is the real part
of the complex source
(S.21)
it follows that the sinusoidal response vet) is just the real part of the complex response
The results obtained in the preceding section may be put in much more compact form VI (t), or
by the use of quantities called phasors, which we introduce in this section. The phasor vet) = Re(Ve jwt )
method of analyzing circuits is credited generally to Charles Proteus Steinmetz (1S65-
1923), a famous electrical engineer with the General Electric Company in the early part Using the polar exponential form for V and taking the real part yields
of the twentieth century. vet) = Re(IVle jO e jwt ) = IVI cos(wt + e) (S.22)
328 Chapter 8 Sinusoidal Sources and Phasors Section 8.4 Phasors 329
Note the direct relationship between the the sinusoid vet) and its phasor V. By To find the forced response i(t) in Fig. 8.9(a), we replace the source
(8.22), IVI is the amplitude of vet) and () its phase angle. The amplitude of the sinusoid 36cos(2t + 30) V by the complex source 36e j (2t+300) V and the
is the magnitude of its phasor, and the phase angle of the sinusoid is the angle of its source 2 sin(2t - 15) A, following Table 8.1, by 2e j (2t-105) A.
phasor. Thus we can immediately write down the sinusoidal current or voltage once its Analyzing Fig. 8.9(b), the single mesh equation is
phasor has been computed.
4il + 3~ [il - 2e j (2t-105)] = 36e j (2t+300)
dt
Suppose in Fig. 8.8 that Vg = 6 cos 2t V. Since Vg = Re(6e j2t ),
we use Vgl = 6e j2t V as our complex exponential source. The or dil + :l:il = 12e j (2t+300) _ j4e j (2t-105)
describing equation is dt 3
Starting with the cosine form Vm cos(wt + ) for each source, we conclude that
each response is also a cosine (8.22) of the same frequency, whose amplitude and phase 36 cos(2t + 30) V 3H 2 sin(2t - 15) A
are just the magnitude and angle of the associated phasor. To preserve this association,
if the sinusoidal source is given in the sine form Vm sin(wt + ), we first convert to the
cosine form using (8.7b), repeated here as
(a)
cos(wt + - 90) = sin(wt + )
Table 8.1 shows the relation between these two sources and their phasors.
A cos(wt + 4 Aej(wtH)
(b)
A sin(wt + 4 Ae j (wtH-900)
330 Chapter 8 Sinusoidal Sources and Phasors Section 8.4 Phasors 331
or i(t) = 4.0cos(2t - 9.2) A
One point of considerable practical significance is illustrated by Example 8.7. Our In this section we show that relationships between phasor voltage and phasor cur-
work with phasors will save us considerable time and effort compared to using real rent for resistors, inductors, and capacitors are very similar to Ohm's law for resis-
sinusoids, but only if we are equipped with the right tools. To use the phasor method, tors. In fact, the phasor voltage is proportional to the phasor current, as in Ohm's
we need to be able to perform calculations such as (8.23), law. Consider a circuit in which all currents and voltages are of the form Ae jwt .
This will be the case when sinusoidal sources have been replaced by complex ex-
9.6~ = 4.0/ -9.20 ponentials and we are interested in the forced response only. For the resistor of
4/3 + j2 Fig. 8.10,
rapidly and efficiently. This requires adequate computational support for the arithmetic v = Ve jwt (S.24a)
operations required and repeated so frequently in the course of phasor circuit calcula-
i = Ie jwt (S.24b)
tions: addition, subtraction, multiplication, and division of complex numbers, and also
rectangular-to-polar and polar-to-rectangular conversion. It is highly desirable to have where V and I are phasors. By Ohm's law applied to (8.24),
access to an electronic calculator that supports complex data types, four-function com-
plex arithmetic, and single-keystroke rectangular-to-polar and polar-to-rectangular con- (S.25)
versions. Although any calculator having sine, cosine, arctangent, and square root keys
will permit us to do all required calculations, those explicitly set up to support com- or, canceling the e jwt factors,
plex arithmetic have a decided edge in convenience. Many fewer keystrokes will be
required to complete a given problem, and keystrokes translate into time and diversion
of mental focus from the concepts in the circuit problem at hand. Each major calcula-
tor manufacturer offers a line of suitable machines, beginning at relatively inexpensive
prices.
Finally, in our work with phasors we rely on the principle that the response to the Thus the phasor or frequency-domain relation for the resistor is exactly like the time-
real part of a complex source (i.e., the response to a cosine source) can be computed domain relation. The voltage-current relations for the resistor are illustrated in Fig. 8.11.
as the real part of the response (i.e., the cosine part of a complex exponential). It is With V = VmeN>v and I = ImeN>l, (8.26) becomes
also possible to use another result from Section 8.3, that the response to the imaginary
part of a complex source is the imaginary part of the response, to develop a second VmeN>v = (RIm)eN>1
type of phasor analysis based on sines and imaginary parts rather than cosines and real
that is, the magnitude of the voltage phasor equals the magnitude of the current pha-
parts. Since either type of phasor analysis is sufficient, the alternative form of phasor
sor scaled by R, and the angles are the same. Recalling that the magnitude of a
analysis is not considered in this book, except in one of the problems at the end of the
phasor is the amplitude of its sinusoid, and its angle is the phase angle of the sinu-
chapter. We stick with phasors based on cosines and real parts, as in (8.22) and in the
soid, we have Fig. 8.11. Note that since <Pv = <P[' the current and voltage are in
examples. phase.
I
~
+ +
EXERCISES
v=Ri V=RI
8.4.1. Find the phasor representation of (a) 6 cos(2t + 45), (b) 4 cos 2t+
3 sin2t, and (c) -6 sin(5t - 65).
Answer (a) 6/45; (b) 5/-36.9; (c) 6/25
8.4.2. Find the time-domain function represented by the phasors
(a) 10/-17, (b) 6 + j8, and (c) - j6. In all cases, 0) = 3. (a) (b)
Answer (a) lOcos(3t - 17); (b) 10cos(3t + 53. F);
(c) 6 cos(3t - 90) FIGURE 8.10 Voltage-current relations for a resistor R in the (a) time and
(b) frequency domains.
332 Chapter 8 Sinusoidal Sources and Phasors Section 8.5 Current-Voltage Laws for Phasors 333
v,i i
-?
h<
+ +
(a) (b)
Thus the phasor voltage V, as in Ohm's law, is proportional to the phasor current I, with
fiGURE 8.11 Voltage and current waveforms for a resistor.
the proportionality factor jwL. The voltage-current relations for the inductor are shown
in Fig. 8.12.
As an illustration, suppose that the voltage
If the current in the inductor is given by i = 1m cos(wt + (PI), then by (8.27) the
v = 10cos(100t + 30) V phasor voltage is
is applied across a 5-Q resistor, with the polarity indicated in
Fig. 8.1O(a). Then the phasor voltage is V = (jwL)(Im/ (PI)
In the case of the inductor, substituting the complex current and voltage into the
v,i
time-domain relation,
di
v=L- v
dt
gives "_', /i
, ," '/:C'
,,
v:me j(wt+</>v) = L~ [I ej(Wt+</>I)] ,,
,, ,,
,,
dt m
= jwLlmej(Wt+</>I)
..
.
,,
Again, dividing out the factor e jwt and identifying the phasors, we obtain the phasor
relation
334 Chapter 8 Sinusoidal Sources and Phasors Section 8.5 Current-Voltage Laws for Phasors 335
Finally, let us consider the capacitor. Substituting the complex current and voltage v,i
i = C dv
dt
gives the complex relation
Again dividing by e jwt and identifying the phasors, we obtain the phasor relation
\
1= jwCV (8.28)
FIGURE IU 5 Voltage and current waveforms for a capacitor.
or
w?ich indicates that in the case of a capacitor the current and voltage are out of phase,
WIth the current leading the voltage by 90, This is shown graphically in Fig. 8.15.
I
V=- (8.29)
jwC
As an example, if v = IOcos(100t + 30) is applied across a l-/LF
capacitor, then by (8.28) the phasor current is
Thus the phasor voltage V is proportional to the phasor current I, with the proportionality
1= j (100)(10- 6 )(10/30) A
factor given by 1/ j wC. The voltage-current relations for a capacitor in the time and
frequency domains are shown in Fig. 8.14. = 1/120 rnA
In the general case, if the capacitor voltage is given by v = Vm cos(wt + v), then
The time-domain current is then
by (8.28) the phasor current is
i = cos(100t + 120) rnA
1= (jwC)(Vm/ v)
= wCVm/ v + 90 and therefore the current leads the voltage by 90, as it must in all
capacitors.
Therefore, in the time domain we have
v
!
for L = 15 mH, and (c) Fig. 8.14(a) for C = fJ,F.
v Answer (a) 3 cos(1000t + 30) rnA; (b) 0.8 cos(lOOOt - 60) A;
(c) 6cos(1000t + 120) rnA
8.5.2. In Exercise 8.5.1, find i in each case at t = 1 ms.
Answer (a) 0.142 rnA; (b) 0.799 A; (c) -5.993 rnA
(a) (b) 8.5.3. For what L in Fig. 8.12 will a sinusoidal voltage of amplitude
20 V produce a 4-rnA amplitude current at (j) = 500 rad/s?
FIGURE 8.14 Voltage-current relations for a capacitor in the (a) time and Answer 10 H
(b) frequency domains.
336 Chapter 8 Sinusoidal Sources and Phasors Section 8.5 Current-Voltage Laws for Phasors 337
It is important to stress that impedance is a complex number, being the ratio of two
complex numbers, but it is not a phasor. That is, it has no corresponding sinusoidal time-
domain function as current and voltage phasors do. Impedance is a complex constant
Let us now consider a general circuit with two accessible terminals, as shown in Fig. 8.16. that scales one phasor to produce another.
If the time-domain voltage and current at the terminals are given by The impedance Z is written in polar form in (8.33); in rectangular form it is
generally denoted by
v = Vm cos(wt + <Pv) (8.30a)
V = Vm / <Pv
(8.31) where R = Re Z is the resistive component, or simply resistance, and X = 1m Z is
1= 1m/J!.i \ the reactive component, or reactance. In general, Z = Z(jw) is a complex function of
We define the ratio of the phasor voltage to the phasor current as the impedance of jw, but R = R(w) and X = X(w) are real functions of w. Like Z, both R and X are
the circuit, which we denote by Z. That is, measured in ohms. Evidently, comparing (8.33) and (8.34), we may write
Z = 1O~ = 5/36.90 Q
<Pz = <Pv - <PI 2/20
In rectangular form this is
The magnitude of the impedance is the ratio of magnitudes of voltage and current phasors;
Z = 5(cos 36.9 + j sin 36.9)
the angle is the difference of the voltage and current phasor angles. = 4+ j3 Q
Impedance, as is seen from (8.32), plays the role, in a general circuit, played .by
resistance in resistive circuits. Indeed, (8.32) looks very much like Ohm's law; also lIke
resistance, impedance is measured in ohms, being a ratio of volts to amperes. The impedances of resistors, inductors, and capacitors are readily found from their
V-I relations of (8.26), (8.27), and (8.29). Distinguishing their impedances with sub-
scripts R, L, and C, respectively, we have, from these equations and (8.32),
+
v ZR =R
ZL = jwL = wL190 (8.35)
1 1 1
Zc = - = - j - = -1-90
jwC wC wC
FIGURE 8.16 General phasor circuit.
1
YL = - -
jwL
Yc = jWC
is called admittance and is analogous to conductance (the reciprocal of resistance) in which are the admittances of a resistor, with R = I/G, an inductor,
resistive circuits. Evidently, since Z is a complex number, then so is Y. The standard and a capacitor.
representation is
EXERCISES
Y = G+jB (8.39)
8.6.1. Find the impedance seen at the terminals of a series RL subcircuit
in both rectangular and polar form.
The quantities G = Re(Y) and B = Im(Y) are called conductance and susceptance, Answer R + jwL; vi R2 + w2L2/ tan- 1 wLjR
~---
340 Chapter 8 Sinusoidal Sources and Phasors Section 8.6 Impedance and Admittance 341
8.6.2. Find the admittance seen at the terminals of a series RL subcircuit As.an example, consider the circuit of Fig. 8.18, which consists of
in both rectangular and polar form. N Impedances connected in series. By KCL for phasors, the single
R wL phasor current I flows in each element. Therefore, the voltages
Answer R2 + w2L2 - j R2 + w2L2; shown across each element are
1
/ tan- 1 (wL/ R)
JR2 + w 2 L2
8.6.3. Find the conductance and susceptance if Z is (a) 3 + j4, (b)
0.4 + jO.3, and (c) (J2/2)/45 .
Answer (a) 0.12, -0.16; (b) 1.6, -1.2; (c) 1, -1
V N = ZNI
and by KVL around the circuit,
\
V=V 1 +V 2 +",+V N
= (ZI + Z2 + ... + ZN)I
If a complex excitation, say Vmej(wt+l:l), is applied to a circuit, then complex voltages,
such as Vl ej (wt+e1), V2ej(wt+e2), and so on, appear across the elements in the circuit. KVL Since we must also have, from Fig. 8.18,
applied around a typical loop results in an equation such as v= ZeqI
Vlej(wt+e1) + V2 ej (wt+e2) + ... + VNej(wt+e N ) = 0
where Zeq is the equivalent impedance seen at the terminals, it fol-
Dividing out the common factor e jwt , we have lows that
(8.40)
11 + 12 + ... + IN = 0
I
~
where +
n = 1,2, ... , N + v2 -
v
Thus KCL holds for phasors.
In circuits having sinusoidal excitations with a common frequency w, if we are in-
terested only in the forced, or ac steady-state, response, we may find the phasor voltages
or currents of every element and use Kirchhoff's laws to complete the analysis. The
analysis is therefore identical to the resistive circuit analysis of Chapters 2, 4, and 5, with
impedances replacing resistances and phasors replacing time-domain currents and volt-
ages. Once we have found the phasors, we can convert immediately to the time-domain
FIGURE 8.18 Impedances connected in series.
sinusoidal answers in the usual fashion.
I 1 as obtained earlier.
Zeq= - = - - - (8.42)
Yeq YI + Y2
This ability to treat inductors, capacitors, and resistors alike as elements of the
In like manner, voltage and current division rules hold for generic impedance without distinction among them when writing circuit equations is one
phasor circuits, with impedances and frequency-domain quantities, of the main strengths of the phasor method. RLC circuits may be simplified using series-
in exactly the same way that they held for resistive circuits, with parallel equivalents, current-voltage dividers, and Thevenin-Norton transformations just
resistances and time-domain quantities. The reader is asked to es- as pure resistive circuits were earlier. Series impedances add, currents through parallel
tablish these rules in Exercise 8.7.2. impedances divide in proportion to their admittances, and so on, regardless of the specific
identities (R, L, or C) of the impedances involved.
For example, let us return to the R L C circuit considered in Sec-
tion 8.2. The circuit and phasor circuit are shown in Fig. 8.19(a)
and (b). By KVL in the phasor circuit, We wish to find i (t) in steady state in Fig. 8.20. Using series-parallel
impedance equivalents, we have
ZcI + ZLI + RI = 15LQ
3(2 - j2) .
or (2 - j)I = 15LQ ZI = 3 + (2 _ j2) = 1.45 - JO.621 Q
from which the phasor current is ( 'I)(Z '1)
Z2 = -. J I +J = 0.583 - '0.75 Q
I = ~ = 3./5/26.6 A
-Jl+(ZI+jl) J
2-J
Therefore, in the time domain we have the same result as
before, although now computed with considerably less effort:
2H j4Q
15 cos 2tV
'J 2Q 15Loo V
'J 2Q
12~V 2Q
I -j5 Q
iO F
(a) (b) (b)
FIGURE 8.19 (a) Time-domain circuit; (b) phasor circuit. FIGURE 8.20 (a) Time-domain circuit; (b) phasor circuit.
344 Chapter 8 Sinusoidal Sources and Phasors Section 8.7 Kirchhoff's Laws and Impedance Equivalents 345
so the equivalent impedance is 8.7.4. Find the steady-state voltage v in Exercise 8.7.3 using phasors and
Zeq = ! + Z2 = 1.083 - jO.75 Q
voltage division.
Answer lOcos(8t - 126.9) V
and
1= 12/14 = 12~ = 9.11/48.7 A
Zeq 1.083 - jO.75
Recalling that w = 1 rad/s yields
i(t) = 9.11 cos(t + 48.7) A
This is the desired ac steady-state current.
As the discussion in Section 8.7 suggests, we may omit the steps of finding the describing
equation in the time domain, replacing the excitations and responses by their complex
EXERCISES forcing functions and then dividing the equation through by e jwt to obtain the phasor
equation. We may simply start with the phasor circuit, which we define as the time-
8.7.1. Derive (8.41). domain circuit with the voltages and currents replaced by their phasors and the elements
8.7.2. Show for circuit (a) that the voltage-division rule, identified by their impedances. The describing equation obtained from this circuit is then
the phasor equation. Solving this equation yields the phasor of the answer, which then
22
V= Vg may be converted to the time-domain answer.
21 +22 The procedure from starting with the phasor circuit to obtaining the phasor answer
and for circuit (b) that the current-division rule, is identical to that used earlier in resistive circuits. The only difference is that impedances
Y2 21 replace resistances.
1= - - - Ig = Ig
YI+Y2 21+22
are valid, where 21 = 1/YI and 22 = 1/Y2 As an example, let us find the steady-state current i in Fig. 8.21(a).
The phasor circuit, shown in Fig. 8.21(b), is obtained by replacing
I
---?>- the voltage source and the currents by their phasors and labeling the
+
In i
+ ---?>-
22 V
3n
5 cos 3tV !p
9
(a) (b) IH
EXERCISE 8.7.2
(a)
8.7.3. Find the steady-state current i using phasors.
Answer 2cos(8t - 36.9) A I
In ---?>-
H2 IH
3n
+ 5Lov -j3 n
1
10 cos 8tV 4<)P v j3 n
(b)
346 Chapter 8 Sinusoidal Sources and Phasors Section 8.8 Phasor Circuits 347
elements with their impedances. In the phasor circuit the impedance lp
8
seen from the source terminals is a
Z = 1 + (3 + j3)(-j3)
3 + j3 - j3 +
3cos4tA
= 4 - j3 Q
Therefore, we have
(a)
1 _ 5iJ1 _ 5iJ1 _
I - 4 _ j3 - 5/-36.9 - 1/36.9
-j2 Q
a
and, by current division,
1_ 3 + j3 _
- 3 +]'3 -]'3 11 - .J2/81.9 A
3.LOA
In the time domain, the answer is
i = .J2cos(3t + 81.9) A
(b)
In the case of a dependent source, such as a source kv x volts controlled by a FIGURE 8.22 (a) Circuit containing a dependent source; (b) corre-
voltage Vx. it will appear in the phasor circuit as a source kV x , where Vx is the phasor sponding phasor circuit.
representation of vx , because Vx = Vm cos(wt + ) in the time domain will become
In the case of an op amp, the phasor circuit is the same as the time-domain circuit.
Vm e j (<vt+4 when a complex excitation is applied. Then dividing ej<vt out of the equations
That is, an ideal op amp in the time-domain circuit appears as an ideal op amp in the
leaves Vx represented by its phasor Vme N . In the same way, kvx = kVrn cos(wt + ) is
represented by its phasor kVrne je , which is k times the phasor of vx . phasor circuit, because the time-domain equations
i = 0, v=o
As an example of a circuit containing a dependent source, let us which characterize the current into and the voltage across the input terminals retain the
consider Fig. 8.22(a), in which it is required to find the steady-state identical form,
value of i. The corresponding phasor circuit is shown in Fig. 8.22(b). 1=0, V=o
Since phasor circuits are analyzed exactly like resistive circuits, we
may apply KCL at node a in Fig. 8.22(b), resulting in in the phasor equations.
Phasor analysis is indeed a tool of great power and versatility. This may already
V I - !VI be apparent from our work in this chapter and will be further underscored by our reliance
1+ 2 = 3iJ1 (8.43) on phasor analysis to explore important issues such as ac steady-state power in the next
-j2
several chapters. Like any tool, however, its range of applicability is limited, and it is
By Ohm's law we have VI = 41, which substituted into (8.43) yields useful to remind ourselves of these limits.
The first caution is that phasors are useful for finding the forced response only. If
-j21+ !(41) = -j6 the natural response is desired, another tool must be used. Also, if the total response is
-j6 6~ 3 required, phasor analysis is useful in determining the forced component only. The second
or 1 = 2 _ j2 = 2./i/-45 0 = ./i/-45 A caution concerns the use of phasors to determine steady-state response. For many circuits
it is the case that natural responses all decay to zero as t ~ 00. In such circuits the forced
Therefore, we have response and the steady-state response are synonymous, and phasors (designed for deter-
mining the forced response) may be used to find the steady-state response. Such circuits
. 3 are called stable. For the remaining or unstable circuits, the natural response does not
I = ./i cos(4t - 45) A
in general decay to zero, and thus the steady-state response cannot be identified with the
348 Chapter 8 Sinusoidal Sources and Phasors Section 8.8 Phasor Circuits 349
10
forced response. Phasor analysis may be used to find the steady-state response only with
stable circuits. Examples of stable circuits include all circuits that are made up exclu-
sively of passive elements and independent sources and that happen to have no undamped
natural responses (no uncoupled LC subcircuits lacking resistance). Examples of unstable 4 cos 3tV
circuits are those with undamped natural response, and circuits possessing a natural re-
sponse that grows, rather than decays, with time. Growing natural responses are possible
only in linear circuits containing controlled sources. Stability is discussed in Chapter 14.
In general, unstable circuits do not possess steady-state responses at all (their natural (a)
responses either grow or oscillate periodically, so the circuit never goes to steady state no
matter how long we wait). Thus it is meaningless to try to define a steady-state response 10
for an unstable circuit, and for such circuits phasors may be used to compute the forced,
but not the steady-state, response.
\
Let us determine the describing equation for i in Fig. 8.23. The 4Lov
controlled source has transconductance g = ~ S. By KVL around
the left loop,
I '8 ( 23)
+ 13 I- V = 4LQ V 0.1 F
350 Chapter 8 Sinusoidal Sources and Phasors Section 8.8 Phasor Circuits 351
8.5. In the figure for Problem 8.4, is it true that the cosine
quadrature components of the four currents sum to zero?
The sine components? Justify.
The method of phasors permits efficient determination of the forced response in sinusoidal
circuits while bypassing the difficulties of solving coupled differential equations. For
o 8.6. Find the forced response i if L = 4 mR, R = 6 kQ, i
3F 2F iF
1m Vm = 5 V, and w = 2 X 106 rad/s. 2 cos ltV
stable circuits, the forced response is also the ac steady-state response. Ac steady-state
reveals the long-term behavior of circuits excited by sinusoids, and as the excitation
R
frequency varies, defines the circuit's frequency response.
'-----_--1\/1/1._..... b
(c) 5 - j 12, (d) 10, and (e) - j5.
Find the forced solution by replacing the sinusoidal forcing D
. .:.::. 8.24. Find (V so that i(t) lags vet) by 450. I o - - -_ _ _ _~
function by a complex exponential. Check by using the IillII --?
trial forced solution A cos 7t + B sin 7t. +
8.18. Find i4, using only the properties of sinusoids,
i(t) 3 kn V
if (a) i 1 = 6 cos 3t A, i2 = 4 cos(3t - 30) A, and
i3 = -4-vS cos(3t + 60) A, (b) il = 5 cos(3t + 30) A, --?o---~~r---' fiGURE 1'8.32
+
i2 = 5 sin 3t A, and i3 = 5 cos(3t + 150) A, and
Ql8.33. At what frequency (V is the susceptance B at termi-
(c) il = 25cos(3t - 53.1) A, i2 = 2sin3t A, and IillII nals ab equal to B
= 2 S?
i3 = 13 cos(3t - 22.6) A. (Hint: cos 22.6 = M.) vet) ImH
FIGURE 1'8.27
a---.--~\AP'r-_-'
8.20. d 2x
-dt 2
dx.
+ -dt - 6x = 4smt - 2cos(t - 27.1 )
Find the forced solutions by replacing the sinusoidal forcing
function by a complex exponential. Check by using the trial 20 cos wtV lp
8
forced solution A cos t + B sin t. FIGURE 1'8.25
FIGURE 1'8.30
o 8.21. Write the describing equation as a second-order dif-
IillII ferential equation. Rewrite in terms of the phasor I. Solve 8.31. Find the reactance X so that the impedance seen by
for I, and then for the forced response i(t). 8.26. For phasor current I = lLQ rnA and (V = 1000 rad/s, FIGURE 1'8.34
the source is real. For this case, find the steady-state current
sketch V R, V L and V c in the complex plane. i (t) corresponding to I if (V = 10 rad/s.
g 8.35. Convert all voltages to phasors. Use the fact that
Iill!l KVL holds for phasors to find V, then vet).
1 kn
I 6 cos(2t + 17)
--? - VR + +
12 cos 2000t V IH
48LQV 15 n -j20n
+ jlOn
0.Q1 H VL + +
vet)
2 cos(2t - 29)
500/LP
FIGURE 1'8.21 <E--
0
I fiGURE 1'8.31
- Vc + +
08.22. If vet) = 16 cos (4000t - 27) + 3 sin(4000t - 114) 18 sin(2t + 114)
IillII find the phasor associated with vet). 8.32. Find the impedance Z between a and b as a function
FIGURE 1'8.26 of (v.
FIGURE 1'8.35
FIGURE PB.47
12 cos 2000t V
fiGURE PB.37 +
FIGURE P8.41 g8.48.
8.38. Find the steady-state value of v. IillII
In
v 3P
100mH +
In IH IH 08.45.
....
I:::
FIGURE PB.3B
fiGURE P8.4B
+
8.39. Find the steady-state values of v and VI. 8 cos tV v +
12 n 2P 08.49. (a) Given a source A sin(wt + <p) in an ac steady-
2sin4tA v
IillII state problem, suppose we use Allas the correspond-
In
ing phasor source (rather than AI
<p - 90 0 as prescribed in
Table 8.1). It is no longer true that the ac steady-state re-
+ fiGURE PB.42 sponse will be the real part of the complex response. How
5 cos t A v lp lp 3n 3H must this statement be modified, and why?
3 3
FIGURE PB.45 (b) Use this idea, sometimes called the "sine phasor" ap-
proach, to solve Problem 8.48 without using the 90 0 phase
8.43. Find the steady-state value of i when (a) w = shift in the formula for the source phasor.
1 rad/s, (b) w = 2 rad/s, and (c) w = 4 rad/s. [Note 8.50. Find the complete response i if i (0) = 2 A and
fiGURE PB.39 that (b) is the resonant case.] 08.46.
.... v(O) = 6 V. (Suggestion: Use phasors to get if and the
.... differential equation to get in.)
8.40. Find the steady-state values of i and v.
20 n IOn
+ +
4 cos 2500tV v 20 cos wtV lp 4 cos (t - 17) mV 0.1 P 20 cos 3tV ~p v
4 3
+
1kn 1 kn
v 2H W 2 scott
7n
FIGURE PH.51
Thomas A. Edison
also designed the first electric power station. Hisdiscovery.of the Edison effect,
the movement of electrons in the v-aGuUl:n.()fNs:lig~t. by:lb, also .marked the
beginning of the age of electronics.
I
----7
+ + +
Vj
V
+
In Chapter 8 we saw that the steady-state response of circuits excited by sinusoidal v2
V
sources (ac circuits) could be efficiently computed by converting the time-domain cir-
cuit to its corresponding phasor circuit. The circuits studied were relatively simple and
were analyzed directly from Kirchhoff's laws and use of the basic notion of impedance.
In the present chapter we move beyond these relatively simple circuits. We shall see,
in fact, that all the tools of circuit analysis introduced thus far apply equally well to
phasor circuits as to their time-domain counterparts. First we consider circuit simpli- (a) (b)
In Chapter 2 several useful methods for simplifying resistive circuits were introduced.
These include series-parallel equivalents for resistors and for sources, current-voltage (9.2)
division, and Thevenin-Norton equivalents. Each of these methods was derived from the
same foundation: Ohm's law describing the I-V behavior of individual elements and
Kirchhoff's current and voltage laws governing their interconnection. Comparing the voltage drops across two impedances gives Zi and Zj gives
In ac steady state we may replace the original circuit by its phasor circuit coun-
terpart, in which the sinusoidal sources are represented by their corresponding phasor (9.3)
sources and RLC elements by their impedances Z. In the phasor circuit the same
= (0.743/68.2 )(0.429/83.8)
(9.4) = 0.319/152 A
Series and parallel equivalents may be found for sources as well as impedances in
and comparing two of these parallel currents, phasor circuits. Application of KVL in Fig. 9.3(a) shows that voltage sources in series
\ are equivalent to a single source whose source function is the sum of the individual
(9.5) source functions, or series voltage sources add. Using KCL in Fig. 9.3(b), we have the
corresponding result: parallel current sources add.
Thevenin and Norton equivalents in phasor circuits are found exactly as described
The current through parallel impedances divides in direct proportion to their admit- in Chapter 2 for resistive circuits, with only the substitution of impedance Z for resistance
tances. R and subsequent use of complex arithmetic. Following the development of Section 2.6
with only this change, we have, from (2.20), that the Thevenin and Norton forms shown
in Fig. 9.4 are equivalent if the relations
We seek VIand 12 in the phasor circuit of Fig. 9.2. The impedances
for each element are shown in the circuit diagram. To get Vb we
use voltage division. The admittance of the parallel elements is
(a) ZT = ZN (9.6a)
1 1
Y =-1 +15+4=4+ 110 1 1 1 S
10 (b) V T = ZNIN (9.6b)
1 hold between the circuits. To find the Thevenin or Norton equivalent of any two-terminal
Z = Y= 3.45 - j1.38 n subcircuit of a phasor circuit, we follow the prescription of (2.24), again only replacing R
by Z. The two-terminal circuit A with open-circuit phasor voltage Voe and short-circuit
Then, by (9.2), phasor current Ise shown in Fig. 9.5 is equivalent to the Thevenin and Norton forms
_ 2 (3/450)
VI - 2 - j3 + (3.45 - j1.38)
= (0.286/38.8)(3/45)
= 0.858/83.8 V
4n
(a)
fiGURE 9.3 (a) Series voltage sources and equivalent; (b) parallel current
fiGURE 9.2 Circuit for Example 9.1. sources and equivalent.
a'
(a) (b)
shown in Fig. 9.4 if FIGURE 9.6 (a) Circuit for Example 9.2; (b) after Thevenin equivalent
of circuit to left of terminals 1-2.
ZL=jwL=j~Q
ZT and ZN can also be found as the impedance looking into terminals a-a' with all The common voltage across these inductors is, by voltage division,
independent sources in the phasor circuit A killed. '4/3
Thevenin and Norton equivalents are used in phasor circuits to reduce complicated VL = ] = 20/-163 V
(j~ - j2)(10/17)
multielements circuits to simple two-element circuits, just as they were used in the resis-
tive circuits studied previously. These equivalent circuits are even more generally useful The current through the !-H inductor is the short-circuit current
in phasor circuits, since the restriction that the passive elements all be of the same type
I _ VL 20L-=.!.QJ
does not apply. Any mixture of RLC elements, that is, impedances, will do. se - j(4)(!) (2/90) = 10/107 A
Thus ZT in Fig. 9.6(b) is found to be
ZT = Voe = 20~
10/1070 = 2/-90 Q
Ise
Having reduced the problem to one with a simple circuit, we tum to
Fig. 9.6(b). The current through the resistor is
(a) (b)
I _ 20/17
FIGURE 9.5 Voc and Isc for computing Thevenin and Norton equivalents R - R - j2
of the circuit A. The magnitude of the phasor IR is given by the ratio of magnitudes
of its numerator and denominator,
We wish to determine the value of R in Fig. 9.6(a) that will cause a 20
sinusoidal current of amplitude 1 A to flow through this resistor. Our IIRI= ~
strategy will be to reduce this to the simple one-loop circuit shown in vR-+4
Fig. 9.6(b) by finding the Thevenin equivalent of everything except If IIRI = 1 A, the amplitude of the corresponding sinusoidal current
the resistor. The open-circuit voltage Voe = V 12 with R removed is i R will be 1 A as well. Thus we require that
found by voltage division to be 20
--===
,JR2 +4
= 1
"4(10/17 )
V
oe
= ] j4 - j2
= 20/17 V
or R = 6,JIT Q.
EXERCISE 9.1.2
3Q (2 + j2)Vl - jIV2 = 10
-jiVI + (1- jI)V2 = 5
10 - j1 1
1 5 1 - jl 10 - j5
VI = -,--!---~--'--~ 5 =2-jl V
EXERCISE 9.1.3 2 + j2 - jl 1
1
-jl I-jl
9.1.4. Find the Thevenin equivalent of the circuit of Exercise 9.1.3.
Answer ZT = 3 + j4 Q; VT = 10/63.1 V 12 + j2 I~ 1
V2 = -ji = 1O+j20 =2+j4V
5 5
3000i 2kQ
4 cos 5000t
~"FI
- 2kQ
5LOV 5LOA V
1
JS/-LF
(a)
I-
fIGURE 9.9 Circuit containing a dependent source.
dashed, we have
V- 4 V V + 30001
-jl n -- + + -:-:---- 0
!(10 )
3 ~(1 - j2)(10 3 ) (2 - jl)(10
3 ) =
5LOV
1 +j2 n 5LOA
5 Note that the" gain of the dependent source is 3000 VIA. Also ' f rom
the phasor CIrcUIt we have
4-V
(b) 1=--
!(103)
fiGURE 9.8 Two versions of the phasor circuit corresponding to
Eliminating V between these two equations and solving for I we
Fig. 9.7. have '
In polar form these quantities are 1= 24 x 10-3/53.1 A 0
e= _ tan-1 y'2ev/lOOO
(9.8)
1 - (ev/1000)2
>--.---0 v
In the time domain we have
2kQ 2Vm
v = cos(evt + e) (9.9)
2kQ
.}I + (ev/1000)4
We might note in this example that for low frequencies, say
o< ev < 1000, the amplitude of the output voltage v is relatively
large, and for higher frequencies, its amplitude is relatively small.
FIGURE 9.11 Circuit containing an op amp. Thus the circuit of Fig. 9.11 filters out higher frequencies and allows
lower frequencies to "pass." Such a circuit is called a filter and is
considered in more detail in Chapter 14.
Therefore, v = 2V2, or v2 = v/2, as indicated by the phasor VI2 in
the phasor circuit of Fig. 9.12.
Writing nodal equations at the nodes labeled V 1 and V /2, we
have EXERCISES
IOn
Eliminating VI, by solving the last for V 1 and substituting into the
previous equation, and solving for V results in
10 cos 3tV
2Vm
V---------=---
- [1 - (ev 2/10 6 )] + j(y'2ev/103 )
In polar form this is EXERCISE 9.2.1
2VmL.!l
(9.7) 9.2.2. Find the steady-state value of v using nodal analysis.
V = -';-;=1=+=(ev=/=10'===0:: : :0): : : 4 Answer 2S..j2cos(2t - 81.9) V
-j1000/w kQ
Vl2 +
1
IOQ v
>--+--0 V 30 F
2kQ
-jl000/w kQ
The same shortcut procedures for writing nodal and mesh equations discussed in
Sections 4.3 and 4.5 for resistive circuits apply to phasor circuits. For example, in
Fig. 9.13, if 13 = -5 is the mesh current in the right mesh in the clockwise direction,
As suggested in Section 9.2, the general methods of circuit analysis apply to phasor the two mesh equations are written by inspection as
circuits as if they were resistive circuits, with resistance R replaced by the ~ore general
impedance Z. In this section we illustrate ~e application of ~esh anal.ysls. to phasor (~ - jl) 11 - (- jl)lz = 5
circuits. Once again we note that the generalIzed form of Ohm slaw usmg lmp~dance
V = ZI permits replacement of calculus (differentiation and integration) by the sImpler
operations of algebra (multiplication and division). (
-(-jl)ll+ - j l - j l + -+I -1 '2) z - (1-I-
I + '2) 13=0
These are equivalent to (9.11) and are formed exactly as in the resistive circuit case. That
To illustrate mesh analysis of an ac steady-state circuit, let us find VI
is, in the first equation the coefficient of the first variable is the sum of the impedances
in Fig. 9.7, which was obtained, using nodal analysis, in Section 9.2.
around the first mesh. The other coefficients are the negatives of the impedances common
We shall use the phasor circuit of Fig. 9.8(b), which is redrawn in
to the first mesh and the meshes whose numbers correspond to the currents. The right
Fig. 9.13, with mesh currents II and 12, as indicated. Once the circuit
member is the sum of the voltage sources in the mesh with polarities consistent with
is broken, the phasor voltage V I may be obtained as
the direction of the mesh current. Replacing "first" by "second" applies to the next
II equation, and so on. The dual development, as described in Section 4.3, holds for nodal
VI=5-- (9.10) equations.
2
lQ
2
This example illustrates that series-parallel impedance conversions,
VI
together with Thevenin-Norton transformations, may be of great
-jl Q benefit in simplifying a circuit before writing the general analysis
equations (nodal or mesh). We shall apply mesh analysis to the
~ ~ :3
-jl Q 1 +j2 Q
5LOV
5 circuit of Fig. 9.14(a), in which the desired response is the voltage
V across the 2-H inductor. Since this circuit contains six meshes
and one current source, simultaneous solution of five equations in
five unknowns would be required if the mesh equations were written
FIGURE 9.13 Circuit of Fig. 9.8 redrawn for mesh analysis. directly. By series-parallel conversions, the circuit may be redrawn
fiGURE 9.14 (a) Circuit for Example 9.7; (b) corresponding phasor Thus
circuit.
We have also converted the series combination of Z3 and the Having broken the circuit by finding the mesh currents, the desired
4/-43 V voltage source in Fig. 9.14(b) to the Norton form shown unknowns are next found in terms of the mesh currents. The voltage
across ZI in Fig. 9.15(b) is
V Z1 = ZI(1 1 - 12) = (0.128 + jO.79)(3.58/7.2 - 0.657/88.3)
= 2.83/77.4 V
12L20 V
From Fig. 9.14 we see that this voltage is across an impedance of
j2 n (the desired voltage, V) in series with another impedance of
1 n. By voltage division,
(a)
'2
V = 1 ~ j22.83/77.4 = 2.53/104 V
As a final example, let us consider the circuit of Fig. 9.16(a), where 9.3.1. Find the forced response i in Fig. 9.9 using mesh analysis.
the response is the steady-state value of VI. The controlled source 9.3.2. Solve Exercise 9.2.4 using mesh analysis.
has transconductance g = 2 S. The phasor circuit is shown in 9.3.3. Find the steady-state current i using mesh analysis.
Fig. 9.16(b), with the mesh currents as indicated. Answer .J2cos(2t - 45) A
Applying KVL around the supermesh labeled I, we have
6Q 2H 2Q
-VI - jl(-jl +1) + (1 + j2)(1+2V I) = 0
VI = cos(2t + 143.1) V
sin 2t A
r-----{~ ) - - - - - ,
An ac circuit is, we recall, a circuit whose independent sources are all sinusoids. If all
IQ sources in an ac circuit are of the same frequency w, the corresponding phasor circuit
+
may be used to determine the forced response in the manner described in Sections 9.1
to 9.3. It may be computed by a single unified nodal or mesh analysis or by use of
VI IH
2
IH superposition (summing the component responses to each individual source or group of
sources with all other independent sources killed).
Superposition is a general principle that may always be used to find the response
(a)
of a linear circuit containing more than one source. Even if an ac circuit contains sources
with different frequencies, superposition may still be used to find the forced response.
IL-90= -jl A For purposes of superposition, the sources are grouped so that each component problem
contains only sources of a single frequency. Then, for each resulting component problem,
a phasor circuit may be used to determine the phasor response, which is then converted
E to a sinusoidal response and added with the other component responses as the principle
of superposition requires.
I
2\) 1+ j2 Q
phasor circuits are shown in Fig. 9.17(b) and (c). Note that the
impedances in the two component problems are different for all
elements but the resistors, since for the other (storage) elements
impedance is frequency dependent. By voltage division,
(b)
376 Chapter 9 AC Steady-State Analysis Section 9.4 Sources with Different Frequencies 377
-v+ 20
Note in Example 9.9 that VI and V2 were converted to sinusoids before being added
together. Phasors corresponding to different frequencies cannot be superposed; only their
corresponding sinusoids can be superposed. Recall that a given phasor corresponds
12 cos 2t V IH
to amplitude and phase information on a sinusoid of a specific frequency w. If the
frequencies of two phasors differ, it makes no sense to add them together. The magnitude
of their phasor sum does not correspond to any sinusoidal amplitude, nor does the angle
(a) of their sum to any sinusoidal phase angle. When frequencies differ, the principle of
superposition applies to the summing of time-domain components, not phasors. Within
a component problem corresponding to a single frequency, however, phasors may also
be superposed. This is illustrated in the next example.
lUOoV j20 jl0 \ We seek the current i through the voltage source in the ac circuit
shown in Fig. 9.1S(a). We will use superposition, grouping the
two sources at w = 10 rad/s together and calling this component
(b) (c) of the response i I; the remaining component due to the source at
w = 5 rad/s is i 2 . Figure 9.1S(b) shows the phasor circuit for
FIGURE 9.17 (a) Ac circuit; (b) phasor circuit for the w = 2 compo- computing i l and Fig. 9.1S(c) that for i 2 . In each case the other
nent; (c) phasor circuit for the w = 1 component. sources have been killed, and the value of w corresponding to the
active sources used to compute the impedances. Using superposition,
where Za is the equivalent of the rightmost three impedances in each component problem may be set up and solved independently.
Fig.9.17(b): The right mesh equation in Fig. 9.1S(b) is
Za = (1)(2 + j2) = 0.769 + J0.154 Q 11(- j2 + 1 + j4) + j2(11 - 4LQ) = 5/30 A
3 + j2
Using this in (9.12), VI = 1O.5/13So V. Thus or 1 = 5m + j8 = 2.76/-S.4 A
I 1 + j4
VI = 10.5 cos(2t + 13S0) V
Turning to Fig. 9.17 (c), again by voltage division,
0.05 F 10 O.4H
V2 = ~ (5/4SO)
Zb +1
where Zb is the parallel equivalent of the impedances - j2 and
~ 5 cos (lOt + 30) V
2 + jl, or
Z - (-j2)(2+jl) -16- 12 Q
b - 2 _ jl -. J.
(a)
which, upon substitution, yields
-j40 10
l.6-jl.2
V2 = . (5/45) = 3.5/32.9 V
2.6 - Jl.2
The corresponding sinusoidal component is
V2 = 3.5 cos(t + 32.9) V
and by superposition, the forced response V is the sum of its com- (b) (c)
ponents:
FIGURE 9.18 (a) Ac circuit; (b) phasor circuit for the w = 10 compo-
V = VI + V2 = 10.5 cos(2t + 13S0) + 3.5 cos(t + 32.9) V nent; (c) phasor circuit for the w = 5 component.
i = il + i2 = 2.76cos(10t - 8.4) + 4.47 cos(5t + 163) A 9.4.2. For the phasor circuit corresponding to Exercise 9.3.3, replace the
part to the left of terminals a-b by its Thevenin equivalent and find the
steady-state current i 1.
In Example 9.10 we computed one component for each source frequency and Answer Voc = ~(2 - jl) V, Zth = !(18 + jl) n, il = cos2t A
summed these components. Note that since there are two sources in the phasor cir- 9.4.3. Find VI, 11, and 12. Use the ladder method, assuming V = lLQ.
cuit Fig. 9.18(b) corresponding to the frequency W = 10 rad/s, we may choose to solve Answer 3 V; 3 - j3 A; 3 A
this component problem by once again invoking the principle of superposition. We could
compute the subcomponents of 1\ due to each of the two sources with the other killed 1Q -4 ji Q
L -__ ~ __________ ~
and then sum these two phasors to get 11. Superposition may be freely applied within a
single-phasor circuit, that is, one in which there is a single frequency w. +
In this section we have seen that superposition may be used to decompose ac steady- Vg = 6LOV _ -jl Q -ji Q IQ V
state problems involving independent sources at more than one frequency into component
problems each containing sources at the same frequency. The component problems may
then be solved with the help of phasors. In other instances thus far where superposition
was discussed, it was also possible to choose to bypass superposition, solving in a single EXERCISE 9.4.3
unified manner with all sources in place. This same choice is also available for ac circuits
with multiple frequencies; but if we are to use phasors, we cannot choose to solve the
problem all at once. The difficulty is that a single unified phasor circuit involving sources
of different frequencies cannot be defined.
Phasors are defined as those complex numbers I and V used to specify currents
and voltages when they are of the specific form i = Ie jwt and v = Ve jwt . In circuits
with distinct frequencies WI and W2, the currents and voltages will not be of the required Since phasors are complex numbers, they may be represented by vectors in a plane,
form. They will in fact be the sums of complex exponentials at each of the different where operations such as addition of phasors may be carried out geometrically. Such
frequencies. Thus, in ac circuits containing sources with different frequencies, we cannot a sketch is called a phasor diagram and may be helpful in analyzing ac steady-state
hope to define a unified overall phasor associated with a given current or voltage. If we circuits.
were to try to do so by coverting the original circuit to a phasor circuit containing sources
at both WI and W2, the dilemma would be apparent when we tried to assign values to To illustrate, let us consider the phasor circuit of Fig. 9.19, for which
the impedances. What would we use as the value of W in computing impedance values we shall draw all the voltages and currents on a phasor diagram. To
for those elements whose impedance depends on w? Clearly, neither W\ nor W2 by itself begin, let us observe that the current I is common to all elements
would do. A single phasor circuit with sources at different frequencies is meaningless. and take it as our reference phasor, denoting it by
For sources at different frequencies, superposition is not just a choice; it must be used to
determine the ac steady-state response. 1= IIILQ
Vc -1 1
13 = - = - = -/!N,
- j2 j2 2
______________ Vg
VL
e vR 1
I + V~L2~~~~+-_ _~L--Re
VR I e 1Q Vc
-j2Q Vc
2
-------------- 'Vg
Vc (a) (b)
2 Vmy
principle, scaling the source will scale all phasors by the same factor. x +y 2 = - -
wL
Choosing the scale factor at to match the actual source value 2/30 ,
This result may be rewritten as
2~
at = 1.12/1530 = 1.79/ -123
2~LY = (2~LY
0
we see that all the phasors in the phasor diagram must be enlarged by
x2 + (Y + (9.18)
a factor of 1.79 and rotated 123 clockwise to complete the solution. which is the equation of a circle with center at [0, -(Vm/2wL)] and
In particular, radius Vm /2w L.
The circle (9.18) appears to be the locus, as R varies, of the
Vc = at(1LQ) = 1.79/-123 V
R--+oo phasor I = x + jy. However, by (9.16), x ~ 0; thus the locus
Note that rotating all the phasors in the vector diagram the same is actually the semicircle shown dashed on the phasor diagram of
amount does not change their phase relationships: 12 still leads 13 by o Fig. 9.24. The voltage VmLQ, taken as reference, is also shown,
90 and is still 180 out of phase with V c, and so on. The arbitrarily along with the phasor I. If R = 0, we have, from (9.16) and (9.17),
selected angle of the reference phasor is adequate to determine all x = 0 and y = - Vm/wL. If R -+ 00, then x -+ 0 and y -+ O. Thus
phase relationships without the need for corrections. Corrections are ,, as R varies from 0 to 00, the current phasor moves counterclockwise
needed only to determine the phase angle of a response rather than ,, along the circle.
,,
phase shift between two response variables. If I is as shown in Fig. 9.24, the current phasor may be resolved
ea into two components, one having amplitude 1m cos (j in phase with
, the voltage and one with amplitude 1m sin (j, which is 90 0 out of
As a final example illustrating the use of phasor diagrams, let us find ,,
phase with the voltage. This construction is indicated by the dashed
the locus of I as R varies in Fig. 9.23. The current is given by
vertical line. As we shall see in Chapter 10, the in-phase component
1= Vm Vm(R - jwL) of the current is important in calculating the average power delivered
R + jwL R2 + w 2 L2 --- by the source. Thus the phasor diagram gives us a method of seeing
at a glance the maximum in-phase component of current. Evidently,
Therefore, if R=O this occurs at point a, which corresponds to (j = 45. This is the
(9.15) fiGURE 9.24 Locus of the phasor I. case x = -y, or R = wL.
9.5.1. Eliminate wL in (9.16) and (9.17) and show that as wL varies, the .AC LIN 1 30K 30K
locus of the phasor I = x + j y is a semicircle.
In the ac analysis mode invoked by this statement, SPICE is set up to perform mul- where CVLIS T is a list of circuit variables. This list is formatted exactly as described
tiple ac steady-state analyses at user-specified sets of frequencies. We will have oc- in Chapter 4 for dc analysis, except that we specify the magnitude or phase of a variable
casion to make full use of this capability when studying frequency response in sub- by including M or P after the V (for voltage) or I (for current). For instance,
sequent chapters, but for present purposes, determination of the ac steady state at a
single source frequency is sufficient. Since the control statement is set up to accom- .PRINT AC VM(2) VP(2) IM(VDUMMY)
modate the more general purpose, however, we must be aware of the general format.
FVAR must be replaced by one of the keywords DEC, OCT, or LIN. This keyword results in the printing of the magnitude and phase of the node voltage phasor V 2 and
describes the manner in which the frequency variation of the set of frequencies to the magnitude of the current phasor through the voltage source VDUM MY. If rectangular
be used for repeated ac analysis is to be done: by decade, by octave, or linearly. components of the output are preferred, substitution of R or I for M or P will result in
NP is a number specifying the number of frequencies per decade, per octave, or in the printing of the real or imaginary parts.
the LIN case, the total number of frequencies in the set. FLO Wis the lowest fre- Recall that in some versions of SPICE, only currents through voltage sources may
quency, and FHIGH the highest frequency to be analyzed, with units of hertz (Hz). For be output, and it is necessary to install a zero volt "look-in" or "dummy" voltage source
instance, in series with any other element whose current is required and that does not happen to
have a voltage source already in series. If the rectangular rather than polar representation
.AC DEC 5 10 1000 is desired, inclusion of R or I after the leading V (voltage) or I (current) will result in
printing of the real part or the imaginary part, respectively. SPICE solves phasor circuits
specifies that an ac analysis is to be performed for each of five evenly spaced frequencies only; it is the responsibility of the user to convert the original sinusoidal sources to their
per decade, from 10Hz to 1 kHz. The logarithmic units of decade and octave are not phasor representations and then convert the response phasors back to sinusoids.
386 Chapter 9 AC Steady-State Analysis Section 9.6 SPICE and AC Steady State 387
To illustrate the application of SPICE, consider the circuit of Fig. 9.7. *SOLUTION CONTROL STATEMENT FOR f = 5000/(2*3.1416)
4 Let us find the voltage of node 1 in polar form and the current of .AC LIN 1 79577 795.77
the l-Q resistor in rectangular form. A SPICE deck for calculating .PRINT AC IM(Rl) IP(Rl)
these values is .END
AC STEADY-STATE SOLUTION FOR CIRCUIT OF FIG. 9.7
*DATA STATEMENTS The solution is
Vl lQo 0 AC 5 0
Rl 100 1 .5 FREQ IM(Rl) IP(Rl)
Cl 1 0 .5 7.958E+02 2.4ooE-02 5.313E+ol
C2 1 2 1
Ll 1 2 .5
L2 2 0 .25 Thus the ac steady-state current is 0.024 cos(5000t + 53.1).
R2 2 0 1
11 0 2 AC 5 0
*SOLUTION CONTROL STATEMENT FOR AC ANALYSIS [ f = 2/C2*PI) Hz] In Chapter 4 the use of subcircuits in SPICE was introduced. The subcircuit
.AC LIN 1 .3183 .3183 definition, a set of statements enclosed by SUBCKT and. ENDS control lines, may be
*OUTPUT CONTROL STATEMENT FOR V(1) & ICR2) included in the SPICE input file for the circuit containing the subcircuit, or it may
.PRINT AC VM(1) VP(1) IRCR2) II(R2) be stored as a separate library file. The latter is particularly useful if the subcircuit is to
.END be used in several different circuits. Inclusion of the control statement
The P RI NT statement is formatted for versions of SPICE that ac- .LIB FILENM
cept current references such as Ie R2 ). If your version of SPICE
only outputs currents through voltage sources I CVXXXXX), a
will cause the contents of the text file FI LEN Mto be linked to the main SPICE input
dummy source should be inserted in series with R1. The solution
file. FI LEN Mmust contain only subcircuits and, if we wish to link other library files
printed in this case is to FILENM, . LIB control statements. For instance, an op amp model introduced as a
subcircuit in Example 4.23 and repeated here is
FREQ VM(1) VP(1) IR(R2) II(R2)
3.183E-ol 2.263E+00 -2.657E+ol 2.oooE+00 4.o0oE+00
SUBCKT & OPAMP 1 2 3
The corresponding sinusoids are v = 2.26 cos(2t - 26.6), and since *NODE 1 is the + in, 2 the - in, and 3 the output
IR = 2 + j4 = 4.47(63.4, iR = 4.47 cos(2t + 63.4). RIN 1 2 lMEG
El 4 0 1 2 LOOK
As a second example, consider finding the phasor current I in the cir- RO 4 3 30
9.15 cuit of Fig. 9.9, which contains a controlled source of transresistance .ENDS
r = 3000 Q.
AC STEADY-STATE SOLUTION FOR FIG. 9.9. Since we will have occasion to use this model frequently, assume we have stored these
*DATA STATEMENTS five lines as a separate file named 0 PAM P . CKT. Example 9.16 illustrates how this library
V 1 0 AC 4 0 file can be used.
Rl 1 2 05K
R2 2 0 2K Let us find the phasor output voltage of the op amp circuit of
Cl 2 0 0.2UF Fig. 9.11 if the input voltage is Vg = lOcos(1000t + 30) V. A
H 3 2 V -3000 circuit file for the nodes of the op amp inverting input, op amp out-
R3 3 4 2K put, and input source, assigned as 3, 4, and lO, respectively, with
C2 4 0 0.2UF nodes 1 and 2 as shown, is
388 Chapter 9 AC Steady-State Analysis Section 9.6 SPICE and AC Steady State 389
AC STEADY-STATE SOLUTION OF FIG. 9.11 Consider the single-loop circuit of Fig. 9.25, with describing equa-
*DATA STATEMENTS USING OPAMP.CKT OF CHAPTER 4 tion for the mesh current i as follows:
.LIB OPAMP.CKT di
VG 10 0 AC 10 30 dt + (1 - a)i = Vg
sin2tA
+
,------l-E--l------, lloLooy + Vz
IF
2 In
~
.... 9.2.
....
Find VI using voltage division .
9.5. Find il using current division .
EXERCISE 9.6.2
2kn
8 cos 2tY
4cos2tY
1F 2n
In this chapter phasor analysis is systematically applied to ac steady-state circuits. All FIGURE P9.2
of our familiar tools, nodal analysis, mesh analysis, current and voltage division, series FIGURE P9.5
and parallel equivalents, and so on, pass over to the phasor domain unchanged. Their iDl9.3. Specify three impedances 21, 22, 23 so that in this
use in the phasor domain is greatly simplified by the fact that all currents and voltages Iill!I circuit VI = 6(60, V2 = 61J1, and V3 = 6(-60 V.
are constants, and all nonsource elements satisfy the same simple equation V = ZI. 9.6. Specify three admittances Yl, Y2, Y3 so that in this
circuit 11 = IIJ1 A, 12 = 1/60 A, and 13 = 1/-60 A.
Series impedances add, parallel admittances add.
Thevenin-Norton transformations and current or voltage division are unchanged in the +
phasor domain. 12Loo Y +
Nodal and mesh analysis are unchanged in the phasor domain.
For circuits with sinusoidal sources at two or more frequencies, superpOSItIon must
be invoked to define a separate phasor circuit for each frequency. Each is solved
independently. The overall ac steady-state response is the sum of the ac steady-state
(sinusoidal) responses of all these phasor circuits. FIGURE P9.3 FIGURE P9.6
lO kQ
a
2Q 2 cos 2tV IH
I---.-(.~ 1-----1
4kQ +j20Q
5 cos 2000t rnA
a'~~VV~--~------~ 8 kQ
lQ
fiGURE 1'9.11
FIGURE 1'9.7 FIGURE 1'9.18
9.12. Find the steady-state voltage v using nodal analysis.
D 9.8. Find the Thevenin equivalent to the right of the nGURliE 1'9.15
lillIl dashed line; then use the voltage-divider principle to 6 cos 8tV
FIGURE 1'9.12
lLO A t 4Q
2L180 A
+ 15L-45V
fiGURE 1'9.8 j2Q
9.9. For what ZI is VI = 1/90 V? 9.13. Find the steady-state current il using nodal analysis.
-jlO Q
FIGURE 1'9.16
FIGURE 1'9.19
FIGURE 1'9.9
FIGURE 1'9.13
9.10. Find the steady-state current i using nodal analysis.
For Problems 9.14 through 9.18, solve for the indicated
variables using nodal analysis.
~9.14.
lillIl >-_--0 V
+ v _
FIGURE 1'9.20
+4jQ
FIGURE 1'9.10
FIGURE 1'9.17 9.21. Solve Exercise 9.2.1 using mesh analysis.
9.11. Find the steady-state voltage v using nodal analysis. FiGURE 1'9.14
lp
4
j60
FIGURE P9.33
+
10 v
9.34. Find the steady-state value of v.
2 cos 8tV
FIGURE P9.28
40 IH 10
\
FIGURE P9.25
o 9.29. The controlled source has transconductance g =
lp l!.ml (1/2) S.
4 8 cos 2tV
fiGURE P9.23 For Problems 9.26 through 9.30, solve for the indicated
variables using mesh analysis.
09.26 + FIGURE P9.34
.... .
.... v
9.24. Find the steady-state current i.
~
9.35.
.... Find v .
20000 10000
FIGURE P9.29 +
6 cos tV v
....
09.30
.... 2 cos 2000t A
10 cos 2000t A
FIGURE P9.26
FIGURE P9.35
2H 1 kO -j20
1 kO +
-jl0 v
09.36. The sinusoidal sources Vg and ig operate at w =
rmi] 1000 and w = 2000 rad/s, respectively. If v =
~
9.27.
.... 4 cos lOOOt - 2 sin 2000t, find vg and i g
20 +
10 kO 100 fJ,p
FIGURE P9.30 v
FIGURE P9.24
I-----{+ -/----1 ~ 9.31. Solve Problem 9.18 using mesh analysis.
4kO 12 cos (2000t - 10) V SmH
+
9.25. Show that if Z, Z4 = Z2 Z3 in the "bridge" circuit v 8kO 9.32. Mesh analysis is restricted to planar circuits. Sketch FIGURE P9.36
shown, then I = V = 0 and therefore all the other currents a circuit that cannot be analyzed by the mesh method and
and voltages remain unchanged for any value of Zs. Thus that contains the minimum number of nodes for a nonplanar 9.37. Sketch the three given voltages as a phasor diagram,
it may be replaced by an open circuit, a short circuit, etc. circuit. and find V, the phasor corresponding to v, from the diagram.
In this case, the circuit is said to be a balanced bridge. FIGURE P9.27 9.33. Find the steady-state voltage v. Find v.
IQ
+ lF
8
2sint V
R
2H
+ v ~
fiGURE P9.48
FIGURE P9.37
9.49. Find V2 in ac steady state if VI = cos wt V.
o 9.38. Sketch the impedances of the three elements in a c 4Q
[;ill] phasor diagram. What must L be if the magnitude of the
overall impedance is IZI = 14 Q? Assume w = 10 rad/s. FIGURE P9.43
lOQ
9.44. Find vet) for w = 1, 2, 3, 4, 5 rad/s. Use only +
a
one SPICE run, and use the ideal voltage-controlled voltage
source op amp model of Fig. 3.7 with A = 100,000. +
L
a'
FIGURE P9.47
20kQ I/-tF
IF +
sinwtV IOkQ
FIGURE P9.38
v(t)
g ?.4.8. Find the ac steady-state Thevenin equivalent, spec-
[;ill] Ifymg V T and ZT. fiGURE P9.49
o 9.39. Sketch the locus of the total series impedance shown
[;ill] in Problem 9.38 in the complex plane as L varies from 0
FIGURE P9.44
to 1 H.
9.40. Find the locus of I, the phasor associated with i, as More Challenging Problems
w varies from 0 to 00. For which values of w is III largest?
9.45. Solve Problem 9.44 as specified, but use the im-
Smallest?
proved op amp model of Fig. 3.9 with A = 100,000,
i
--7 Ri = 1 MQ, and Ro = 30 Q. Enter the op amp model
in the SPICE input file as a SUBCKT.
9.46. For what L will i(t) = O?
cos wtV I Q IH IQ
James P. Joule
T =nTp (10.3)
where n is a positive integer.
In this chapter we consider power relationships for networks that are excited by periodic As an example, suppose that a resistor R carries a current i
currents and voltages. We concern ourselves primarily with sinusoidal currents and 1m cos wt with period T = 2]( jw. Then
voltages since nearly all electrical power is generated in this form. Instantaneous power,
as we now well know, is the rate at which energy is absorbed by an element, and it varies p = Ri 2
as a function of time. The instantaneous power is an important quantity in engineering = RI2
m cos wt
2
applications because its maximum value must be limited for all physical devices. For
this reason, the maximum instantaneous power, or peak power, is a commonly used RI2
specification for characterizing electrical devices. In an electronic amplifier, for instance,
= 2 m (1 + cos2wt)
if the specified peak power at the input is exceeded, the output signal will be distorted. Evidently, Tp = ]( jw, and therefore T = 2Tp. Thus, for this case',
Greatly exceeding this input rating may even damage the amplifier permanently. n = 2 in (10.3). This is illustrated by the graph of p and i shown
A more important measure of power, particularly for periodic currents and voltages, in Fig. 10.1.
is that of average power. The average power is equal to the average rate at which energy
is absorbed by an element, and it is independent of time. This power, for example, is
what is monitored by the electric company in determining monthly electricity bills. Aver- i(t), pet) T
age powers may range from a few picowatts, in applications such as satellite communica-
tions, to megawatts, in applications such as supplying the electrical needs of a large city.
Our discussion will begin with a study of average power. After introducing the
convenient root-mean-square (rms) metric, complex power is defined and its uses are
discussed. Superposition is related to power, and the notions of maximum power transfer
and conservation of power explored. The chapter ends with power factor, an idea of
great practical interest to companies supplying electrical power, and a comment on the
use of SPICE for ac power calculations.
In linear networks that have inputs that are periodic functions of time, the steady-state
currents and voltages produced are periodic, each having identical periods. Consider an
instantaneous power fiGURE 10.1 Instantaneous power and current waveforms (sinusoidal
p = vi (10.1) case).
(l0.5) for some 1m and <Pi, since in ac steady state all currents and voltages have the same
frequency.
The average power delivered to the device, taking t1 = 0 for convenience in
If we select m such that T = mTp (the period of v or i), then (10.6), is
wv. 1 (2:n:/w
2: m 10 + <Pv) cos(wt + <Pi)
P = -
T
11 II
11
+T pdt (10.6)
P =
Since the integral runs over two periods of the first term and the average value of any
sinusoid (with w =j:. 0) is zero, the integral of the first term is zero. The second term is
constant, and we have
Vml m
P = -2- cos(<Pv - <Pi) (10.10)
Thus the average power absorbed by a two-terminal device is determined by the ampli-
tudes Vm and 1m and the angle e by which the voltage v leads the current i.
In terms of the phasors of v and i,
V=Vm~ = IVI~
fiGURE 10.2 Periodic instantaneous power (nonsinusoidal case). I = Im/..!h = III/..!h
The average power absorbed by a two-terminal subcircuit is one-half the product of the
----------------~~~._-------------Re
magnitude of their current and voltage phasors times the cosine of the angle between them. <l>a
If the two-terminal device is a resistor R, then v - i = 0, and Vm = RIm, so 1= O.632L-18.4
(10.10) becomes
P R = ~RI;
I
FIGURE 10.5 Phasor diagram for selected currents and voltages from
It is worth noting at this point that if i = Ide, a constant (dc) current, then (j) = v =
the circuit of Fig. 1004.
i = 0, and 1m = Ide in (1O.7b). The instantaneous power vi = (RIde) (Ide) is constant
in this special case, so it equals its average, or
\
in phase. Hence by (10.11),
PR = RIJe
PR = ~ IV R11111 cos(/V R - /JJ.) = ~ (1.26)2 = 0.80 W
Let us determine the average power absorbed by various parts of the
ac steady-state circuit shown in Fig. lO.4(a), whose phasor circuit is For the capacitor we have
drawn in Fig. lO.4(b). Combining impedances yields
z = 2 + (j2)(1 - j2) = 6 + j2 Q
because the angle between voltage and current is Bb = / V c - = /JJ.
so -90 The net average power absorbed by all the impedances is
0
4LQ determined by the voltage 4LQ across the impedance subcircuit and
1= 6+j2 =0.632/-18.4 A 0
current I into it to be
Note from Fig. 10.3 that the terminal current and voltage used to
so V R = 1.26/71.6 V and
0 compute power absorbed by a subcircuit using (10.11) must satisfy
the passive sign convention. If they do not, (10.11) measures the
Vc = (-j2)1 1 =2.52/-18.4 V 0
negative of the power absorbed, that is, the power delivered by the
subcircuit to the rest of the circuit. Considering the voltage source
The phasor diagram for these currents and voltages is shown in
as a subcircuit, the terminal voltage 4LQ V and current I violate the
Fig. 10.5.
passive sign convention since the current reference direction arrow
We first determine the average power absorbed by the l-Q
points out of the positive end of the voltage reference direction.
resistor. The magnitude of its voltage is IVRI = 1.26 V, its current
Thus the 1.20 W absorbed by the impedances above is also the
is lIt I = 1.26 A, and the angle between these phasors is 0; they are
power delivered by the source. Indeed, the total power delivered
to all the elements absorbing power in a circuit is balanced by the
2Q 1Q power supplied, an intuitively reasonable conservation result to be
discussed in more detail in Section 10.6.
+
4 cos 2tV IF
4
4Lov Vc -j2 Q
Let us find the power supplied by the two sources in Fig. 10.6.
Combining the two parallel impedances gives
z= (1)(-jl) =~_j~ Q
FIGURE 10.4 (a) Circuit for Example 10.2; (b) phasor circuit. 1 - jl 2 2
The voltage V SI and the current source function 10LQ together vio-
late the passive sign convention relative to the current source. Thus
EXERCISE 10.1.1
the power delivered by this source, rather than absorbed by it, is
given by (10.11):
10.1.2. For a capacitor of C farads carrying a current i = 1m cos wt, verify
that the average power is zero from (10.6). Repeat this for an inductor of
PSI = !IVsllllOLQlcos44.6 = !(16.5)(10)(0.712) = 58.7 W L henrys.
10.1.3. Find the average power delivered to a lO-Q resistor carrying a
Since I and the voltage source function 2/30 also violate the passive current of:
sign convention, the power delivered by the voltage source is (a) i = 51sin lOti rnA
(b) i = lOsin lOt rnA, 0:::: t < JT/l0 s
= 0, JT /10 :::: t < JT /5 s; T = JT /5 s
PS2 = !III12/30 I cos (30 - 54.3) = !(7.23)(2)(0.911) = 6.59 W (c) i = 5 rnA, 0 :::: t < 10 ms
= -5 rnA, 10 :::: t < 20 ms; T = 20 ms
Both sources supply net power to the rest of the circuit. The power (d) i = 2t, 0:::: t < 2 s; T = 2 s
30
is in tum dissipated by the pair of resistors. Answer (a) 125 ~W; (b) 0.25 mW; (c) 0.25 mW; (d) I~O W
10.1.4. Find the average power absorbed by the capacitor, the two resis-
tors, and the source.
6 cos tV 60 .!.p Answer O.!Q, 3' ~.
3' -4 W
2 10.1.5. If It (t) is periodic of period TI and h (t) is periodic of period Tz,
Since the current and voltage phasors are always in phase for resistors, by (10.11) show that II (t) + h(t) is periodic of period T if relatively prime positive
resistors always absorb net power. For capacitors and inductors, the plus or minus 90 integers m and n exist such that
phase angle of their impedances drives their current and voltage 90 out of phase, so the
storage elements neither absorb nor supply steady-state average power in any circuit. It T =mTI =nTz
EXERCISE 10.1.4
is no surprise that none of the RLC elements supplies average power, since we know Extend this result to the function (1 + coswt)z, considered in this section,
them to be passive elements. Because inductors and capacitors do not absorb net power to find its period, T = 2JT /w.
either, they are referred to as lossless elements. Any element or subcircuit that absorbs
average power is referred to as lossy. Clearly, any LC subcircuit (one containing only
inductors and capacitors) is lossless, and any RLC subcircuit in which there is current
flow into at least one resistor is lossy.
As we saw in Example 10.3, there is no such uniform rule concerning delivery or
absorption of power that we may apply to sources. Sources can do either, depending on
the circuit contexts in which they are placed. For instance, the same automobile battery We have seen in Section 10.1 that periodic currents and voltages deliver an average power
will supply power to the starter motor subcircuit or absorb power from the alternator to resistive loads. The amount of power that is delivered depends on the characteristics
subcircuit depending on the switch positions in the rest of the circuit. of the particular waveform. A method of comparing the power delivered by different
P = RI2 = -1
rms T
iT 0
Ri 2 dt
Thus far we have defined the rms value of time-domain currents and voltages. Next
let V and I be phasors, and define their associated rms phasors by
1
V rms = ,J2 V (lO.l5a)
from which the rms current is
1
I rms = ,J21 (1O.15b)
I rms = liT
-
T 0
i 2 dt ~S phasors thus have the same angle as our usual phasors, but their magnitudes are
reduced by the factor 1/ ,J'2. Thus while the magnitude of a standard or amplitude phasor
equals the amplitude of the associated sinusoid, the magnitude of an rms phasor equals
the rms value of the associated sinusoid [by (1O.12a) and (lO.13a)]. An immediate use
In a similar manner, the rms voltage is of the rms phasor may be seen by substituting (10.13) into (10.10), revealing that
~ ( v2 dt
P = IVrmslllrmsl cos(v - i) (10.16)
T 10 Average power may be computed as the product of rrns phasor magnitudes times the
The term rms is an abbreviation for root mean square. Inspecting (1O.12a), cosine of the angle between current and voltage phasors without the factor of necessary !
that we are indeed taking the square root of the average, or mean, value of the with amplitude phasors as in (10.10).
Note that by dividing all phasors by ,J2, it is easy to show that rms phasors satisfy
the current. From our definition, the rms value of a constant (dc) is simply the
itself. The dc case is a special case (w = 0) of the sinusoidal current or voltage. Kirchhoff's current and voltage laws, and also that
Suppose that we now consider a sinusoidal current i = 1m cos(wt + ). Vrms = Zl rms (10.17)
(1O.12a) and (10.8), we find for any impedance. Thus, by designating all independent sources by their rms phasors
and elements by their usual impedances, we have an rms phasor circuit in which rms
WI2i2:n:/w I
I rms = 2 n:rn cos 2 (wt + ) dt = ~ currents and voltage phasors may be computed exactly like amplitude current and voltage
0 ,J2 phasors in amplitude phasor circuits.
Thus a sinusoidal current having an amplitude 1m delivers the same average
resistance R as does a dc current that is equal to Im/,J'2. We also see that the rms Let us find the power delivered by the 60-Hz ac generator to the load
is independent of the frequency w or the phase of the current i. Similarly, in impedance in the circuit of Fig. 1O.7(a). The rms source phasor is
of a sinusoidal voltage, we find that (325/,J'2)/..!1 = 230/..!1 V rms, and the rms phasor current is shown
in Fig. 1O.7(b). The equivalent load impedance is computed in the
usual way:
1 (- j53)(30) .
ZL = - II 30 = = 22.7 - ] 12.9 n
jwc 30 - j53
In an rms phasor diagram, all phasors are rms, and the mesh cur-
rent is
Substituting these values into (10.10), we have for any two-terminal subcircuit,
I rms = 230/..!1/(24.7 - j11.9) = 8.39/25.7 Arms
so
P = VrmsIrms cos(v - i) V rms = ZLlrms = (22.7 - j12.9) (8.39/25.7) = 219/-3.91 V rms
Then, by (10.16), the average power absorbed by the load impedance
where P is the average power absorbed if the passive sign convention is ZL is
the power delivered by the subcircuit if it is not. P = IVrmslllrmsl cos(<Pv - <Pi) = (219) (8.39) cos( -3.91 - 25.7)
We have seen that the use of phasors reaps great benefits in the study of ac steady-state
+ 22.7 - j12.9 Q circuits. By using complex numbers to represent real time-domain currents and voltages,
SOj.lF 30 Q 230Loo Vrms v;.ms we have dispensed with the frequent invocation of those identities from trigonometry
needed to combine sinusoids and, in addition, replaced the calculus operations of dif-
ferentiation and integration by the far easier arithmetic operations of multiplication and
2Q , division in the complex plane.
1- ___ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ I
In our present efforts to extend ac steady-state analysis to include power calcula-
AC generator Load impedance
tions, the complications of trigonometry and calculus have reappeared, for instance in
(a) (b)
(10.8) and (10.9). To extend the full benefits of phasor analysis to the study of power in
FIGURE "10.7 (a) Circuit for Example 10.4; (b) phasor circuit. ac steady-state circuits, we will once again define a new complex quantity, which we will
\
call complex power. Similar to the concise way in which phasors represent sinusoids,
complex power will encapsulate two essential power-related quantities in one compact,
or P = 1598 W. This is also the average power delivered by the ac
easily manipulated representation.
generator.
Given a two-terminal subcircuit with rms voltage and current phasors Vrms and
Irms, where
Vrms = IVrmsl/ v
EXERCISES I rms = II rms IL.!l!i
we define the complex power into the subcircuit as
10.2.1. Find the rms value of a periodic current for which one period is
defined by: S = Vrmsl~s (10.18)
(a) i = I, 0 ::5 t < 2 s where I~s is the complex conjugate of the rms current phasor. Exploring this definition,
= -I, 2::5t<48
the magnitude of the complex number S is given by
(b) i = 2t, 0::5 t < T
(c) i = 1m sinwt, 0::5 t ::5 nlw (l0.19a)
= 0, nlw ::5 t ::5 2nlw (T = 2nlw).
Answer (a) I; (b) 2T 1./3; (c) Iml2 and its angle is
10.2.2. Find the rms values of (a) i = lOsinwt+20cos(wt+300), (b)i = (10.19b)
8 sin wt + 6 cos (2wt + 10), and (c) i = 1(1 + cos 377t).
Since the angle of complex conjugate of any complex number is the negative of the angle
Answer (a) 12.25; (b) 7.07; (c) 1ft
of the number itself,
10.2.3. Find Vrms .
Answer 4 V ~ = fVrms -fIrms = v - i
Turning to the rectangular form for S, its real part is Re(S) = lSI cos~, which, by
(10.19), is
(10.20)
Comparing the last with (10.6), we have an important connection with our earlier results.
The real part of the complex power is the average power,
2 cos 4tV 6cos4tV
Re(S) = P (10.21)
Thus we have another way to compute the average power: compute the complex power
S; then take the real part to get P.
EXERCISE 10.2.3 To avoid sign errors, the rms current and voltage phasors used to define the complex
power S into a subcircuit will be assumed to satisfy the passive sign convention. In this
= 82.7/-135
Re(Z) 2
P = ~lVrmsl (l0.24b)
Note that reversing the sign of a complex number adds or subtracts
Note that if the impedance has an angle of plus or minus 90, P = O. This is the lossless
180 to its angle in the complex plane, while taking complex con-
case discussed earlier. If the real part of Z is positive, the impedance absorbs net power
and is lossy. Combining these cases, impedances with angles from minus to plus 90
IQ
correspond to passive elements, while those with angles outside this range must be active
elements. Any interconnection of passive elements is passive; thus we expect the angles
-jl Q
of impedances constructed from passive elements to be 90 or less in magnitude.
In terms of the admittance, if the subcircuit satisfies
Irms = YVrms
- 12L30 V rms
then, by the definition of complex power,
S = VrmsI~s = Y*I VrmsI 2 (10.25)
2Q and
(a) (b)
To determine the power into the load in Fig. 1O.9(a), we use the rms
phasor circuit shown, where FIGURE 10.10 (a) Circuit for Example 10.7; (b) rms phasor circuit.
ZL=2+jlQ
\
Y 1 = jl + (- j5).(2) = 1.72 + jO.31 S
2 - J5
By voltage division,
Since G and YI are in parallel, the total admittance is
50/./2 35.4(2 + jl)
VL nns = ZL = -----=-- Y= G + Yl = (G + 1.72) + jO.31 S
ZL + ZI 2.92 - jO.38
Substituting this into (10.27) yields
= 26.9/34.1 V rms
4(G + 1.72) = 1
Then by (1O.26a) with
(G + 1.72)2 + 0.312
1 1
Y L = ZL = 2 + jl = 0.4 - jO.2 S This yields a quadratic equation for G:
G2 - 0.56G - 3.83 = 0
we have
The roots of this quadratic are G = 2.25 and G = -3.39. Since
PL = Re(Y) IV L nns 12 = 0.4(26.9 2 ) = 289 W
G > 0 for passive conductors, the unique solution is G = 2.25 S,
which is a 1/2.25 = 0.444-Q resistor.
1p
2
Having identified the real part of the complex power S with the average power P,
we next tum to the complex part, which we define as the reactive power Q. Thus
s = VnnsI::ms = P + j Q (10.28)
50 cos t 50 LOVrms + where P is the average power in watts and Q the reactive power,
V -..{2
P = IVnnsllInnsl cos(<Pv - <Pi) (1O.29a)
Reversing the reference direction ofI nns to make it satisfy the passive
= J2IVnns l sine cos(wt 90 0 ) sign convention together with Vs nn for calculation of the power
The phase angle is written as 90 since the sign of this term depends on whether Vnns absorbed by the source,
happens to lead Inns, the case shown in Fig. 10.11, or lag. The instantaneous power in
the quadrature component is then the product
Sv = Pv + jQv = (l2LQ)(8.5/-135)* = -72 + j72
Pq(t) = vq(t)i(t) = [2lVrms llInnsl sine] coswt cos(wt 90 0 )
The reactive power of the resistor is zero, as anticipated since current
and voltage in resistors are in phase. The storage elements are pure
which may be simplified using the trigonometric identity introduced in Section 10.1 to be reactances and have 90 0 phase shift between their currents and volt-
ages, and thus the complex power is purely reactive (zero average
Pq(t) = IVrmsllInnsl sine cos(2wt 90 0
)
power). Note that the magnitude of the reactive power Qc of the
From this expression we see that the instantaneous power in the quadrature com- capacitor is twice that of the inductor QL. Since the impedance is
ponent Pq(t) is a sinusoid with amplitude equal' to IVnnsllInnsl1 sinel. By (10.29), this larger by a factor of 2, the voltage across the capacitor is twice that
equals the magnitude of Q. Thus we arrive at a physical interpretation for the quantity Q: across the inductor, and since both these voltages are in quadrature
the magnitude of the reactive power Q is the amplitude of the instantaneous power in the with the mesh current, the amplitude of the instantaneous power in
quadrature i-v component. Since the instantaneous power in the quadrature component the quadrature component is twice as large for the capacitor as for
is a sinusoid centered at zero, it indicates energy being transferred periodically to the the inductor. The capacitor exchanges more energy per period, back
subcircuit and then back from the subcircuit to the rest of the network in equal amount. and forth with the rest of the circuit, by a factor of 2. Finally, note
From earlier chapters we know this behavior to be characteristic of energy storage ele- that since the reactive power summed over the passive elements does
ments, neither supplying nor dissipating power on average, but rather exchanging it back not equal zero, the source must supply vars (as well as watts). That
and forth with the rest of the circuit. Reactive power Q is thus a measure of the amount is, the source must develop an out-of-phase as well as an in-phase
of periodic energy exchange taking place between a given subcircuit and the rest of the component. How the quadrature component may be adjusted to re-
circuit. By (1O.29b), the sign of Q is positive for inductive reactances and negative for duce the total current-voltage requirements at the source is discussed
capacitive ones. in Section 10.7.
P = PI + P2 + Re(V11; + V2 m
Since the last term in general is not zero, superposition does not in general hold for
average power P. That is, the watts absorbed will not be the simple sum of those due to
each source separately. Similarly, equating the imaginary parts of (lO.32b) yields
In this section we consider power in networks containing two or more independent
sources. Let i and v be the terminal current and voltage into a two-terminal subcir-
Q = QI + Q2 + Im(VII~ + V2 m
cuit in a circuit containing two independent sources. The principle of superposition Superposition does not in general hold for reactive power Q.
always applies to currents and voltages in any linear circuit, so The failure of superposition shown above clearly extends to the case of more than
two sources. Thus, in ac steady-state circuits containing multiple sources at the same
i = i l + i2 frequency, power cannot be computed by superposing powers due to individual sources.
This applies equally to all forms of power: instantaneous, average, reactive, or complex.
v = VI + V2 Superposition may be used to find currents and voltages, but for sources at the same
where the subscripts refer to the responses due to each source separately with the other frequency, these components must be superposed before power is computed.
killed. The instantaneous power absorbed by the subnetwork is, assuming i and v have
reference directions that satisfy the passive sign convention, To determine the net power supplied by the voltage source in
Fig. lO.I2(a), we note that since both sources in the circuit are at the
p = iv = (il + i2)(vl + V2) same frequency w = 2 radls, we may determine the ac steady-state
or p = i l VI + i2v2 + (il V2 + i2 v I) (lO.30a) response using a single phasor circuit. The rms phasor circuit for
w = 2 radls is shown in Fig. lO.I2(b). Let II be the component ofI
p = PI + P2 + (ilv2 + Vl i2) (lO.30b)
2H j4 n
where PI and P2 are the instantaneous powers due to each source separately. If su-
perposition were to apply to power, we would have p as the sum of PI and P2,
which from the last equation is not in general the case. Superposition does not in
general hold for instantaneous power. That is, the instantaneous power into a sub- In In
network is not in general simply the sum of the instantaneous powers due to each 2 c08(21 + 20) A
excitation separately. Failure of superposition to work in general for power is a con-
sequence of the nonlinear nature of power, its definition as the product of two circuit
variables. n
lp -jl
Next consider the same circuit in ac steady state. Suppose that both sources operate 2
(a) (b)
at the same frequency w. If 11 and 12 are the rms current phasors and VI and V 2 the
corresponding voltage phasors produced by the two sources, applying superposition to fiGURE 10.12 (a) Circuit diagram; (b) rms phasor diagram.
420 Chapter 10 AC Steady-State Power Section 10.4 Superposition and Power 421
due to the voltage source and 12 that due to the current source. Then Using the identity (10.8), for WI =1= W2, both resulting terms in the integrand are sinusoids
with nonzero frequency, both have zero average values, and thus (10.35) evaluates to
3~LQ)
II =- 1 + j3 = -0.424 + j1.27 Arms zero. By identical reasoning, the second term in the integral of (10.34) also evaluates to
zero, and (10.35) simplifies to
and by current division
~~
12 = 1 + j3 = 0.278 - jO.35 A rms In other words, superposition holds for average power P when the sources are of different
frequencies. Thus to compute the average ac steady-state power in a circuit excited by
Thus sources at two or more frequencies, we compute the average power for each frequency
separately and then add average powers. By a similar argument, it can be shown that
I = II + 12 = -0.146 + jO.92 Arms
superposition also holds for reactive power Q when the sources are of different frequen-
and the complex power absorbed by the voltage source is \ cies. Interestingly, when phasors can be superposed (WI = (2), neither P nor Q can
be, and when phasors cannot be superposed (WI =1= (2), superposition works for both P
S = VI* = (3.J2LQ) (-0. 146 - jO.92) = -0.62 - j3.9 and Q.
Finally, since both P and Q obey the superposition principle for distinct frequen-
We conclude that the voltage source delivers 620 mW of average
cies, so must S, the complex power. Two examples illustrating the use of superposition
power and 3.9 vars of reactive power to the rest of the circuit.
of power in ac steady-state circuits excited by sources of different frequencies follow.
j80 j40
1 iT Iml Vm2iT
- (i l V2) dt = - - - COS(Wlt + il) COS(W2t + v2) dt (10.35) FIGURE 10.13 (a) Circuit diagram; (b) rms phasor diagram (w
ToT 0 4 rad/s); (c) rms phasor diagram (w = 2 rad/s).
422 Chapter 10 AC Steady-State Power Section 10.4 Superposition and Power 423
differ, we must compute their individual responses from different
phasor circuits. Then we may apply superposition to the average
powers produced at each frequency.
By current division 10.4.1. Find the average power delivered to the resistor if R = 10 Q and:
R
'15/2 (a) VgI = 10 cos lOOt and Vg2 = 20 cos(100t + 60) V
II = (h/200) } = 1.40/27.6 Arms (b) VgI = 100eos(t + 60) and Vg2 = 50sin(2t - 30) V
1+j15/2
and the power into the resistor at w = 4 is
(c) VgI = 50cos(t + 30) and Vg2 = 100sin(t + 30) V
(d) VgI = 20cos(t + 25) and Vg2 = 30sin(5t - 35) V
PI = Re(Z)1111 2 = RIII12 Answer (a) 15 W; (b) 625 W; (c) 625 W; (d) 65 W
10.4.2. Find the average power absorbed by each resistor and each source.
= (1)(1.40)2 = 1.96 W
Answer 8 W; 24 W; -8 W; -24 W
EXERCISE 10.4.1
For w = 2 the mesh current is
12 = -3v'2/J1
- = 1.34/-71.6 Arms
1 + j3 -
and the power into the resistor at w = 2 is
P2 = RI1212
IQ 8cos4tV
= (1)(1.34)2 = 1.80 W
Thus the average power absorbed by the resistor is 1.96 + 1.80 =
3.76 W.
EXERCISE 10.4.2
We wish to find the average and reactive power absorbed by the load
in Fig. 10.14. To determine the power at w = 2, we kill the source 10.4.3. Find the average power absorbed by the resistor and each source.
at w = 3, and since the two remaining voltage sources are in series, Answer 8 W; -4 W; -4 W
the rms phasor current at w = 2 is
2Q
424 Chapter 10 AC Steady-State Power Section 10.5 Maximum Power Transfer 425
,---------
: I are complex conjugates:
----?>-
(l0.37)
Thus maximum power is transferred to a load if the magnitude of its impedance
is matched to that of the source impedance ZT and the angle of the load impedance
matched to the negative of the angle of the source impedance. Load impedances whose
magnitudes are large will sharply limit the current I, resulting in less power delivery,
,
~ _________ J while those with small impedance will have only a small fraction of the voltage V T
Two-terminal snbcircnit Load impedance dropped across it, again reducing the power draw.
To determine the maximum power draw, using ZL = Zf in (1O.36b),
fiGURE 10.15 Load impedance connected to a subcircuit.
P _ Re(ZT)IVTI2
max - 12 Re(ZT) 12
which it is attached. The process of specifying the load for maximum power from a
given subcircuit is sometimes called impedance matching, for reasons that will soon be IVTI2
=
evident. 4 Re(ZT)
To simplify the analysis, we assume that the given two-terminal subcircuit is spec- or, in terms of YT = IjZT,
ified by its Thevenin equivalent, as shown in Fig. 10.15. The current into the load is
given in terms of rms phasors by Pmax = Re(YT)IVTI 2
4
VT
1=--- IITI2
ZL +ZT
4 Re(YT )
and the average power absorbed by the load is, using (1O.24b),
It should be remembered that rms phasors are used in these expressions.
PL = Re(ZL)III 2 (1O.36a)
Given the rms phasor circuit of Fig. 10. 16(a), we first convert all
Re(ZdIV TI2
(lO.36b) but the load impedance to its Thevenin equivalent, as shown in
IZT + ZLI 2 Fig. 10. 16(b). ZT is the series equivalent of the impedances 3 and
Let ZL = RL + jX L and ZT = RT + jX T . In terms of these variables, (l0.36b) may be j4, or 3 + j4. The open-circuit voltage with ZL removed is
written V T = (3)(4LQ) = 12LQ V rms
P _ RL IV 12 Thus, for maximum power transfer, we select ZL = Zf = 3 - j4 Q.
L - (RL + RT)2 + (XL + X T )2 T The maximum power draw is
We wish to know the values of RL and XL that maximize P for fixed values of the 122
remaining quantities. The maximizing choice for XL is clearly XL = -XT' since this Pmax =- - = 12 W
(4)(3)
makes the denominator as small as possible. Setting XL = -XT gives
j4n
the maximizing choice for RL is found by setting this partial derivative to zero. It is easily
checked that the partial above is zero for RL = RT (and that the second partial derivative
is negative, indicating a maximum). We conclude that the maximum power is transferred 12Lo v nns + zr,
to a load impedance ZL when the real parts of the source impedance RT = Re(ZT) and
load impedance RL = Re(Zd are equal, and their imaginary parts X T = Im(ZT) and
XL = Im(Zd are equal and opposite in sign. We may state both requirements together
(a) (b)
by concluding that the maximum power is transferred to a load ZL from a given source
with Thevenin equivalent impedance ZT if the load and Thevenin equivalent impedance fiGURE 10.16 (a) Circuit for Example 10.12; (b) Thevenin equivalent.
426 Chapter 10 AC Steady-State Power Section 10.5 Maximum Power Transfer 427
where the term containing the reference node current is shown separately. There is one
EXERCISES such equation for each node j = 1, ... , n. Adding all n equations together gives
4mH 10.5.1. Find values for load resistance R and and load inductance L so
that the source with amplitude A = 2",fi V and frequency w = 1000 rad/s
delivers maximum power to the load. Find the value of the power delivered
t [t Vjljk + Vjljo] = 0
to the load at these maximizing values for Rand L. The current between every pair of nonreference nodes appears twice in the first double
Answer 1.2 Q; 4.73 mB; 833 mW sum, as Ijk and its negative, Ikj . Replacing I kj by - Ijk for k < j, these two terms may
A cos wt 10.5.2. Repeat Exercise 10.5.1 for A = 4",fi V and w = 1000 rad/s. Note be combined, producing the equation
V that the optimal values for the load impedance do not depend on the source
strength A, but the maximum power delivered does.
Answer 1.2 Q; 4.73 mH; 3.33 W
t [t
j=1 k=j+1
(Vj - Vk)ljk + Vjljo] = 0
10.5.3. For the given rms phasor circuit, determine the load impedance
EXERCISE 10.5.1 Each term in this equation is recognized to be the complex power Sjk absorbed by the
Z L that would absorb maximum power.
Answer 4.0/41.6 Q subcircuit connecting a particular node pair (j, k), and each node pair in the circuit
appears exactly once. Thus this equation may be simplified to
(10.38)
node pairs (j, k)
Finally, we note that since we have already counted all nodes in the circuit, the subcircuit
connecting nodes j and k can only consist of elements in parallel. Suppose that there are
m such, for m > O. These individual elements are shown in Fig. 10.17. The complex
power into this parallel subcircuit is
The terms inside the summation are the complex power absorbed by each of the m
elements separately. Combining this with (10.38), we have the principle of conservation
Conservation laws play key roles throughout the physical sciences. In circuit theory, no of complex power: the sum of the complex powers absorbed by all the elements in a circuit
principle is more fundamental than the conservation of electrical charge, an immediate equals zero.
consequence of which is Kirchhoff's current law. In this section we see how the general Conservation of complex power immediately implies conservation of average power
principles of conservation of power and energy apply to electrical circuits in ac steady P and reactive power Q, since these are the real and imaginary parts of S. By putting
state.
Consider an arbitrary n + 1 node circuit with nodes numbered 0, 1, ... , n. Let Vj Ijkt Vj (node))
be the node voltage rms phasor at node j = 1, ... , n and node 0 be the reference node.
The current flow from node j to node k will be designated Ijk. By KCL at node j,
Ijklt
t 1jk2 t 1jkm
n
L Ijk = 0 Em
k=O
where Ijo is the current flow into the reference node and Ijj = O. Taking the complex
conjugate of this KCL equation and mUltiplying by the node voltage Vj yields
(nodek)
n
L Vjljk + Vjljo = 0
FIGURE 10.17 Elements connecting two nodes.
k=1
Se = 0 - jO.50
432 Chapter 10 AC Steady-State Power Section 10.7 Reactive Power and Power Factor 433
1m in parallel with a 25-/LF capacitor connected to a 120-V fIllS 60-Hz
source.
Answer (a) 2.3 kVA; (b) 197.9 VA
y ,""-:--:rIr;:---,-- Re 10.7.2. Find the power factor for (a) a load consisting of a series con-
nection of a lO-Q resistor and a lO-mH inductor operating at 60 Hz, (b) a
capacitive load requiring 25 A fIllS and 5 kW at 230 V fIllS, and (c) a load
that is a parallel connection of a 5-kW load with a power factor of 0.9
(a) leading and a lO-kW load with a 0.95 lagging power factor.
Answer (a) 0.936 lagging; (b) 0.87 leading; (c) 0.998 lagging
FIGURE 10.19 (a) Circuit for power factor correction; (b) phasor diagram. 10.7.3. A load consisting of a series combination of a 100-Q resistor and
I-H inductor is connected to a source delivering power at w = 100 rad/s.
In other words, Y 1 must reduce the imaginary part of Y so that with fixed real part the What capacitor C across this load will correct the power factor to (a) 0.9
angle of YT will have the desired power factor PF. leading; (b) 0.9 lagging?
\
Answer (a) C = 74.22 /LF; (b) C = 25.78 /LF
As an example, let us change the power factor for the circuit of Ex-
ample 10.14 (shown in Fig. 10.18) to 0.98 leading. The impedance
of this circuit was found in Example 10.13 to be Z = 0.4 + j 1.2 Q,
so its admittance is
Y= ~= 0.25 - jO.75 = 0.791/-71.6 S
Z
Thus the circuit as given has a power factor of cos(71.6) = 0.32, When an ac analysis is requested of SPICE by inclusion of the .AC control statement,
which is a lagging power factor (the angle of an admittance in this some additional information labeled "small-signal bias solution" accompanies the output.
case -71.6, is the angle by which the current leads the voltage). This includes a "total power dissipated" value. For instance, Example 9.17 used the
The desired YT must have an angle of 8d = cos- 1 0.98 = +11.5 (YT source file
of phase angle -US would be 0.98 Zagging rather than leading).
The unique complex number that has real part 0.25 and angle + 11.5 Circuit of Fig. 9.25
has as its imaginary part *
Im(YT) = (0.25) (tan US) = 0.051 S
V 1AC 1 0 0
L 11 2
Thus we must change the admittance from Y = 0.25 - jO.75 S to H 2
3 V 0.5
YT = 0.25 + jO.051 S. This requires that R 3 0 1
Y1 = YT - Y = j(0.051 + 0.75) = jO.801 S .AC LIN 1 .159 .159
.PRINT AC IM(V) IP(V)
Thus, to correct the power factor to the desired value, we need to
.END
install an admittance of jO.801 S or, equivalently, an impedance of
1/ jO.801 = - j 1.25 Q, in parallel with the given impedance. For
The following data were found in the SPICE printout above the printout that we requested
instance, if the working frequency for the circuit is w = 1 rad/s, then
the power factor correction would be achieved by using a capacitor (the magnitude and phase angle of the current though the voltage source labeled V).
with impedance
1 . *** SMALL SIGNAL BIAS SOLUTION TEMP = 27.0 DEG C ***
- - = -J1.25 Q
j(I)C
NODE VOLTAGE NODE VOLTAGE NODE VOLTAGE
( 1) 0.0000 ( 2) 0.0000 ( 3) 0.0000
or C = 0.80 F.
VOLTAGE SOURCE CURRENTS
NAME CURRENT
EXERCISES V O.OOOE+OO
10.7.1. Find the apparent power for (a) a load that requires 20 A fIllS TOTAL POWER DISSIPATION O.OOE+OO WATTS
from a 1l5-V fIllS line and (b) a load consisting of a 100-Q resistor
434 Chapter 10 AC Steady-State Power Section 10.8 SPICE and AC Steady-State Power 435
The power dissipation reported in this portion of the printout is not related to the absorbing this 400 mW. In addition, the independent source supplies
ac power dissipated by the elements of the circuit. Whenever an ac analysis is requested, 0.4 var to the inductor.
SPICE automatically first determines the dc operating point of the circuit, that is, the
dc or constant values of all currents and voltages when the ac sources are killed. This Finally, note that rms values may be used to describe the magnitude of sources in
operating point information is of great value in the case that nonlinear elements, such as SPICE input files. In this case all calculated current and voltage phasor magnitudes will
transistors and diodes, are included in the circuit, and dc supplies are used to power them. !
be in rms as well, and the factor of should be omitted from calculations of complex
This is the "small-signal bias solution" reported. The "total power dissipation" reported power such as (1O.40a).
refers to steady-state dc power required to support this operating point. In a circuit with
no dc sources all these values will be zero. Thus the power dissipation automatically
reported by SPICE is of no help in determining ac steady-state power dissipations at EXERCISES
nonzero frequencies.
Sadly, SPICE offers no specific control card or keyword for printing out ac steady-
state power at any other frequency than w = 0 (dc) as described above. To determine \ 10.8.1. Find the power delivered to the 80-Q resistor.
Answer 13.0 mW
the ac power absorbed by a circuit element using SPICE, we must request output of the
current and voltage phasors; then use the formulas studied in this chapter to get complex, 172 Q
average, or reactive power (as required) from knowledge of the element current and
voltage. 24Q 50J.IF 2mH
The current phasor associated with the counterclockwise mesh cur- 10.8.2. Check the results of Example 10.11 (Fig. 10.14) using SPICE.
rent i in Fig. 10.20 is thus I = 0.895/116.6 A. Since this current Answer 3.02 W, 2.74 var. Note that two SPICE runs are required,
phasor satisfies the passive sign convention relative to the indepen- one at w = 2 radls and one at w = 3 radls (0.318 and 0.477 Hz).
1Q
dent source phasor voltage V = lLQ, the complex power absorbed
FIGURE 10.20 Circuit for Example 10.16. by the independent source is given by
!VI* = !(1LQ)(0.895/-116.6) = -0.2 - JOA (1O.40a)
and the complex power absorbed by the dependent source is given
by The power calculations are important not just to the electrical power generation and
! HI) (-1)* = -0.2 + jO (l0.40b) distribution industries, but also in the design of efficient circuits of all kinds and of
miniaturized circuits which must dissipate heat over very small areas. Just as was the
Note that we used -I, not I, in the last calculation, since -I satis- case with currents and voltages in Chapters 8 and 9, the calculation of power in ac
fies the passive sign convention for the dependent source (together steady-state circuits is greatly simplified by the use of phasors.
!
with its voltage I). Also note that the phasors reported by SPICE
are amplitude, not rms, phasors since we used amplitude phasors For terminal variables v(t) and i (t) that satisfy the passive sign convention, instanta-
to describe the source. Thus we must include the factor !in the neous power pet) = v(t)i(t) greater than zero implies the circuit between the terminals
power formulas (10.40) consistent with the use of non-rms standard is absorbing power at t, less than zero that it is supplying power to the rest of the circuit
or amplitude phasors. Examining our result, we see that the inde- att.
pendent and dependent sources each deliver 200 mW to the rest of Average power equals the product of current amplitude, voltage amplitude, and the
the circuit. By conservation of average power, the resistor must be cosine of the difference of their phase angles. If current and voltage satisfy the passive
10.2. Let v(t) = 2cosnt V for 3n :::: t < 3n + 1, n = 0, 10.6. Find the average power absorbed by the resistors,
1, 2, ... , and vet) = 0 elsewhere. If vet) is across a the inductor, and the source. 10.10. Find the average power absorbed by the 3-kQ re-
l-Q resistor, sketch the instantaneous power pet) and find sistor and the dependent source.
the average power P. 30
5 cos 2000t V 10 cos 2000t V
10.3. One cycle of a periodic current is given by 1 kO 3kO
+
i=lOA,O::::t<lms 6cos4tV 60 FIGURE P10.15
20 cos 10,ODOt V OAH
= 0, 1 :::: t < 4 ms
10.16. Determine the rms current; i (t) is periodic with
If the current flows in a 20-Q resistor, find the average period 1 ms, and equals sin 2m for 0 :::: t < 0.5, and zero
power. FIGURE P10.6 FIGURE P10.10 for 0.5 :::: t < 1 ms.
20
10.33. For a Thevenin equivalent circuit consisting of a
voltage source Vg and an impedance Zg = Rg + jX g,
FIGURE 1'10.20 (a) show that the circuit delivers maximum average power
to a load ZL = RL + jXL when RL = Rg and XL = -X g,
10.21. Find the complex power delivered to each inductor ) "." = 2 ''"'' V and (b) show that the maximum average power is delivered
and resistor if Vg = 0.10 sin (2t + 30) V. to a load RL when RL = IZgl. This is the maximum power
FIGURE 1>10.25
transfer theorem for ac circuits. (In both cases, V g and Zg
FIGURE P10.29 are fixed, and the load is variable.)
10.26. Find the rms value of the steady-state current in R.
10.34. What must the amplitude A of the source be to
iii 10.30. Compute the watts supplied to the load ZL.
supply 100 W to the purely resistive load ZL = 2 Q?
What would it be reduced to if we could adjust Z L to absorb
+ 100 W using the lowest possible value source amplitude A?
5 cos tV
IH 2H
+ 18cos4tV R=10 t 8A 30
Acos2tV + 10
FIGURE 1'10.21
300 !iF
400mH
FIGURE Pl0.38 I
12 -{2 cos 4t V + 2H 10 cos 8tV i6 F
10.39. Verify conservation of complex power for this rms
phasor circuit.
FIGURE Pl0.49
FIGURE Pl0.35 FIGURt P10.44
+j60 +j40
10.45. Find the power factor seen from the terminals of 10.50. Repeat Problem 10.21 using SPICE. Use the ideal
010.36. Find the load impedance ZL that will absorb max- the independent source and the reactance which must be voltage amplifier op amp model of Fig. 3.7 with open-loop
[ill;] imum real average power. Find the value of the maximum connected in parallel with the independent source to change gain A = 106 .
real average power into the load. ISLO V the power factor to 0.8 lagging.
More Challenging Problems
j2 Q 10.51. At what frequency w is the rms value of V2 maxi-
mized?
FIGURE P10.39
10 LOoVnns
SPICE Problems
10.48. A load impedance ZL = 1 + jO.5 is connected
to a lO-mA rms ideal current source. What is the power
120 4cos(t-I7")V
28 cos 16tV factor of the load? How much could the voltage rating
FIGURE Pl0.37
of the current source be reduced if a capacitor was put in
series with the load, creating an overall power factor of 0.95
o 10.38. Find the reactive power absorbed by the inductor. lagging? Use SPICE to determine the capacitance which
corrects the pf to 0.95 lagging at 60 Hz. fiGURE P10.52
[ill;] (Hint: Find the complex power supplied by the source.) fiGURE P10.43
II .,
In
fiGURE P10.53
Nikola Tesla
1856-1943
g 10.54. Find the complex power delivered by each source. He [Teslal was the great-
Iill!l il= 2 cos 2t A, i2 = 6 sin 2t A, VI = 12 sin(3t - 45) V, fiGURE 1>10.54 est inventor in the realm of
V2 = 24 sin(3t - 45) V.
electrical engineering.
W. H. Eccles
As we have already noted, one very important use of ac steady-state analysis is its ap- are the rms phasors associated with an element having impedance,
plication to power systems, most of which are alternating current systems. One principal Z = IZILQ Q 01.2)
reason for this is that it is economically feasible to transmit power over long distances
only if the voltages involved are very high, and it is easier to raise and lower voltages the average power delivered to the element is
in ac systems than in dc systems. Alternating voltage can be stepped up for transmis- p = IVI . III cose
sion and stepped down for distribution with transformers, as we shall see in Chapter 15.
Transformers have no moving parts and are relatively simple to construct. In a dc system, = 1112 Re(Z) W (11.3)
on the other hand, the waveforms must first be converted to ac before transformers may In the time domain the voltage and current are
be used to change voltage levels and then converted back to dc. This process requires
expensive high-power switches and dissipates some of the energy it is converting. v= hivi cos wt V
Also, for reasons of economics and performance, almost all electric power is pro-
duced by polyphase sources (those generating voltages with more than one phase). In i = hili cos(wt - e) A
a single-phase circuit, the instantaneous power delivered to a load is pulsating, even if All phasors in this chapter will be understood to be rms phasors, whose magnitude
the current and voltage are in phase. A polyphase system, on the other hand, is some- corresponds to the rms value of the associated current or voltage, as introduced in Sec-
what like a multicylinder automobile engine in that the power delivered is much steadier. tion 10.2.
An economic advantage is that the weight of the conductors and associated components
required in a polyphase system is appreciably less than that required in a single-phase The use of double subscripts makes it easier to handle phasors both
system that delivers the same power. Virtually all the power produced in the world is analytically and geometrically. For example, in Fig. 11.2(a), the
polyphase power at 50 or 60 Hz. In the United States, 60 Hz is the standard frequency. voltage Vah is
In this chapter we begin with single-phase three-wire systems, but we concentrate on
three-phase circuits, which are by far the most common of the polyphase systems. In the Vah = Van + Vnh
latter case the sources are three-phase generators that produce a balanced set of voltages, This is evident without referring to a circuit since by KVL the voltage
by which we mean three sinusoidal voltages having the same amplitude and frequency between two points II and b is the same regardless of the path,
but displaced in phase by 120. Thus the three-phase source is equivalent to three which in this case is the path a-n-b. Also, since Vnh = -Vhn, we
interconnected single-phase sources, each generating a voltage with a different phase. If have
the three currents drawn from the sources also constitute a balanced set, the system is
said to be a balanced three-phase system. A balanced system is required to reap the full
benefits of polyphase power, and this is the case on which we concentrate our attention. = 100 - 100/-120
446 Chapter 11 Three-Phase Circuits Section 11.1 Single-Phase Three-Wire Systems 447
a a v"b v"b and bB are
+
;,r-----------
laA = Van = VI
,, ZI ZI
,,
IbB - = -VI
= Vbn - = - laA
ZI ZI
6-----1- + I-----<l
Therefore, the current in the neutral wire, nN, by KCL is
n b
Vbn= lOoL-120 V nns InN = -(laA + IbB ) = 0
(a) (b) Now it is always the case that a wire carrying no current may be freely removed in
any circuit without affecting any current or voltage, since its removal changes no KVL,
FIGURE 11.2 (a) Phasor circuit; (b) corresponding phasor diagram.
KCL, or element law equation for that circuit. Thus in Fig. 11.3(b), the neutral could be
\ removed without changing any current or voltage in the system. We use this procedure
which, after simplification, is to simplify circuits throughout the chapter.
If the lines aA and bB are not perfect conductors but have equal impedances Z2,
Vab = 100.J3"/30 V rms then InN is still zero because we may simply add the series impedances ZI and Z2 and
have essentially the same situation as in Fig. 11.3(b). Indeed, in the more general case
These steps are shown graphically in Fig. 11.2(b).
shown in Fig. 11.4, the neutral current InN is still zero. This may be seen by writing the
two mesh equations
. A single-phase three-wire source, as shown in Fig. 11.3, is one having three output
termmals a, b, and a neutral terminal n that center-taps the voltage Vab:
(ZI + Z2 + Z3)laA + Z3 1bB - Z1 13 = VI
Z31aA + (ZI + Z2 + Z3)lbB + Z1 13 = -VI
(11.4)
and adding the result, which yields
This is a common arrangement in a North American residence supplied with both 115 V
and 230 V rms, since if IVan I = IVII = 115 V rms, then IVabl = 12 Vd = 230 V rms. (ZI + Z2 + Z3)(laA + IbB ) + Z3(laA + IbB ) = 0
Let us now consider the source of Fig. 11.3(a) loaded with two identical loads or (11.5)
both having an impedance Z1. as shown in Fig. 11.3(b). The currents in the lines aA
Since by KCL the left side of (11.5) equals -InN, the neutral current is zero. This is a
a laA consequence of the symmetry of Fig. 11.4.
A
n InN
n~---~- _ _~
nt----I.)Zi~)::.I__--... N
b
B
(a) (b)
B
FIGURE 11.3 (a) Single-phase three-wire circuit; (b) single-phase three-
wire circuit with two identical loads. FIGURE 11.4 Symmetrical single-phase three-wire system.
448 Chapter 11 Three-Phase Circuits Section 11.1 Single-Phase Three-Wire Systems 449
If the symmetry of Fig. 11.4 is broken by having unequal loads at terminals A-N
and N-B or unequal line impedances in lines aA and bB, there will be a neutral current.
For example, let us consider the circuit of Fig. 11.5. Note that the
loads across terminals AN and NB are not equal, and the symmetry
of Fig. 11.4 is broken. The mesh equations, in vector-matrix form, r-----ob
lQ
A are
42+ }
39 Q [ -2 ][Il][IISLQ]
-2 SO + } 12 l1SLQ .~~~~_--~n
115Loo v rms
:J jQ
The determinant is
fl = (42 + j)(SO + j) - 4 = 2097/2.S1
n
and solving by matrix inversion yields
(a) (b)
47 Q
][1]
115Loo Vrms
:J jQ
[ II] = l1SLQ [SO +}
12 fl
so the neutral current
zero.
+2
InN
+2
42+} 1 -
[2.8S/-1.4P]
2.41/-1.21
= 12 - II = 0.44/177 A rms and is not
fiGURE 11.6 Two representations of a wye-connected source.
The voltages Van, V bn , and Ven between the line terminals and the neutral terminal
are called phase voltages and in most cases we shall consider are given by
450 Chapter 11 Three-Phase Circuits Section 11.2 Three-Phase Wye-Wye Systems 451
v"b = - \),n v"b
11-----------------
,,
,,
,, ,,
,
,,
\<: -- - -- - -- - -- - - -",-1fL-------";.-'
v"a = - v"n
,,
overtime
,
,
vbn "~c=-~n
,
(a) (b) \ ,,
,,
FIGURE 11.7 (a) Positive and (b) negative phase sequences. ,
,
choice of the tenninallabels, a, b, and c. Thus without loss in generality we consider
only the positive sequence.
By Fig. 11.7(a), the voltages in the abc sequence may each be related to Van. The FIGURE 11.8 Phasor diagram showing phase and line voltages.
relationships, which will be useful later, are
Vbn = Van/-l20
(11.9) Examining Fig. 11.8, the line voltages are also a balanced set, whose magnitudes exceed
Ven = V an /120
the phase voltages by a factor of v'3 and which are 30 out of phase with them (line
The line-ta-line voltages, or simply line voltages, in Fig. 11.6 are Vab, V be, and voltage Vab leads phase voltage Van by 30, Vea leads Ven by 30, and Vbe leads Vbn
V ea , which may be found from the phase voltages. For example, by 30). These results also may be obtained graphically from the phasor diagram shown
Vab = Van + Vnb in Fig. 11.8.
Let us now consider the system of Fig. 11.9, which is a balanced wye-wye three-
phase four-wire system if the source voltages are given by (11.6). The term wye-wye
applies since both the source and the load are wye-connected. The system is said to
be balanced since the source voltages constitute a balanced set and the load is balanced
(each phase impedance is equal, in this case, to the common value Zp). The fourth wire
= J3 v'3 +J 1)
Vp ( 2 2
= J3 Vp /30 a
In like manner,
V be = J3 Vp/-90
Vea = J3Vp /-210
b
If we denote the magnitude of the line voltages by VL, we have
(lUOa)
c
and thus
(ll.lOb) fiGURE 11.9 Balanced wye-wye system.
452 Chapter 11 Three-Phase Circuits Section 11,2 Three-Phase Wye-Wye Systems 453
is the neutral line n-N, which since it carries no current in this symmetric circuit, may arbitrary ZN, no term in any equation is changed, since every term containing Zn is
be omitted to form a three-phase three-wire system. multiplied by zero (InN or VnN ). By this argument we see that in any circuit containing
Applying KVL around loops containing the neutral wire, the line currents of an impedance Z that draws no current, Z may be changed arbitrarily without affecting
Fig. 11.9 are evidently any current or voltage. In particular, we may replace Z by a short circuit (Z = 0) or an
open circuit (Z = 00).
laA = Van
Zp
As an example, let us find the line currents in Fig. 11.10. We may
I bB = V bn = V an/-120
- - - - = laA/-120 (11.11) combine the 1-Q line impedance and (3 + j3)-Q phase impedance
Zp Zp to obtain
Vcn V an(120
Icc = - = = laA(120 Zp = 4 + j3 = 5/36.9 Q
Zp Zp
as the effective phase load. Since by the foregoing discussion there
The last two results are a consequence of (11.9) and show that the line currents also form
is no neutral current, we have
a balanced set. Therefore, their sum is
I _ 100LQ _ _
- InN = laA + IbB + Icc = 0 aA - 5/36.9 - 20/ 36.9 Arms
confirming that the neutral carries no current in a balanced wye-wye four-wire system. The currents form a balanced, positive sequence set, so we also have
In the case of wye-connected loads, the currents in the lines aA, bB, and cC are
also the phase currents (the currents carried by the phase impedances). If the magnitudes IbB = 20/-156.9 Arms, IcC = 20/-276.9 Arms
of the phase and line currents are Ip and h, respectively, then h = Ip, and (11.11)
This example was solved on a "per-phase" basis. Since the
becomes
impedance in the neutral is immaterial in a balanced wye-wye sys-
laA = h/-e = Ip/-e FIGURE 11.10 Balanced system with line tem, we may imagine the neutral line to be a short circuit. We may
(11.12) impedances. do this if it contains an impedance or even if the neutral wire is
IbB = h/-e - 120 = Ip/-e - 120
not present (a three-wire system). We may then look at only one
Icc = h/-e + 120 = Ip/-e + 120 a A
phase, say phase A, consisting of the source Van in series with ZL
and Zp, as shown in Fig. ll.ll. (The line nN is replaced by a
where e is the angle of Zp.
short circuit.) The line current laA, the phase voltage laAZp, and the
The average power Pp delivered to each phase of Fig. 11.9 is
voltage drop in the line laA Z L may all be found from this single-
Pp = VpIpcose phase analysis. The other voltages and currents in the system may
(11.13) be found similarly, or from the previous results, since the system is
= I; Re(Zp) n N balanced.
and the total power delivered to the load is
fiGURE 11.11 Single phase for a per-phase
P =3Pp analysis.
The angle e of the phase impedance is thus the power factor angle of the three-phase
As another example, suppose that we have a balanced wye-connected
load as well as that of a single phase.
Suppose now that instead of assuming that they are perfect conductors, a line source, having line voltage VL = 200 V rms that is supplying a
impedance ZL is used to model each of the lines aA, bB, and cC and that a neutral balanced wye-connected load with P = 900 W at a power factor of
line impedance ZN, not necessarily equal to ZL, is inserted in line nN in series with 0.9 lagging. Let us find the line current h and the phase impedance
Zp. Since the power supplied to the load is 900 W, the power
the phase impedances. The two sets of impedances may be combined to form perfect
supplied to each phase is Pp = 9~O = 300 W, and from
conducting lines aA, bB, and cC with a load impedance Zp + ZL in each phase.
To determine the effect of Z N, consider the full set of KCL, KVL, and element laws Pp = VpIp cose
for this circuit. The currents and voltages solving these equations for the case ZN = 0
include InN = 0, since the circuit in this case is balanced. But these same currents and we have
voltages must solve the circuit equations for any Z N, not just Z N = O. This follows from
the fact that substituting the solutions for the case ZN = 0 into the equations containing
454 Chapter 11 Three-Phase Circuits Section 11.2 Three-Phase Wye-Wye Systems 455
Therefore, since for a wye-connected load the phase current is also
vLoo
the line current, we have A
3v'3
h = Ip = 2(0.9) = 2.89 A rrns
The magnitude of Zp is given by
Vp 200/v'3
=40 Q
IZpl = Ip = 3v'3/(2)(0.9)
and since e = cos- 1 0.9 = 25.84 is the angle of Zp, we have vLoo
,,
----------,
Zp = 40/25.84 Q Load
EXERCISES immediate observation is that the three-phase circuit has a higher element count: more
wires, more phase sources, and more phase impedances. Simplicity favors the single-
A
11.2.1. Vab = 100L!) V nns is a line voltage of a balanced wye-connected phase circuit. But we will look beyond element count in comparing the relative virtues
three-phase source. (a) If the phase sequence is abc, find the phase voltages. of single- and three-phase systems.
(b) Repeat for the acb phase sequence. Consider the line losses incurred by each system while producing the same amount
Answer (a) 57.7/-30, 57.7/-150, 57.7/+90 V nns; of power at the same voltage level and power factor. In the single-phase case, InN = 0,
(b) 57.7/-30,57.71+90,57.7/-150 V nns and replacing ZN by a short circuit in Fig. 11.12(a) gives
11.2.2. In Fig. 11.9 the source voltages are determined by Exer-
B cise 11.2.1 (a) and the load in each phase is a series combination of a 30-Q re- I _ VLQ
sistor, a 500-{tF capacitor, and a 0.25-H inductor. The frequency is w = 200 aN - ZI + ZL
rad/s. Find the line currents and the power delivered to the load. V2 COSe
Answer 1.15/-83.1; 1.15/-203.1; 1.15/36.9 A nns; 120 W PaN = IVaN IIIaN I cos e = - - - - - ,
11.2.3. Show that if a balanced three-phase three-wire system has two bal- IZ1+ ZLI
anced three-phase loads connected in parallel, as shown, the load is equiv- where cos e is the power factor of the series linelload impedance. The total power
c alent to that of Fig. 11.9 with delivered is
Z1 Z2 2V2 COSe
Zp=---
ZI +Z2 PaN + PbN = 2PaN = - - - - (11.14)
IZ1+ ZLI
11.2.4. If, in Exercise 11.2.3, ZI = 4 + j3 Q, Z2 = 4 - j3 Q, and the
line voltage is VL = 200,J3 V nns, find the current h in each line. In the three-phase case of Fig. 11.12(b) again, the neutral current In' N' = 0 and the
Answer 64 A nns per-phase calculation is as above. The total power delivered by the three-phase source is
EXERCISE 11.2.3 3V 2 cose
Pa'N' + Pb'N' + Pc'N' = 3Pa'N' = - - - - (11.15)
IZ 3 + ZLI
The powers delivered by the single- and three-phase sources are equal if (11.14) and
(11.15) are equal, or
3
- (11.16)
It is interesting to compare the relative merits of the single-phase three-wire circuit 2
of Section ILl and the three-phase circuit of Section 11.2 for delivering power to a For equal power delivery, the series phase impedance in the three-phase circuit, Z3 + ZL,
load. Figure 11.12 shows the two circuits to be considered. Examining this figure, one is larger than the single-phase impedance by half. Since the phase voltages are the same,
456 Chapter 11 Three-Phase Circuits Section 11.3 Single-Phase versus Three-Phase Power Delivery 457
the rms current in the single-phase lines must be larger by the same ratio: Examination of the instantaneous power waveforms associated with these two cir-
IlaNI V /IZI + ZLI 3 cuits reveals another difference relevant to many practical applications. From Fig. 11.12(a),
(11.17) in the single-phase case
(l1.21a)
(11.18) and
11.3.1. A three-phase source and load 100 Ian apart are connected by lines
with impedance 0.01 Q/km, as shown in the figure. What is the efficiency
of this system, defined as power delivered to the load divided by total power
produced by the source? At what distance is the efficiency 90%?
Answer 0.95; 211 Ian
P3T(t) = ( J2 y )2 190
IZ3 + ZLI
x Re(Z3) [cos 2 (wt + e) + cos2 (wt + e + 120) + cos2 (wt + e - 120)]
EXERCISE 11.3.1
2y2 Re(Z3)
v, 11.3.2. The figure shows a single-phase system of the same source voltage
IZ3 + ZLI 2 IVsl as in Exercise 11.3.1. For 100-Ian separation, for which RL would the
X [~ + cos 2(wt + e) + cos 2(wt + e + 120) + cos 2(wt + e - 120)] same power be delivered to the load as delivered by the three-phase system
of Exercise 11.3.1 to its three-phase load? What is the efficiency of this
(11.22)
system? (See Exercise 11.3.1.)
The cosine terms in (11.22) are evidently time functions representing a balanced set of Answer 11.95 Q or 0.0837 Q; 0.923 for RL = 11.95 Q; 0.0772
phasors. Since a balanced set of phasors sums to zero, so must these three terms. The for RL = 0.0837 Q
remaining term is constant, and thus P3T(t), the total instantaneous three-phase power 11.3.3. Consider a motor powered by a single-phase 50-Hz source. What
delivered to the load, is constant, as shown in Fig. 11.13(b). is the period of the vibrational motion the motor will experience? Repeat for
We conclude that single-phase power is delivered unevenly, varying periodically a balanced three-phase source and an unbalanced three-phase source both at
50 Hz.
between zero and its maximum, while three-phase power is constant over time. Suppose
EXERCISE 11.3.2 Answer 10 ms; no vibrational motion; 10 ms
that we were delivering power to an electric motor. Since the torque produced varies with
the power delivered to it, a single-phase motor shakes as the power delivery fluctuates
periodically, while a three-phase motor "pulls" as hard each instant of time as every other,
with no induced vibrations. Or consider powering fluorescent lights, whose brightness
varies with instantaneous power delivery. A pair of single-phase three-wire fluorescent
fixtures would flicker periodically, while three three-phase fixtures mounted together
would illuminate without flicker. Another method of connecting a three-phase load to a line is the delta, or ~, connection,
Thus not only does three-phase power delivery have an efficiency advantage over A balanced ~-connected load (with equal phase impedance Zp) is shown in Fig. 11.1S(a),
single phase, but a smooth power delivery advantage as well. It is no surprise that three- in a way that resembles a ~, and in an equivalent way in Fig. 11.IS(b). If the source is
phase is favored the world over for delivery of bulk power. Only where small amounts y- or ~-connected, the system is a y-~ or ~-~ system.
of power and small fractional horsepower motors are sufficient is single-phase power the An advantage of a ~-connected load over a Y-connected load is that loads may be
method of choice. This is the case for most residential wiring, for which the simplicity added or removed more readily on a single phase of a ~, since the loads are connected
and low element weight of single-phase circuits are decisive. Dc is sometimes used to directly across the lines. This may not be possible in the Y connection, since the neutral
transport power over long distances, but mUltiphase ac is dominant in almost all power may not be accessible. Also, for a given power delivered to the load, the phase currents
delivery systems. in a ~ are smaller than those in a Y. On the other hand, the ~ phase voltages are
(a) (b)
higher than those of the Y connection. Sources are rarely ~-connected, because if the
voltages are not perfectly balanced, there will be a net source voltage, and consequently
a circulating current, around the delta. This causes undesirable heating effects in the
generating machinery. Also, the phase voltages are lower in the Y-connected generator, FIGURE 11.16 Phasor diagram for a delta-connected load.
and thus less insulation is required. Obviously, systems with ~-connected loads are
three-wire systems, since there is no neutral connection. The current in line aA is
From Fig. 11.16 we see that in the case of a ~-connected load the line voltages
are the same as the phase voltages. Therefore, if the line voltages are given by (11.1Ob),
as before, the phase voltages are which after some simplification is
V CA = Vd150 (11.23) laA = ,J3 Ip/-e
where The other line currents, obtained similarly, are
lAB = -V AB = Ip/30 - e
Zp
(11.27)
V BC
I Bc =- = Ip/-90 - e (11.25)
Zp
V CA and the line currents are thus
ICA =- = Ip/150 - e
Zp
Icc = h/-240 - e (11.28)
where the rrns phase current magnitude is
Thus the currents and voltages are balanced sets, as expected. The relations between line
I _ VL (11.26)
and phase currents for the ~ -connected load are summed up in the phasor diagram of
P - IZpl Fig. 11.16.
EXERCISES
11.4.1. Solve Exercise 11.2.2 if the source and load are unchanged except
that the load is .6.-connected. [Suggestion: Note that in (11.23), (11.25),
and (11.28), 30 must be subtracted from every angle.]
Answer 2-vG/ -83.1 0; 2-vG/156.9; 2-vG/36.9 Arms, 360 W
11.4.2. A balanced .6.-connected load has Zp = 4 + j3 n, and the line
voltage is VL = 200 V rms at the load terminals. Find the total power
delivered to the load.
Answer 19.2 kW
11.4.3. A balanced delta-connected load has a line voltage of VL
100 V rms at the load terminals and absorbs a total power of 4.8 kW.
If the power factor of the load is 0.8 leading, find the phase impedance.
Answer 4 - j3 n
FIGURE 11.17 Wye- and delta-connected loads in parallel.
Za = ZabZca
Zab + Zbc + Zca
Zb = ZbcZab
(11.32)
Zab + Zbc + Zca
(b) Z _ ZcaZbc
\ c - Zab + Zbc + Zca
where the Z's are the reciprocals of the Y's of Fig. 11.18. The rule is as follows: The
impedance of an arm of the wye is equal to the product of the impedances of the adjacent
arms of the delta divided by the sum of the delta impedances. (By adjacent here, we mean
"on each side of and terminating on the same node as." For example, in the superimposed
drawing of the Y and ~, Fig. 11.18(c), Za lies between Zab and Zca and all three have
a common terminal A. Thus Zab and Zca are adjacent arms of Za.)
The Y-~ conversion formulas (11.31) and (11.32) do not require balanced loads
(e)
on even three-phase circuits. They are often of considerable use in simplifying general
fiGURE 11.18 (a) Wye connection; (b) delta connection; (c) the two circuits containing impedances that are neither in series nor parallel, as illustrated in the
superimposed. next example.
Equating coefficients of like terms in these equations and (11.30), we have the Y-~ 2Q 2Q
a
transformation:
I
---7>
jQ
-j2Q -j2 Q
(11.31) +
IlSLoo Vnns +
6Q
lQ
4
+jQ
If we imagine the Y and ~ circuits superimposed on a single diagram as in
Fig. 11.18(c), then Ya and Yb are adjacent to Yab, Yb and Yc are adjacent to Ybc, and so c c
on. Thus we may state (11.31) in words, as follows: The admittance of an arm of the (a) (b)
delta is equal to the product of the admittances of the adjacent arms of the wye divided
by the sum of the wye admittances. fiGURE 11.19 Using y-~ impedance conversion to simplify a circuit.
466 Chapter 11 Three-Phase Ci rcu its Section 11.5 Wye-Delta Transformations 467
Ya = - }I, Yb = }2, and Ye =4 S. Then Ya + Yb + Ye = 4 +} Thus
and
115LQ 115LQ
1= - Z - = = 40.7/-25.8 Arms
y _ (- }I)(j2) _ _2_ 2.54 + }1.23
ab - 4+} - 4+}
The use of Y-Cl conversion in this unbalanced single-phase circuit
}8 permitted us to bypass writing and solving three complex equations
Ybe = 4 +} in three unknowns.
-4}
Yea = 4 +} If the load is balanced, the Y-Cl conversion rules simplify greatly. Setting Ya =
Yb = Ye = Yy in (11.31), where Yy is the common phase admittance of the Y -connected
By Fig. 11.I9(b) we see that after this Y -Cl transformation, Yab balanced load, we see from (11.31) that Yab = Ybe = Yea = Y,:;, the common admittance
is in parallel with the admittance 1/(2 - }2), leading to a parallel of the Cl-connected load, or
equivalent admittance of (Yy)2 1
Y,:; = - - = -Yy (11.33)
2 1 8 - 3} 3Yy 3
4+} + 2- }2 = to - 6} or, inverting, we have
12Loo Vrms
n
11SLoo v rms
'----..... c
FIGURE 11.20 Equivalent of Fig. 11 .19(b) with impedances shown. FIGURE 11.21 Circuit for Example 11.8.
120L120 Y rms
The analysis of three-phase networks presented previously has been restricted to balanced
c
systems whose solutions can be expressed in terms of a single phase. SPICE is easily
used for both balanced or unbalanced systems when applied to the entire network. SPICE
makes no distinction between single- and three-phase circuits. FIGURE 11.22 Balanced Y-Y system for SPICE analysis.
470 Chapter 11 Three-Phase Circuits Section 11.6 SPICE and Three-Phase Circuits 471
Also, IM (VAN) = IM (VBN) = IM (VOn 3.033 Arms; 11.6.2. Repeat Exercise 11.6.1 if the 2-Q neutral line between nodes 0
IP(VAN) = 107.7, IP(VBN) = -12.3, IP(VCN) = -132.3 and 10 is removed.
forms a balanced set of phase currents. The magnitude of the neutral Answer 204.9/32.8 V; 5.33/82.41 A
current is computed to be IM (RLOSSN) = 2.08 X 10- 15 Arms,
very close to its exact value of zero.
that if the neutral wire were removed, conditions would change. connected load with impedances Z\ = 3Zp .
This is unlike the balanced case in which INn = 0, and this wire A Y-~ system is best analyzed by first performing a ~- Y load transformation, then
can be removed without altering any output. Removing the neutral using per-phase calculations.
wire requires only removing (or "commenting out" by adding an SPICE has no special routines for three-phase circuits. It does not recognize balanced
asterisk in the first column) the RLOS SN element line. With the circuits nor do per-phase calculations.
neutral wire removed, rerunning SPICE, the currents have changed
to IAa = 2.83/107.7 , IBb = 2.99/ -10.7 , and Icc = 2.99/ -134.
Note that the degree of imbalance caused by this modest (10%) PROBLEMS
source imbalance is itself relatively small. Assuming balanced cir- 11.1. If in Fig. 11.3 Van = Vnb = 100LQ V rms, the
cuit conditions will not lead to large errors in the presence of minor impedance between terminals A-N is 10/60 Q, and that
source, line, or load imbalances. between terminals N-B is 10/-60 Q, find the neutral
current InN.
EXERCISES
11.2. For the symmetrical single-phase power delivery cir-
11.6.1. Find the line voltage and phase current for the load of phase A of cuit shown, derive an expression for the loss fraction I,
the system of Fig. 11.22 if the load of phase C is short-circuited (called a defined as the ratio of the real average power dissipated
phase fault). by the line impedances Z \, to the total real average power
Answer 204.9/32.8 V; 4.07/90.54 A produced by the sources, in terms of Z\ and Z2. FIGURE 1'11.2
in both circuits if this Zl is used in the single-phase case. 0.5 Q, and Zp = 8 + j2 Q. Find the line currents laA, IbR,
FIGURE 1'11.3 n N and lec and the complex power delivered to the load. Do
11.21. Why is the frequency of mechanical vibration of a not use wye-delta transformation.
o 11.4. Find the real average power supplied by each of two single-phase motor twice that of its electrical supply?
Iill!l voltage sources in the circuit of Problem 11.3. 11.22. Consider Fig. 11.12 with V = 4.4 kV rms, each
D 11.5. Determine the currents laA, IbR, and InN, and sketch source operating at a power factor of 1.0, and each cir-
Iill!l in a phasor diagram. The source frequency is 60 Hz. cuit supplying 660 kW to its load. Sketch phasor diagrams
for the line currents laA, IbR; la' A', Ib'R', le'c' and the to-
1Q FIGURE 1'11.11 tal instantaneous power absorbed by the single-phase and
three-phase loads as a function of time.
~ 11.12. Repeat Problem 11.11 for negative phase sequence. 11.23. In Fig. 11.15 the source is balanced with positive
l1sLoo V rms + 0.01 F phase sequence and Van = 100LQ V rms. If the phase
D 11.13. If in the balanced three-phase circuit shown in impedance is 3vS/30 Q, find the line current and the
0.2H
Iill!l Problem 11.11, we have phase sources of magnitude power delivered to the load.
4400 V rms, Zz = 0.01 QIkm, and Zp = 20 + j4, find 011.24. In Fig. 11.15 the positive sequence system has
l1sLoo V rms + om F
an equation relating line length in kilometers to rms line IlliU Van = 200LQ V rms. Find Zp if the source delivers 2.4 kW
current, and sketch as a graph. at a power factor of 0.8 lagging.
11.14. A balanced wye-wye system with 60-Hz phase 11.25. In the Y-b. system shown, the source is positive se-
sources of 1200 V rms and no line losses delivers 3.00 kW quence with Van = 100LQ V rms and the phase impedance
to a load consisting of a series combination of a lO-Q re- is Zp = 3 - j4 Q. Find the line voltage VL, the line current fiGURE 1'11.29
FIGURE P11.5 sistor and L-henry inductor. Find L. h, and the power delivered to the load.
p. 11.15. What. is the magnitude of the source phase voltage
D 11.6. Determine the Thevenin equivalent of the circuit of Iill!l needed to dehver 100 W at 60 Hz to a small three-phase Van
Iill!l Problem 11.5 at the terminals AB when the inductor is motor whose Zp is equivalent to a series combination of a a A
removed. What value of inductance connected between A 12-Q resistor and lO-mH inductor? -+
p'11.30. A balanced wy~-connected source with line volt-
and B causes the complex power delivered by the sources D 11.16. A balanced wye-wye three-phase system with Iill!l age
115 V rms supphes power to a balanced delta-
to be purely real? Iill!l phase voltage of 115 V rms supplies power at a power connected load with phase loads Zp each consisting of
11.7. In Fig. 11.9 a balanced, positive-sequence source factor of 0.85 lagging. Find the complex power supplied to Vbn a parallel combination of a 75-Q resistor, 2.5-mH induc-
b tor, and lO-f1,F capacitor. Find the complex power de-
has Vab = 120LQ V rms and laA = 101-60 A rms. Find 0 the three-phase circuit if the line current is 2.3 Arms. n -+
Zp and the power delivered to the three-phase load. 11.17. A balanced Y-Y three-wire, positive-sequence sys- livered to the load and the power factor at 60 Hz and at
11.8. A balanced three-phase Y -connected load draws tem has Van = 200LQ V rms and Zp = 3 + j4 Q. The lines 400 Hz.
1.2 kW at a power factor of 0.6 leading. If the line voltages each have a resistance of 1 Q. Find the line current h, the g 11.31. For the circuit of Problem 11.29, suppose V~n =
are a balanced 200-V rms set, find the line current h. power delivered to the load, and the power dissipated in the e 1ill!l1200LQ, ZZ = 0 Q, and the three-phase source dehvers
11.9. A balanced Y-Y system with Zp = 3vS/30 Q lines. -+~~~--------~~------~ +2 kvars of reactive power to the load. Find Zp and the
c real average power in watts delivered if the power factor is
delivers 9.6 kW to the load. Find the line voltage VL and 11.18. A balanced Y -connected source, Van = 200LQ
the line current h. V rms, positive sequence, is connected by four perfect FIGURE 1'11.25 0.92 lagging.
120LQ Y nus
C
FIGURE 1'11.39
120 /-120 Y rms \ FIGURE 1'11.46
FIGURE 1'11.43
11.47. For the circuit of Problem 11.46, suppose the lines
are protected by circuit breakers set to trip when a line
11.40. Find the equivalent delta- and wye-connected loads current laA, hB, or IcC reaches 100 A. If the 100-Q phase
if ZI = -j30 Q and Z2 = 60+ j15 Q. impedance between A and C were replaced by an R-Q
11.44. Find the total power delivered to the three-phase resistor, for what approximate value of R will a circuit
load by the unbalanced source. All resistors are 10 Q. breaker trip? Solve using SPICE. Guess at a trial value for
FIGURE 1'11.33 R, reduce it if the limit is not exceeded and increase if it
Check using SPICE.
is. Find R to two significant digits.
Chapter Contents
III!I 12.1 The s-Domain III!I 12.5 Solving Integrodifferential
Laplace transformation L is defined as a rule L[f(t)] = F(s) that associates with the
III!I 12.2 Singularity Functions Equations
time function f(t) a function F(s), of a complex variable s, which is given by
III!I 12.3 Other Transform Properties and II1II Summary
Pairs l1li Problems
l1li 12.4 Partial Fraction Expansion F(s) = 1~ f(t)e- st dt (12.l)
F(s) is called the Laplace transform of f(t), and f(t) the inverse Laplace transform of
F(s). Together, f(t) and F(s) are called a Laplace transform pair. The complex plane
over which s ranges and F(s) takes value is called the s-domain.
Several comments are in order with respect to the definition (12.1). Note first that
the lower limit is specified as 0-. The quantities 0- and 0+ were introduced in Chapter 5
as times just before and after t = O. The domain of integration in (12.1) begins just to
the left of t = 0 and continues tot = 00. For ordinary functions f(t), the lower limit 0-
can freely be replaced by either 0 (or even 0+) without changing the value of the integral,
In Chapters 8 to 11 our goal was to determine the behavior of ac steady-state circuits. since there can be no nonzero area under any finite integrand in the infinitesimal interval
We began with basic circuit analysis equations containing time integrals and derivatives, t = 0- to t = 0+. However, there is a class of decidedly unordinary functions called
or integrodifferential equations. While solving these sets of coupled integrodifferential the singularity functions to be introduced in the next section. In a singUlarity function,
equations directly proved somewhat tedious, we discovered an indirect strategy that paid area may "pile up" right at a single time instant such as t = 0, and we choose to include
great dividends. By first transforming the integrodifferential equations into the phasor this area in the definition of the Laplace transform by setting the lower limit to 0-.
domain, we managed to replace the derivative and integral operations of calculus with the Since values of f(t) for t < 0- do not enter into (12.1), the Laplace transform of
much simpler multiplication and division operations of ordinary algebra. After solving a time function f(t) does not depend on values of f(t) prior to 0-. In circuits with tran-
the circuit equations in this more convenient algebraic phasor form, the results were then sients, there is good justification for making transforms depend only on values of currents
transformed back into the original sinusoidal time-domain format without difficulty. The and voltages after a specific time. In transient circuits there is some initial time, which we
strategy of transformation of circuit equations to phasor form proved effective in reducing usually call t = 0, before which the detailed time histories of the currents and voltages
computations and brought us a conceptual bonus as well. It permitted extension of the are not of interest. For example, a circuit may have no sources acting until t = 0, at
idea of resistance R into impedance Z(jw), which forms the basis for a unified treatment which time a voltage source is switched in and a transient response begins. Before t = 0,
of all RLC elements in ac steady state. nothing of interest is happening. Our choice of lower limit makes (12.1) well suited to
Our goal in this chapter is to introduce a tool that will permit us to generalize study such circuits, making the Laplace transform properly indifferent to behavior prior
this successful strategy beyond the narrow confines of ac steady state. That tool is to the initial time of interest. The earlier behavior will be summarized effectively by
Laplace transformation. Clearly, such a tool is needed. The same sets of coupled specifying initial conditions at time 0-. Equation (12.1) is sometimes referred to as the
integrodifferential equations that proved inconvenient in ac steady-state problems also "one-sided" Laplace transform to distinguish it from the "two-sided" Laplace transform
govern the behavior of circuits containing nonsinusoidal sources and circuits in which defined with lower limit equal to -00. We will have no need for the two-sided Laplace
transients are to be determined. We cannot transform these circuit problems using phasors, transform in this book and will take Laplace transform to be synonymous with one-sided
which apply in the ac steady-state case. The Laplace transform supplies what is needed Laplace transform as defined in (12.1).
to convert from integrodifferential to algebraic equations and to generalize the important By existence of the Laplace transform F(s) of a given f(t), we mean that the
notion of impedance to all linear circuit problems, in all cases, not just ac steady state. defining integral (12.1) exists for at least some values of its argument s. That is, there
We first define the Laplace transform and the s-domain in which it takes value. exists a region of the s-domain, called the region of convergence, for which the integral
The singularity functions are next introduced, a set of functions useful in describing in (12.1) is well defined. A function f(t) whose Laplace transform F(s) exists is said to
transient behavior. In succeeding sections, the Laplace transforms of common time be Laplace transformable. Sufficient conditions for Laplace transformability of f(t) are:
functions are calculated and properties of the Laplace transform explored. Return back
from the s-domain to the time domain (inverse transformation) by partial fractions is next 1. f(t) is a piecewise continuous function; that is, f(t) is continuous except perhaps
considered, and the chapter concludes with the solution of coupled integrodifferential at a set of isolated finite discontinuities [points {til, where f (ti +) and f (ti -) are
equations by this method. distinct finite numbers].
480 Chapter 12 The Laplace Transform Section 12.1 The s-Domain 481
2. f (t) is of exponential order; that is, there is some real exponential function M errt we have determined the Laplace transform of f(t) = e-alu(t) to be
that for all t sufficiently large is greater in magnitude than f(t):
If(t)1 ::s Meat
Fortunately, almost all functions of practical interest satisfy these conditions and are
therefore Laplace transformable. * Those few that are not contain infinite discontinuities,
(12.2)
F(s) = - -
s+a
1
(12.4) I
such as f(t) = 1/t 2 , and thus are not piecewise continuous or grow too fast to be of
A sketch of the magnitude of F(s) over the s-domain is shown in Fig. 12.1(b). The
exponential order, such as f(t) = tl. While they are of interest mathematically, we
region of convergence for F(s) is the portion of the s-domain Re(s) > -a. Note that
are not likely to encounter such exotic functions in practical circuits, and we will not
consider non-Laplace-transformable functions further. the function f(t) = e- at has the same F(s) as that for e-atu(t) calculated above, since
these two functions disagree only in negative time. Behavior in negative time is never
As a first Laplace transform calculation, consider the real function of time f(t) =
accounted for by the one-sided Laplace transform.
e-alu(t) shown in Fig. 12.1(a). u(t) is the unit step function introduced in Section 6.6.
\ With a = 0, this result (12.4) reveals a particularly useful transform pair: the
Recall that multiplication by the unit step function has the effect of forcing the product
to zero for t < O. Using (12.1), L[f(t)] = F(s) is Laplace transform of the unit step function f(t) = u(t) is F(s) = l/s, or
482 Chapter 12 The Laplace Transform Section 12.1 The s-Domain 483
Example 12.1 suggests a general property of Laplace transforms that will greatly aid equations. These properties, beginning with the linearity property above, are summarized
in their calculation. Let f(t) = cllt(t) + c2!2(t). Then, by the defining integral (12.1), in Table 12.1. Table 12.2 gathers together all the important transform pairs we derive
throughout this chapter. These tables are located in Section 12.3, and expanded versions
F(s) = 1~ [cdl (t) + c2!2(t) Je- st dt of these tables are found on the inside of the back cover and its facing page.
Definition (12.1) of the Laplace transform permits us to move from the time-domain
= CI roo It (t)e- st dt + roo h(t)e-st dt
C2
description f(t) of a time function to its s-domain description F(s). The relation that
Jo- Jo- explicitly defines passage back from the s-domain to the time domain is the Laplace
inversion integral
Recognizing the two integrals as FI(s) and F2(S), we have the linearity property.
L(coshat) = -
2
1( 1 + -+1)
--
s- a
-
s a
=- s-
s2 - a 2
.
L(slnhat) = -1( -1- + -1-) = -a -2
2 s- a s +a s2 - a
2 3 4 7
In subsequent sections of this chapter we introduce other properties that will be
helpful in finding Laplace transforms and their inverse transforms and solving circuit EXERCISE 12.1.2
484 Chapter 12 The Laplace Transform Section 12,1 The s-Domain 485
12.1.3. Is J(t) = t 2 e+ t Laplace transformable? If so, find suitable M u(t)
and a in (12.2).
Answer Yes; for instance, M = I and any a > +1. MeO't >
IJ(t)1 for all
t 2
->-- o
lnt a-I
(a)
r(t) p(t)
o 0
Repeated integration and differentiation of the unit step function produce a family of -~-----+~~--~
functions called the singularity functions. In this section we introduce this family, develop
general rules for finding the Laplace transform of derivatives and integrals, and use these
(b) (c)
rules to find the Laplace transforms of the singularity functions. In the application of
Laplace transforms to circuit analysis, singularity functions arise naturally and playa fiGURE 12.2 Unit step, ramp, and parabola.
central role. Indeed, this family of functions is prominent in the study of all manner of
physical systems that are modeled by integrodifferential equations.
The only singularity function we have encountered so far is the unit step u(t). We further integration of tp-e unit parabola yields the unit cubic function, equal to ~u(t),
next consider singularity functions produced by repeated integration of u(t). The first then the unit quartic, hu(t), and so on. Each is a scaled power of t in positive time.
integral of the unit step is referred to as the unit ramp function ret). Since u(t) = 0 for One practical application of this branch of the family of singularity functions is in
t < 0, its integral ret) = 0 for t < O. For t 2: 0, representing piecewise polynomial functions, those functions that are fixed polynomials
over subintervals. This use is shown in the next example and the exercises at the end of
this section.
ret) = t u(i)di = t Idi = t, t2:0 (12.10) Consider the function f(t) of Fig. 12.3, given by
10- 10- 0, -00 < t < 1
Examining Fig. 12.2(b), ret) is zero until t = 0, then "ramps" upward at a constant slope
f(t)=
1 t 2 -2t+l,
-2t+1O,
l::st<3
3::st<6
(12.12)
of unity. Integrating once again, we have the unit parabola function pet). Since ret) is 2, 6::st<00
zero for t < 0, its integral pet) also will be zero for t < 0, and for t 2: 0:
f(t)
pet) = 1t
0-
r(i)di =
1t
0-
t2
idi = - ,
2
(12.11)
4
The unit parabola is sketched in Fig 12.2(c).
-2t+ 10
Note in Fig. 12.2 that the unit step, ramp, and parabola are each zero for t < O. 2 2
Functions produced by further integrations will share this property, as will those produced 0 0
by differentiation, however many times repeated. The singularity functions are all zero
3
for t < O. Thus ret) = tu(t) and pet) = ~t2u(t). Note also that each is infinitely smooth, -2
that is, possesses derivatives of all orders, everywhere except at the point t = O. This is
the singular point of these functions from which their family name derives. Continuing,
fiGURE 12.3 f(t) for Example 12.3.
J(t) = 2p(t - 1) - 2p(t - 3) - 6r(t - 3), t:s6 FIGURE 12.4 Functions approaching the unit step and impulse.
Finally, adding +2r(t - 6) corrects the slope to 0 in the final subin-
terval, and the constant term is corrected from + 10 + 2(-6) = -2 and base 2~ (and thus area always equal to 1) centered at t = O. These functions are
to +2 by adding +4u(t - 6). This is the desired description, valid sketched in Fig. 12.4. Note two characteristics of (l(t) that emerge by noting that (l(t) is
for all t, of J(t) as the weighted sum: the limit of d(t) as ~ goes to zero:
J(t) = 2p(t - 1) - 2p(t - 3) - 6r(t - 3) + 2r(t - 6) + 4u(t - 6) (l(t) = 0 for all t =1= 0 (l2.1Sa)
r
1 A8(t-to)
vet) = -[ret -~) - ret + ~)] (l2.l4a)
d
2~
where ret) is the unit ramp. Let d(t) = (djdt)v(t) be its derivative. Since u(t) =
(djdt)r(t), it follows from (12.14a) that
1
"T I
;:. t
to
d(t) = dt vet) = 2~ [u(t - ~) - u(t + ~)] (l2.14b) (a) (b)
As ~ gets smaller and smaller, vet) approaches the unit step, and its derivative d(t) thus FIGURE 12.5 (a) Unit impulse function 8(t); (b) shifted impulse of area At
approaches the unit impulse (l(t). The unit impulselike d(t) is a pulse with height Ij2~ A8(t - to).
488 Chapter 12 The Laplace Transform Section 12.2 Sillgularity Functions 489
the equation 1/00 = 0 is correct and obvious. However, we must be careful when using Evaluate the generalized derivative (d/dt)[cos tu(t)] and show that
the generalized number 00. For instance, although the equation x/x = 1 is valid for all it obeys the fundamental theorem of calculus; that is,
real numbers x =1= 0, it is not valid for the generalized number 00. Even though 00 is not
zero, 00/00 is not in general equal to 1.
o(t) is to ordinary functions as 00 is to ordinary numbers, an extension that if used
I-00
t d
dr [cos(r)u(r)] dr = cos(t)u(t) (12.18)
in the right equations will yield accurate results. Which are the right equations for o(t)?
To answer this question thoroughly and rigorously would require an extended digression We first note that cos(t)u(t) is not differentiable in the ordinary
into an area of mathematical analysis called the theory of generalized functions. Indeed, sense, since it has a jump discontinuity at t = O. One approach is
the status of the unit impulse was the subject of very lively debate among great thinkers to express this function as the sum of a continuous function whose
for a considerable period of time. Two of the most prominent analysts of the last hundred derivative is an ordinary derivative and a unit step whose generalized
years, Oliver Heaviside (1850-1925), a self-educated English engineer, and Paul Dirac derivative is known to be the unit impulse o(t).
(1902- ), English physicist and Nobel Prize winner, played key roles in developing the
theory and reconciling the controversies. A thumbnail biography of Heaviside introduces \ cos(t)u(t) = f(t) + u(t)
Chapter 13. where f(t) is the continuous function:
Here we limit ourselves to identifying some specific conditions under which the
unit impulse may be used with accuracy and rigor. The generalized function o(t) may
be integrated, as in (12.15b), and evaluated at all times except its singular time, as in f(t) = { ~~S(t) - 1,
t < 0
t2:0
(12.15a). Identifying the unit impulse as the derivative of the unit step (12.13) is correct,
with the understanding that since o(t) is a generalized function, we mean the generalized f(t) has no jumps and is thus differentiable in the ordinary sense,
derivative of the unit step (its ordinary derivative does not exist). Often, for brevity, with derivative equal to zero in negative time and - sin t in positive
we will omit the word generalized, referring to o(t) as simply the derivative of the unit time. Thus
step.
d d d
An immediate consequence of the fact that the unit impulse is the derivative of the - cos(t)u(t) = - f(t) + -u(t)
unit step is that the unit step is the integral of the unit impulse: dt dt dt (12.19)
=- sin(t)u(t) + o(t)
As shown in Fig. 12.6, the derivative has a unit impulse located
(12.16) where the function being differentiated has a jump discontinuity.
Finally, integrating this derivative we have calculated yields
For any upper limit t < 0, there is no area under the impulse within the integral's limits,
and as t passes through zero, the unit area "piled up" at t = 0 is added, so the value of
It -00
d
dr [cos(r)u(r)]d(r) =- It
-00 sin(r)u(r)d(r) + It
-00 oCr) dr
the integral jumps to 1. The function that is zero for t < 0 and 1 for t > 0 is, of course,
the unit step. f(t) = cos t u(t)
df .
A consequence of the fact that o(t) = 0 for all t except t = 0 is that if we have a - =o(t) - smt
+1 dt
product f(t)o(t), where f(t) is an ordinary function continuous at t = 0, then
f(t)o(t) = f(O)o(t)
This follows by comparing the two sides for t not 0 (they are both 0) and for t = 0
(the two sides are identical). If the impulse is time shifted so that its singular point is
t = to, then, by reasoning identical to the above, the continuous function f(t) may again
be replaced by its value at the singular point of the impulse, to:
-1
f(t)o(t - to) = f(to)o(t - to) (12.17)
(a) (b)
This rule, the impulse product rule, will be helpful in simplifying expressions containing
o(t). FIGURE 12.6 Functions for Example 12.4.
490 Chapter 12 The Laplace Transform Section 12.2 Singularity Functions 491
The first integral evaluates to 0 in negative time and cos(t) - 1 in impulse function is 0 in the added interval. Then, applying the sifting
positive time. The second is the integral of 8(t), which is u(t). property (2.20) with singular point to = 10 gives
I t d
-[cos(r)u(r)] dr = [cos(t) - l]u(t)
-00 dr
+ u(t) = cos(t)u(t) 1 +00
-9 Slog(t)8(t - 10) dt = Slog(lO) = S
1-00 +00
f(t)8(t - to) dt
singular point, the unit doublet 8' (t), its derivative, must inherit this property. 8' (t) has
a sifting property,
where f(t) is assumed continuous at t = to. Applying the impulse product rule, we
obtain
1-00 +00
f(t)8'(t - to)dt = - f'(to) (12.21)
which can be verified by integration by parts and use of the impulse sifting property
:00 f(t)8(t _ to) dt = :00 f(to)8(t - to) dt (2.20). Just as the unit impulse sifts out the value of a function at its singular point, the
unit doublet, its derivative, sifts out the derivative there (multiplied by -1).
Taking the constant f (to) out of the integral and noting that the integral is just the total The remaining higher-order singularity functions 8(2) (t), 8(3)(t), ... , are each deriva-
area under this (shifted) unit impulse, which is 1, we arrive at tives of the previous one. Each has a sifting property, selecting out the values of the
higher-order derivatives at the singular point (see Problem 12.47).
SIFTING PROPERTY Having introduced the family of singularity functions, we turn to computing their
1-00 +00
f(t)8(t - to) dt = f(to) (12.20)
Laplace transforms. To do so, we first develop rules for writing down the Laplace
transform of the derivative and of the integral of any function whose Laplace transform
is known.
DIFFERENTIATION
The sifting property (12.20) shows that the effect of the impulse in the integral is to sift
out, or select, the value of the rest of the integrand at its singular point as the value of the L [:t f(t) ] = sF(s) - f(O-) (12.22)
integral. This results holds for all ordinary functions f(t) continuous at to
Evaluate the following integral (log means log base 10): The Laplace transform of a derivative of a function is s times the Laplace transform of
1 +00
S(log t)8(t - 10) dt
the function minus its initial value. Integrating by parts gives
1
-9
The lower limit of the integral may be extended to -00 since the
100
0-
d f(t)e- st dt = f(t)e- st 1
-d
i
00
0-
+s 00
0-
f(t)e- st dt
492 Chapter 12 The Laplace Transform Section 12.2 Singularity Functions 493
The last term is clearly sF(s), and evaluation of the first at infinity yields 0 for all s and differentiating once again yields
within the region of convergence of the transform. Equation (12.22) follows.
INTEGRATION
L [:t22 sinh fU(t)] = sL [dd sinh tU(t)] -
t dt
sinh(t) I
t=O-
~u(O-)
(12.23) ]
=
S2 - 12
We may verify these calculations by differentiating first,
The Laplace transform of the integral of a function is the Laplace transform of the function
divided by s. This may also be verified by an integration of the defining integral by parts ddt coshatu(t) = ~
dt
[~(et + e-t)u(t)]
2
as above. (The reader is asked to do so in Problem 12.19.)
Armed with these rules, we may easily evaluate the Laplace transforms of all I
= 2[(et - e-t)u(t) + (e t + e- t )8(t)]
the singularity functions. We begin with lis, the transform of u(t), derived in Sec-
tion 12.1. Since the unit ramp is the integral of the unit step, by the integration rule = sinhtu(t) + 8(t)
the transform of ret) is R(s) = I/s2. Similarly, the transform of the unit parabola
is P(s) = 1I s3, and so on, for those singularity functions produced by further in- and then transforming.
tegrations. To find the transform of 8(t), apply the differentiation rule with f(t) = 1 s2
u(t): L[sinhtu(t) + 8(t)] = -- + 1= --
s2 - 1 S2 - 1
which, of course, gives the same result. Note that the transforms of
(dldt)coshtu(t) and d 2Idt 2 sinhtu(t) are the same, so the inverse
transforms must be identical. Indeed, these two time functions are
identical, as can be verified by direct calculation.
The transform of the unit impulse is simply the constant 1. The unit doublet 8' (t), be-
ing the derivative of 8(t), has transform s(l) - 0 = s. Similarly, 8(2)(t) has transform
s2, 8(3) (t) has transform s3, and so on. The transforms of the singularity functions are all
EXERCISES
powers ofs.
12.2.1. Find the Laplace transform of 8(t - to) using the sifting theorem.
/(t)
Note that the lower limit on the integral in the sifting theorem is -00. Find
d the Laplace transform of u(t-to) from the defining integral (12.1). Consider
Let us determine the Laplace transforms of dt [cosh tu(t)] and
d2 to < 0 and to ::: 0 as separate cases.
dt 2 [sinh tu(t)], where cosh and sinh are the hyperbolic sine and Answer L[8(t - to] = 0 for to < 0, e- sto for to ::: 0; L[u(t - to] =
cosine functions discussed in Example 12.2. Noting the transforms lis for to < 0, e- sto Is for to ::: 0
2 3 7/2 4 12.2.2. Sketch g(t) = df Idt for the f(t) shown. Express f(t) and g(t)
given there and by the differentiation rule we have
(a) as linear combinations of shifted singUlarity functions. Use Exercise 12.2.1
to find G(s).
L [:t coshtU(t)] = sL[coshtu(t)] - cosh(O-)u(O-)
g(t) 2r- Answer f(t) = u(t-l)-u(t-2)+2r(t-3)-4r(t-~)+2r(t-4);
s2 g(t) = 8(t - 1) - 8(t - 2) + 2u(t - 3) - 4u(t - ~) + 2u(t - 4);
7 G(s) = e- s - e- 2s + 2e- 3s Is - 4e- 7 / 2s Is + 2e- 4s Is
= S2 - 12 "2 4
since u(O-) = O. Similarly,
2
3 -- 12.2.3. Express e(t), the integral of f(t) in Exercise 12.2.2,
e(t) = t f(r:)dr:
L [:t SinhtU(t)] = sL[sinhtu(t)] - sinh(O-)u(O-) -1 ' - - -2 Jo-
(b)
as a linear combination of shifted singUlarity functions. Find E(s).
s
EXERCISE 12.2.2 Answer e(t) = r(t-l)-r(t-2)+2p(t-3)-4p(t- $+2p(t-4);
S2 - 12
E(s) = e- s Is2 - e- 2s Is2 + 4e- 3s Is 3 - 8e 7 / 2s Is 3 + 4e- 4s I s 3
494 Chapter 12 The Laplace Transform Section 12.2 Singularity Functions 495
12.2.4. Compute d 3 /dt 3 [sin 3 tu(t)]. For what values n > 0 do the nth FREQUENCY SHIFT
derivatives of sin3 tu(t) exist as ordinary functions? For what values as
generalized functions?
Answer 6cos 3 tu(t) - 21 sin2 tcostu(t); n = 1,2,3; n > 3
The Laplace transform of a time function made zero for t < 0, then shifted to the right To evaluate the transform of the product coswtf(t), where f(t) is
(delayed), is multiplied by a complex exponential. To verify, we evaluate the transform any Laplace transformable function, note that
of f(t - to)u(t - to):
coswtf(t) = !e jwt f(t) + !e- jwl f{t)
roo f(t _ to)u(t _ to)e- st dt = roo f(t - to)e- st dt
h- ~ Using the frequency shift property with So = jw gives
Let i = t - to and change variables in this integral.
L[cos wtf(t)] = ![F(s - jw) + F(s + jw)] (12.27)
10 00
f(i)e- sto di = e-stoF(s)
f(t) and its time-shifted version f(t - to)u(t - to) are shown in Fig. 12.7.
g(t)
4
f(t)
,
1 3 5 7
o
(a) (b)
fiGURE 12.7 (a) Function f(t); (b) f(t) made zero for t < 0 and delayed
by to. fiGURE 12.8 Time functions for Example 12.7.
496 Chapter 12 The Laplace Transform Section 12.3 Other Transform Properties and Pairs 497
For instance, the transform of f(t), the unit . ramp f(t) = ret) = Thus for a time-compression factor c = 2, G(s) = (2Is)(1 - e- s / 2 ).
tu(t), is F(s) = I/s2, as shown in Section 12.2, and using (12.27), It is interesting to note that as c gets larger and larger, get) ap-
we obtain proaches the unit impulse function oCt) [its area is always 1 for any
c, and get) goes to zero for any nonzero t]. To evaluate its trans-
L[tcoswtu(t)]=-1 [ 1 . 2+( 1 . ]2 form G(s) as c gets very large, we may apply the well-known Taylor
2 (s - JW) S + JW) series expansion of the exponential function,
(12.28)
= ----:------::---:- x2 x3
(s2 + ( 2)2 e X =I+x+-+-+
2! 3!
which the desired Laplace transform for Fig. 12.8(b). Although with x = -sic to get
intended to illustrate the use of the time and frequency shift proper-
ties, both these results required the use of the linearity property as
well. \
G(s) = ~s [1 - (1 - ~ + ~ - ~ +
C 2!c 2 3!c3
00 .)]
S s2
=1--+-- 00
2!c 3!c 2
TIME-FREQU ENCY
As c gets very large, G(s) converges to 1. This is consistent with the
SCALING L[f(ct)] = ~F (~), c>O (12.29) transform of the function that get) is approaching, since we know
that L[8(t)] = 1.
Compressing (expanding) the time scale expands (compresses) the frequency scale by an
T-MUlTIPUCATION
equal factor. To verify, compute the left side of (12.29) with r = ct:
-d
L[tf(t)] = -F(s) (12.30)
L[f(ct)] = fo~ f(ct)e- st dt ds
= 1 00
0-
dr
f(r)e- ST / C -
c
t-multiplication corresponds to negative differentiation in the s-domain. This property
follows by computing the negative derivative of both sides of (12.1) with respect
Taking l/c out of the integral, the result follows. Note that values of c > 1 cor- to s:
respond to time compression and c < 1 to time expansion. For instance, the graph
of the function get) = f(3t) is compressed by a factor of 3 relative to that of f(t). -~F(s) = _~ [ roo f(t)e- st dt]
The real positive constant c in (12.29) is sometimes called the time-compression
ds ds Jo-
factor. = roo tf(t)e-st dt
Jo-
Consider the square pulse f(t) = u(t) - u(t - 1) shown. If we
compress it by a factor of c > 1 and at the same time amplitude- Let us find the Laplace transforms of tne- at for n = 1,2,3, .... In
scale by the same factor c, we get get) = cf(ct). get) for c = 2 Section 12.1 we determined that f(t) = e- at and F(s) = 1/(s + a)
is shown in Fig. 12.9. The transform of f(t), using the time-shift were a transform pair. Then, by the t-multiplication property,
2 g(t) for C =2
property (12.25), is
L[te- at ] = _~_1_ = 1
I--+-..... f(t) 1 e- s ds s +a (s + a)2
F(s) = -S - -
S t-multiplying once again gives
112
Then, with get) = cf(ct), by time-frequency scaling, L[t 2e- at ] = L[t(te- at )]
fiGURE 12.9 f(t) and a compressed f(t) for
Example 12.8.
G(s) = c -[1(1
c
e- s-
- - -
sic sic
/C )] c
= -(1-
s
e- s / C ) =
d 1
ds(s+a)2 (s
2
+ a)3
500 Chapter 12 The Laplace Transform Section 12.3 Other Transform Properties and Pairs 501
pairs may be scanned to find a similar table entry, say get). Properties that may be used This result can also be gained by long division of F(s), as in Prob-
to transform get) to J(t) are then sought in consultation with the table of Laplace trans- lem 12.28.
form properties. Once a sequence of transformations yielding J(t) from get) has been
discovered, the properties table specifies the corresponding changes to G(s) that yield
F(s). The same strategy may be used given a transform F(s) whose inverse transform EXERCISES
J(t) is sought. We conclude this section with examples of the use of these tables.
12.3.1. Use Table 12.2 entries 14 and 15 for pure sine and cosine, together
Find F(s) for the time function J(t) = (t - 3)2e- 2t u(t - 3). The with linearity, to write fonnulas for L[sin(wt + e)] and L[cos(wt + e)].
closest entry in Table 12.2 appears to be the repeated t-multiplied Answer (s sin e +w cos e)/(s2+w2); (s cos e -w sin e)/(s2 +w2)
exponential. With n = 3 and a = 2, 12.3.2. Use time-shift to compute L [~(t -1)2e -(t-l)u(t - 1)].
Answer e- s /(s + 1)2
L [~t2e-2tJ = (s: 2)3 12.3.3. Use the t-multiplication property to find F(s) forf(t)
te- 2t cos4t.
\
Consulting Table 12.1, a time shift by to = 3 yields multiplication Answer ([s + 2]2 - 16)/([s + 2]2 + 16)2
of the transform by e- sto = e- 3s , or
3s - 3
L- I [-l+ 2
s2
s +
4+23 -
s +4
J=~Cos2t+~sin2t
dt 2 S2 + s + 21 3s 3 + 2s + 1
But we know that adding 1 to the transform corresponds to adding
-(3s 3 + 3s 2 + 6s)
o(t) to the time function, or - 3s 2 - 4s + 1
-( - 3s 2 - 3s - 6)
L- I [1-1 + s2s~4 + S2 :4J = o(t) + :t cos2t + ~ sin2t - s+7
The result of the long division is 3s - 3 with remainder -s + 7. Thus
- 2+
[S2 -s+7
or L- I --3J = o(t) - -1 sin2t F(s) = 3s - 3 + - - -
s +4 2 S2 + s + 2
502 Chapter 12 The Laplace Transform Section 12.4 Partial Fraction Expansion 503
Long division converts an improper rational function into the sum of a polynomial SIMPLE REAL POLE For each simple real pole pin F(s), there will be a term AI(s - p) in its partial fraction
(3s - 3 in the example) and a proper rational function. Finding the inverse Laplace expansion, where A = (s - p)F(s)ls=p, and a term Ae+ pt in its inverse transform.
transform of the polynomial part is easy, since all powers of s have singularity func-
tions as their inverse transforms (entry 7 in Table 12.2). For instance, the polynomial Complex poles occur in conjugate pairs and their corresponding coefficients in the
FI(s) = 3s - 3 has inverse transform fI(t) = 30'(t) - 30(t). Thus the problem reduces expansion are complex conjugates. For example, if F(s) has simple poles s = a j fJ
to finding the inverse transform of the other part, the proper rational function, for which and an expansion
there is no obvious table entry. For the remainder of this section we assume that F(s) is A B
a proper rational function. F(s) = +----
s-a-jfJ s-a+jfJ
The basic idea of partial fraction expansion is to write F(s) as a sum of simple
rational functions, or "partial fractions," each of which may be then easily inverse trans- then
formed. As an illustration, let us invert the transform
A = (s -a - jfJ)F(S)1
2(s + 10) S=Ol+j{3
F(s) - --'----'-- \
(s + 1)(s + 4)
Obtaining a partial fraction expansion is the opposite operation of getting a common B = (s - a + jfJ)F(S)ls=Ol_j{3
denominator. That is, we ask ourselves what simple fractions add together to yield F(s).
Since the transform is proper, the partial fractions will be proper and must therefore be From this we see that B = A *, because F(s) is a ratio of polynomials in s with real
of the form coefficients. The inverse transform is
2(s + 10) A B f(t) = Ae(a+j{3lt + A*e(Ol-j{3lt
F(s) = = -- + --
(s + 1)(s + 4) s+ 1 s+4 which is the sum of a complex number and its conjugate. Therefore,
The constants A and B are determined so as to make this an identity in s.
f(t) = 2 Re[Ae(Ol+j{3lt]
The simplest means of determining A and B is to note that
2(s + 10) B(s + 1) If A = IAle jO , we have
(s + I)F(s) = = A + ---
f(t) = 2 Re[IAle Olt ej({3t+Ol]
s+4 s+4
2(s + 10) A(s + 4) = 21AleOit cos(fJt + e)
(s +4)F(s) = = + B
s+1 s+1
SIMPLE PAIR For each simple (unrepeated) pair of complex conjugate poles p = a + j fJ, p* = a - j fJ
Since these must be identities for all s, let us evaluate the first at s = -1, which eliminates
OF COMPLEX in F(s), there will be a pair of terms AI(s - p) + A* I(s - p*) in its partial fraction
B, and the second at s = -4, which eliminates A. The results are
CONJUGATE POLES expansion, where A = (s - p)F(s)ls=p' There will be a single term 21Al eOi t cos(fJt + LA)
A = (s + I)F(S)1 = 2(9) =6 in its inverse transform due to the pair of poles.
s=-I 3
As an example, the transform
2(6)
B = (s + 4)F(s) 1 =- = -4 s
F(s) = - - - - : : - - - -
s=-4 -3 (s + 1)(s2 + 2s + 2)
Therefore, we have has the partial fraction expansion
6 4 s
F(s) = - - - - - F(s) = . . . , - - - - - - - - - - -
s+1 s+4 (s + 1)(s + 1 - jl)(s + 1 + jl)
and by Table 12.2,
A B B*
f(t) = 6e- t - 4e- 4t = s+1 + s+l-jl + s+l+jl
The poles p = -1 and p = -4 in this case are simple poles or poles of order 1, where
and in general for each such simple pole the partial fraction expansion contains a term
AI(s - p). In the case of a denominator factor (s - p)n, where n = 2, 3, 4, ... , the pole A = s2 + ~s + 2Is=_1 = -1
s = p is a multiple pole, or a pole of order n. The factors comprising (s - p)n are thus
not distinct, and the partial fraction expansion must be modified. We consider multiple s 1 1 - jl 1
poles later. B = (s + l)(s + 1 + jl) s=-!+jl = - 2 - = .;2/-45
504 Chapter 12 The Laplace Transform Section 12.4 Partial Fraction Expansion 505
Thus we have We wish to find the inverse transform of F(s) = 3/(s + 1)2(s + 2).
f(t) = Ae- t + 2Re [Be(-1+ j l)tJ There is a simple pole at -2 and a repeated pole of order 2 at -l.
Thus the partial fraction expansion is of the form
= _e- t + 2Re (~e-t/1t - 450) F(s) =
ABC
+ -- + -- (12.34)
(s + 1)2 S + 1 s+ 2
= _e- t + he- t cos(t - 45) Both methods for computing the terms associated with the repeated
pole begin by evaluating A as
An alternative form is
A=(S+1)2F(S)1 =_3_1 =3 (12.35)
f (t) = -e -t + he -t (cos t cos 45 + sin t sin 45) s=-1 s + 2 s=-1
= _e- t + e- t (cos t + sin t) The repeated subtraction method continues by subtracting 3/(s + 1)2
from both sides of (12.34), leaving
3 3 B C
Associated with a repeated pole p of order r, there will be r terms in the partial
----:--- - = -- + --
(s+1)2(s+2) (s+1)2 s+1 s+2
fraction expansion:
Combining terms on the left yields
-3 B
F(s) = N(s) Ar
---+
Ar-l Al
++--+Fl(S) (12.33) ----- = -- + -C- (12.36)
DJ(s)(s - p)' (s-p)' (S-p),-1 s-p (s+1)(s+2) s+1 s+2
But (12.36) is just the partial fraction expansion of Fl(S) =
Multiplying both sides of (12.33) by (s - p)' and evaluating at s = p, each term on
-3/(s + l)(s + 2), so
the right is zero except the first, which is A r . To get the remaining Ar-l, Ar-2, and
so on, either of two procedures may be used. The term just found involving Ar may
B = (s + I)Fl(S)1 = ~I =-3 (12.37)
be subtracted from both sides, leaving a new equation similar to the original (12.33) s=-1 s +2 s=-1
except that the pole at s = p is now repeated only r - 1 times. Ar-l may then be
found as described above, by multiplying by (s - py- 1 and evaluating at s = p. The C = (s + 2)Fl(S)1 = ~I = +3 (12.38)
process descends through the Ak'S until Al is found. We refer to this as repeated s=-2 s + 1 s=-2
subtraction. The repeated differentiation method finds B by computing the deriva-
The second procedure for finding Ar-l, Ar-2, involves repeated differentiation. If tive,
both sides of (12.33) are multiplied by (s - p)' and the result is differentiated with respect d 2 d 3 -3
to s, we have ds (s + 1) F(s) = ds s + 2 = (s + 2)2
506 Chapter 12 The Laplace Transform Section 12.4 Partial Fraction Expansion 507
Next we shall find the inverse transform of F(s) = 1/(s + 1)(s2 + facilitate their calculation, since half the coefficients define the other half. Terms in the
2s + 2)2. The roots of the quadratic are -1 j, so there are repeated inverse transform due to these poles will also always come in complex conjugate pairs, as
complex conjugate poles. Let p = -1 + j. Then in the example, and can conveniently be combined into a single phase-shifted sinusoidal
form.
A B A* B* C
F(s) =
(s - p)2
+--+
S - P (s - p*)2
+--+--
S - p* s + 1
(1241)
.
First compute A: EXERCISES
A- 1 I = 1 12.4.1. Find the PFE for F(s) = (s3 - l)/[s(s + 2)].
(s + 1)(s - p*)2 s=p (p + 1)(p _ p*)2 1 7
Answer s - 2 + 1. + ~
1 1 s s +2
= (+ j)(2j)2 = j"4 12.4.2. Find the inverse transform of F(s) = 5(s + 2)/[(s - 3)(s2 + 2s +
10)].
Continuing by the repeated differentiation method yields Answer e3t + 1.05e- t eos(3t - 161.6)
d 1 _ -[(s - p*)2 + 2(s + 1)(s - p*)] 12.4.3. Find the inverse transform of F(s) = (2s2 +s + 2)/(s5 + 2s 3 +s).
ds (s + 1)(s - p*)2 - (s + 1)2(s _ p*)4 Answer 2 - 2eost + ~(sint - t cos t)
Evaluating at s = p, we obtain
B = -[(p - p*)2 + 2(p + 1)(p - p*)]
(p + 1)2(p _ p*)4
Using the fact that p - p* = j2 Im(p) = 2j and p +1= j gives
In this chapter we have introduced the basic tools needed to work in the s-domain:
B = -[-4 + 2(j)(2j)] _ ~ definitions, transform properties and pairs, and partial fractions. Thus armed, we close
(-1)(16) - 2 the chapter by discussing a most important application of these ideas.
We also need C: Laplace transforms may be used to solve differential and integrodifferential equa-
tions, such as the equations of basic electric circuit analysis. The equation or equations
C= 2 1 I =1 are first Laplace transformed and then solved by straightforward algebraic means. The
s + 2s + 2 s=-1
inverse transform of the solution is identified by partial fraction expansion.
We now have the full partial fraction expansion (12.41). It remains
to find the inverse transform. From Table 12.2, As a first example, we seek the solution of the second-order differ-
f(t) = Ate pt + Be pt + A*tep*t + B*ep*t + Ce- t ential equation
d 2x dx
We recognize that the terms involving A and A * are conjugate terms, - +4- +3x = e- 2t (12.42)
as are those involving Band B*, so dt 2 dt
for t > 0, given the initial conditions x (0-) = 1, x' (0-) = 2.
f(t) = 2 Re(Ate pt + Be Pt ) + Ce- t First, note that (12.42) is an equality between time functions; thus
To simplify the final substitution, converting to polar form, we have their transforms must be equal. Both sides of a time-domain equa-
A = ~/90, B = ~LQ. Then tion may be Laplace transformed, or an s-domain equation inverse
transformed, while retaining equality. Transforming (12.42) to the
f(t) = 2Re [t G/900) (e- t e+ jt ) + GLQ) (e- t e+ jt )] + e- t s-domain and noting that the linearity property allows transforming
= 1/2te-t Re(e+jt+90o) + e- t Re(e+ jt ) + e- t a weighted sum as a weighted sum of transforms,
508 Chapter 12 The Laplace Transform Section 12.5 Solving Integrodifferential Equations 509
Solving for the unknown transform Xes) yields This transform has complex conjugate poles at -1 j2 in the
s-domain. Then
Xes) = s2 + 8s + 13 A A*
(s + l)(s + 2)(s + 3) I(s) =
s+ 1-
+---
j2 s + 1 + j2
The partial fraction expansion is, using the method of Section 12.4, where
311
Xes) = - - - - - - - -
A = 1 I j
s+l s+2 s+3 + 1 + j2 s=p=-I+j2
s 4
which has inverse transform x(t) solving the original differential Then i(t) = Meat cos(f3t + e), where p = IX + jf3 -1 + j2,
equation M = 21AI, e = LA, and
x(t) = 3e- t - e- 2t _ e- 3t i (t) = !e- t cos(2t - 90) = !e- t sin 2t
This value for x(t) is not necessarily valid for negative time, since As our final example, we will find the node voltages VI and V2 in
the differential equation we solved is required to be valid only for the circuit shown in Fig. 12.11. The nodal analysis equations are
t ~ O. We may verify this solution by direct substitution into
1 d d
(12.42). -(VI - 3) + -VI + 2-(VI - V2) = 0
2 dt dt
d2 d
-(3e- t _ e- 2t _ e- 3t ) + 4-(3e- t _ e- 2t _ e- 3t ) d 1
dt 2 dt 2-(V2 - VI) + -V2 = ret)
dt 3
+ 3(3e- t - e- 2t - e- 3t ) = (3e- t _ 4e- 2t _ ge- 3t ) Since neither source switches on until t = 0, the values of all currents
+ 4( -3e- t + 2e-2t + 3e- 3t ) + 3(3e-t _ e- 2t _ e- 3t ) = e- 2t and voltages at t = 0- are zero. Transforming the equations with
these initial conditions gives
as (12.42) requires. Checking the initial conditions, we note that x(t)
and its derivative are continuous at t = 0, so x (0-) = x (0) = 1 and
x' (0-) = x' (0) = 2, as required.
~ [ V I (s) - n + s V I(s) + 2s [V I(s) - V2 (s)] = 0
1 1
2s[V2(s) - VIeS)] + 3"V2(s) = S2
In Example 12.17, the initial conditions were specified at time t = 0- and the
differential equation was required to be satisfied for t > O. This is a useful way to This is a pair of coupled algebraic equations in the unknowns V I (s)
specify the initial conditions for a circuit problem, since we may compute the initial fiGURE 12.11 Circuit for Example 12.18. and V 2 (s), which may be solved by several methods. Choosing
values at t = 0- using steady-state calculations based on the t < 0 circuit (the circuit matrix inversion, we rewrite the node equations as
as it appears before the switches and unit steps act at t = 0), as described in Chapters 5
and 6. This procedure is shown in the next two examples.
[ 3S +! -2s ] [VIeS)] = [;s]
-2s 2s + ~ V2(S) ~
Consider the single-mesh circuit of Fig. 12.10. Its mesh equation s2
for t > 0 is
Performing the matrix inversion, the determinant is 2s2 +2s +! and
IH di
dt [t
+2i+ 5 Jo_ i ('r)dr+l ] = l+u(t) (12.43)
Just before the unit step acts, the t < 0 circuit is in dc steady state [ VIeS)] _ 1 [2S +~
IV
with i (0-) = 0 A and Vc (0-) = 1 V. The net source voltage for
V 2 (s) - 2s2 + 2s + ! 2s
t > 0 is the sum of the sources 1 + u(t). Transforming (12.43)
gives ~s
2
+ 4~
5I(s) 1 1 1
[sICs) - 0] + 2I(s) + - - + - = - + -
S(S2 +s+ fz)
s s s s ~s2
2
+ ~s
2
+ !4
fiGURE 12.10 Circuit for Example 12.17. or I(s) _ --:-_1__ s2(s2 + s + fz)
- s2 + 2s + 5
510 Chapter 12 The Laplace Transform Section 12.S Solving Integrodifferential Equations 511
The partial fraction expansion for V I (s), with real poles at 0, approach to solving these equations is almost always more efficient than the conventional
I !1 .
-2 V6' IS
time-domain approach.
ABC
V 1 (s) = - + + ---=
s s+!- f! s+!+ f! EXERCISES
2 '16 2 '16
where 12.5.1. Find x(t), t > 0: d 2x/dt 2 + 3(dx/dt) + 2x = 2, x(O-) = 0,
dx/dtlo- = O.
A= i -7-12 = 15 Answer 1 - 2e- 2t + e- t
and resorting to finite precision on our calculator, 12.5.2. Find x(t), t > 0: d 2x/dt 2 = 2t + 1, x(O-) = 0, dx/dtlo_ = l.
Answer ~t3 + ~t2 + t
B = (D (-0.0918) + i = -14.848 12.5.3. Solve the following pair of differential equations for t > 0:
( -0.0918)(0.816)
2
-d x - x - 5y dx I = 0
= 0, x(O-) = 0, -d
(l) (-0.908) + ~ dt 2 t 0-
C= -0 152
-
2 4 -
(-0.908)(-0.816) - .
-~y - 3x +y= 0, y(O-) = 0, -~I - 64
or VI (t) = 15 - 14.848e-0.0918t - 0.152e-0.908t dt 2 dt 0-
The partial fraction expansion for V2 (s) is Answer x(t) = lOe 2t _lOe- 2t -20sin2t
A2 Al B C yet) = 6e 2t - 6e- 2t + 20sin2t
V 2 (s) = 2 + - + +---=
s s 1!1 1!1
s+2-V6 s+2+V6
where
Subtracting yields
Laplace transformation is a way of converting functions of time i~to func.tion~ of a c~m
ls2
2
+ ls
2
+ !4 plex variable s. The calculus operations of differentiatio.n ~d. mtegratIO~ .1~ th~ tIme
S2 (s2 + s + 12) domain go over to simpler algebraic operations of multIphcatIon and dl~lSlon 1~ the
s domain. Thus a key use of Laplace transformation is to convert sets of mtegrodlffer-
The remaining values AI, B, and C may now be found using the sim- ential equations, such as circuit equations, to the s-domain, solve using si~ple algebra,
ple real pole formula on the rational function written on the right side and convert the solution back to its desired time-domain format. ConversIOn back, or
of the equation above, and the inverse transform is computed to be inverse transformation, is accomplished by a straightforward manipulation called partial
V2(t) = 3t - 18 + 18.1ge-0.0918t - 0.190e-0.908t fraction expansion.
The s-domain approach to the solution of differential or integrodifferential equa- The Laplace transform of f(t) is defined by the integral transform
tions has some clear advantages over classical differential equation techniques worked in F(s) = to f(t)e- st dt
the time domain. First and foremost, derivatives and integrals of unknown time functions 10-
transform to powers of s times unknown transforms, effectively converting a problem The Laplace transform of a sum is the sum of Laplac~ tr~sfo~s; scalin~ a time
in calculus to one in algebra. Initial conditions at t = 0- are taken into account nat- function scales its transform equally. Laplace transformatIOn IS a hnear operatIOn.
urally by the method and do not have to be "fit" after a general form for the solution
The unit step u(t) and its integrals and derivatives are called s~ngularity function~. Their
is found. The price paid for these advantages is that the equations must first be trans-
Laplace transforms are sn, where the integer n is 1 for the umt step, 2 for the umt ramp,
formed and then, at the end, their s-domain solutions returned back to the time domain
3 for the unit parabola, and 0 for the unit impulse.
(inverse transformed). This price is seldom very high, since the s-domain version of the
original equations can always be written by inspection, and partial fractions can easily be The unit impulse oCt) is the generalized derivative of the unit step. oCt) = 0 for all
computed, at least for low-order rational functions. High-order rational functions imply t =I 0, and has a total area of unity.
high-order integrodifferential equations or complicated source functions, either of which The Laplace transform of a derivative df /dt is s times the Laplace transform of the
would add greatly to the complexity of time-domain solutions as well. The s-domain function f(t) minus its initial value f(O-).
1t21
of simple terms each one of which has a known inverse transform.
The PFE term due to a simple pole p is A/(s - p).
The PFE terms due to an r-fold repeated pole p are of the form Ak/(S - p)k for e-(t-2) I> 2
k=I, ... ,r. /' ' > 1
4
Integrodifferential equations may be solved by converting the equations to the s-domain (b)
term by term, solving for the unknown transforms, and using partial fraction expansion 2
to recover the inverse transforms. (a)
(c)
\
(b)
PROBLEMS
12.1. Determine F(s) for the following f(t)'s: (a)
(a) 5e- 9t - 2e- t + e+ t
(b) (5e- 9t - 2e- t + e+t)u(t + 1)
if:::::::l::::::::f-------I-------~,
(c) e j2t _ e- j2t
+--~~5
(b) 6r(t) - u(t - 2)
12.5. Explain why f(t) = 1/(t + 1) is Laplace trans- (c) 4[u(t - ~) - u(t - ~)] + pet)
formable but 1/(t - 1) is not. (d) ret) - ret - 1) - r(t - 2) + r(t - 3)
12.6. Specify a periodic function f(t) that is not Laplace 12.14. Evaluate the following and determine if they are
5 (c)
transformable. Explain why not. ordinary derivatives or generalized derivatives.
(b)
(a) ~ [sin2 tu(t)]
g 12.7. ~hich function converges to zero faster: tlOe- lOt or
FIGURE 1'12.11
dt
IIilll e- 9t ?
Fmd a t* such that the one you select is always less (c) d
(b) -[u(t) - te-tu(t)]
than the other for t > t*. This defines faster convergence. 12.12. Express the following f (t)' s as linear combina- dt
tions of time-shifted unit steps u(t), ramps r(t), and parabo- d2
12.8. Considering the defining integral, what is the value las pet). All are zero for t < O. (c) dt 2 cos 2t u(t)
of F(s) if f(t) = 0 everywhere except f(to) = 1, to ::: O? 21-----. (a) 12.15. Evaluate the first and second derivatives.
What does this imply about the difference fl (t) - h(t)
(a) f(t) of Problem 12.1(a)
of two time functions with the same Laplace transform
(b) f(t) of Problem 12.1(b)
Fl(s) = F2(S)?
(c) f(t) of Problem 12.1(c)
'I I~
3
12.9. Use the defining integral (12.1) to find the Laplace 12.16. Evaluate the first three derivatives df /dt, d 2 f /dt 2 ,
transform of 5e- 2t cos(3t + 12). Hint: Use the Euler iden- -2 and d 2 f /dt 3 .
tity and linearity. (c) >1 (a) f(t) of Problem 12.12(a)
4 6 (b) f(t) of Problem 12.12(b)
12.10. Determine F(s) for the following f(t)'s: FIGURE 1'12.10 (a) (c) f(t) of Problem 12.12(c)
520 Chapter 13 Circuit Analysis in the s-Domain Section 13.1 Element and Kirchhoff's Laws 521
The element law for an inductor is We wish to find vc(t), t > 0 for the circuit shown in Fig. 13.3. By
d. KVL in the s-domain, summing voltage drops around the loop, we
VL(t) = L-l(t) (l3Aa) obtain
dt
Using the derivative rule, in the s-domain this equation becomes (13.8)
+ vc(t)_
V LCs) = sLIL(s) - L iLCO-) (13Ab) Since there are no sources in the circuit for t < 0, all initial condi-
tions at 0- are zero. Each voltage above is therefore the product of
The element law for a capacitor can be written impedance Z(s) and current I(s). The impedance of the capacitor
d
(l3.Sa) is Zc(s) = 1/sC = 2/s, for the inductor it is ZL(S) = sL = s, and
ic(t) = C -vc(t) 3!.1
dt for the resistor, ZR(S) = R = 3. Then
Transforming gives
Ic(s) = sCVc(s) - Cvc(O-) (l3.5b)
~I(S) + 3 [I(S) + ~ ] + sICs) = 0
(13.9)
Solving the last for Vc(s), we obtain FIGURE 13.3 Circuit for Example 13.1. I(s) _ -31s = -3
1 1 s + 3 + (2/s) S2 + 3s + 2
Vc(s) = - Ic(s) + - vc(O-) (l3.5c) Vc (s) may be recovered by multiplication of I(s) by the impedance
sC s
It has been assumed in writing these element laws that the current and voltage variables of the capacitor.
satisfy the passive sign convention. -6
Equations (13.3b), (13Ab), and (13.5c) show that the voltage V(s) across the RLC V c(s) = Zc(s)I(s) = s(s2 + 3s + 2)
elements is the sum of a term proportional to its current I(s) and a term that depends on This completes the s-domain solution for the capacitive voltage
its initial condition (this term is zero in the case of a resistor). Vc(s). To recover the time-domain solution vc(t), we require the
The factor of proportionality between voltage and current in the first term is the inverse transform. By partial fractions we have that
impedance Z(s). We define the impedance Z(s) of an RLC element as the ratio ofWs)
-6 -3 6 3
to I(s) when all initial conditions are zero. Vc(s) = = - + -- - --
s(s + 1) (s + 2) s s +1 s +2
so
V(s) = Z(s)I(s) (all ICs = 0) (13.6) vc(t) = 6e- t - 3e-2t - 3 (t > 0) (13.10)
The inverse of the impedance Z(s) is called the admittance Yes), where
Applying this to the equations cited above yields impedances
I(s) = Y(s)V(s) (all ICs = 0) (13.11)
Admittance Y (s) is the ratio of current I(s) to voltage V (s) when all initial conditions are
ZR(S) = R (13.7a) zero. In light of the corresponding impedances in (13.7), the admittances of a resistor R,
ZLCs) = sL (13.7b) inductor L, and capacitor C are
1
Zc(s) = sC (13.7c)
1
YR(s) = - = G (13.l2a)
R
as shown in Fig. 13.2. 1
YL(s) = - (13.12b)
sL
Yc(s) = sC (l3.l2c)
The Kirchhoff laws and the element laws are the basis for all other useful circuit
theorems, such as the rules for series-parallel combination, current and voltage division,
and Thevenin-Norton equivalents. Since the Kirchhoff laws are identical in the time do-
FIGURE 13.2 Impedance Z(s). main and s-domain, and the element law V(s) = Z(s)I(s) is of the same form as Ohm's
522 Chapter 13 Circuit Analysis in the s-Domain Section 13.1 Element and Kirchhoff's Laws 523
law in the time domain, we expect that these rules would carry over to the s-domain Z(s). Impedances may be combined in series and parallel by the usual rules. Other
unchanged. This is indeed the case. familiar workhorse theorems of circuit analysis, such as current and voltage division and
Thevenin and Norton transformation, carry over into the s-domain as well. If the initial
Find the equivalent impedance Z(s) of the subnetwork shown in conditions are all zero, they carry over unchanged in form. If the initial conditions are
Fig. 13.4. Since the 1-H inductor and 2-0 resistor are in series, not identically zero, an initial condition term must be added to the corresponding element
their impedances s L = sand R = 2 add: law V(s) = Z(s)I(s). In Section 13.2 we shall see how to take nonzero initial conditions
Zj(s) = s +2 into account without difficulty.
I(s)
The impedance Zj (s) is in parallel with a ~-F capacitor whose
----7 2H IH impedance is Zds) = l/sC = 3/s. The equivalent is the inverse EXERCISES
of the sum of inverses, or
+
1 3(s + 2) 13.1.1. Find I(s) in Fig. 13.3 by s-domain current division. Assume zero
\ initial conditions.
Yes) !p 2n Z2(S) = [l/(s + 2)] + (s/3) = s2 + 2s + 3
3 Answer
Z(s) Z2 (s) could also have been found by adding parallel admittances
= sj(s2 + 2) (_~) = _
----7
4n tic Yj(s) = l/Z j (s) = l/(s + 2) and Yds) = s/3 and then inverting
the equivalent admittance Y2(S). Or, since there are exactly two
I(s)
[sj(s2+2)]+~ s
-3
s2+3s+2
: ----7 : ---7> 13.1.2. What is the equivalent admittance Y(s) of a lO-Q resistor, 1-F
impedances Zj (s) and Zds) in parallel, the equivalent impedance
Z2(s) Z1(S)
Z2 (s) could have been computed as the product of these impedances capacitor, and 5-H inductor all in series? Express as a rational function.
Circuit for Example 13.2. divided by their sum. Answer s j(5s 2 + lOs + 1)
FIGURE 13.4
Combining Z2 (s) with the impedances it is in series with gives 13.1.3. Find the Thevenin and Norton equivalents in the s-domain at the
terminals of the inductor of Fig. 13.3. Assume zero initial conditions.
Z(s) = Z2(S) + 2s + 4 (l3.13a) Answer
3(s + 2) + (2s + 4)(s2 + 2s + 3) 2
s2 + 2s + 3 S
3
~r------~V'IfV--_
2s3+8s 2 +17s+18
(13.13b) 3
s2 + 2s + 3
3
Note that all the equivalent forms for combining series-parallel elements are valid 3s + 2
2
for impedances Z(s): the product-by-sum shortcut for two parallel impedances, summing S
admittances for parallel admittances, and so on. + +
(a) (b)
Find Ids) in the circuit of Example 13.2 for an input voltage
V(s) = 1/s2, assuming that all initial conditions are zero. Using EXERCISE 13.1.3
(l3.13b), we have
S2 + 2s + 3 1
I(s) = Y(s)V(s) = 2s 3 + 8s2 + 17s + 18 s2
By the current divider rule, current divides in proportion to admit-
tance. The admittance of the capacitor is sC = s/3, and for the Since the Kirchhoff laws remain valid in the s-domain, we can write circuit equations in
other arm of the current divider dividing I(s) it is l/(s L + R) = the s-domain directly by inspection of the circuit diagram. Voltage drops around loops in
l/(s + 2), or the s-domain still sum to zero, as do s-domain currents into each node or closed region.
s/3 s +2 We need only relabel the original time-domain circuit diagram to reflect the necessary
Ids) = (s/3) + [l/(s + 2)]I(s) = s(2s 3 + 8s 2 + 17s + 18) changes: s replaces t in the unknown currents and voltages, and independent source
functions are replaced by their s-domain transform pairs.
In summary, the Kirchhoff laws are unchanged in s-domain analysis, and the no- One additional change is necessary before the s-domain circuit diagram is com-
tion of impedance, first defined for phasor circuits, carries over into the s-domain as plete. Let us return to the s-domain RLC element laws (13.3) to (13.5). Repeating the
524 Chapter 13 Circuit Analysis in the s-Domain Section 13.2 The s-Domain Circuit 525
v-forms of these element laws, 60
30H
(a) (b)
The first of these equations, that for a resistor, is of the form V(s) = Z(s)l(s), where
Z(s) = R is the impedance of an R-ohm resistor. The other two are of a form which FIGURE 13.6 (a) Original circuit; (b) s-domain circuit.
suggests that the storage elements in the original time-domain circuit are each equivalent
to two elements in series in the s-domain: a pure impedance and an initial condition
generator. Circuit diagrams for these equivalents are shown in Fig. 13.5. Each labeled 12s2 + lOs + 4
V (s)=--~-.,..-
impedance, whether R or sL or lIsC, is considered a two-terminal device satisfying a pure
impedance relationship V(s) = Z(s)l(s). The initial condition is accounted for separately
S (s2 + ~s + n
Since Vc(s) = V(s), this completes the solution in the s-domain. To
as if a second element, the initial condition generator, was present.
recover the time-domain value, the inverse transform is computed
This is the only other change between time-domain and s-domain. The s-domain
by the usual method.
circuit is simply the original time-domain circuit with s-domain unknowns and source
24
junctions replacing their time-domain counterparts and combinations of impedances and V(s) = - + -24- - -36-
initial condition generators replacing the RLC elements. The equations of s-domain s s+!2 s+!3
circuit analysis follow from inspection of this circuit diagram. and for t > 0,
(13.15)
Let us find the s-domain mesh currents 11 (s) and 12(s). The origi-
R nal circuit is shown in Fig. 13.7(a) and the corresponding s-domain
circuit in Fig. 13.7(b).
2
2A s
(a) (b) (c) .------i ~l_--___. r------(~.I__----_,
We seek the voltage across the capacitor in Fig. 13.6(a) for t > O.
Example 13.4
Noting that the source jumps from 12 to 24 V at t = 0, the dc
steady-state values at 0- are computed to be i L (0-) = 2 A and
vc(O-) = 12 V. The s-domain circuit is shown in Fig. 13.6(b). e- 3t u(t) V + 4Q 8
s
Note that impedances are labeled here, together with initial condi-
tion generators LidO-) = 60 and vc(O-)ls = 121s for the storage
elements. The node equation at the upper-right node in Fig. 13.6(b)
is then lQ 2Q 2
V(s) - (12/s) V(s) V(s) - [60 + (24/s)] _ 0 (a) (b)
Sis + -6- + 30s -
FIGURE 13.7 (a) Original circuit; (b) s-domain circuit.
526 Chapter 13 Circuit Analysis in the s-Domain Section 13.2 The s-Domain Circuit 527
To determine the equivalent circuit for each inductor, its initial
current at 0- is needed. For t < 0, the voltage source in Fig. 13.7(a)
is zero and the inductors act as short circuits, so iLl (0-) = iL2(0-) =
2 A. Thus the initial condition generator for LI is Lli Ll (0-) =
(1)(2) = 2 and for L2 is L2iL2(0-) = (2)(2) = 4. As shown in
Fig. 13.5, these initial condition generators are in series with their re-
spective inductive impedances ZI (s) = s L I = sand Z2 (s) = SL2 =
2s. The capacitive voltage at 0-, again examining Fig. 13.7(a), is
oV. No initial condition generator is needed for this element. Noting (a) (b) (c)
that the mesh current in the upper mesh is 2/s, the mesh equations are
[
s+
-4
5 +-4]
+ -;
2s
[I6
(S)] [ _1 ]
8 I
12 (s)
= s+3
0
FIGURE 13.8 s-Domain equivalents of RLC elements: alternative forms.
\ The number of equations and unknowns is reduced and our ability to recover all currents
and voltages once the circuit is broken is unimpaired.
Solving by matrix inversion gives
[ 12(S)
11(S)]=_I_[2S+6+~
Do(S) 4
4][_1_]
s+ 5 s -:; 3
EXERCISES
+6+~) -
+ I 13.2.2. Write the mesh equations for I(s) and solve for i(t), t > O.
Do(s) = (s +5) (2S (-4)(-4) 40 3H v
Answer i (t) = (1 - t)e- t
u: F
Carrying out the algebra, the vector of mesh currents is 20
which completes the required calculation. Note that each mesh cur- IH IF
rent has the same four poles but different zeros, so they will contain
the same four time functions but with different partial fraction coef-
ficients. One pole is due to the independent source. The other three
60 2H
are due to the three storage elements in the circuit. EXERCISE 13.2.2
Thevenin-Norton transformation in the s-domain works exactly as described in 13.2.3. Using series-parallel equivalents in the s-domain for both impe-
v4
the time domain in Chapter 2, with Z(s) replacing resistanC'e R. Consider, for instance, + dances and sources, reduce the given circuit to one containing only two
V2 unknown node voltages VI(S) and V2(S), and write the node equations in
the networks shown in Fig. 13.5. Since each is a series combination of impedance
and voltage source, we will call these the Thevenin forms of the s-domain equivalents 10
IF ~i2 vector matrix form.
3H Answer
(the Thevenin equivalent voltage source in the resistive case is defined to be 0). If we
20 IH
-(S+2S~6)
perform Thevenin-Norton transformation on these circuits, the equivalent Norton forms 1 4s 1
have current sources equal to the ratio of the Thevenin equivalent voltage divided by the 20 2s+6+ s + 4s+1 +3"
4F [ ] [VI(S)]
impedance and are in parallel with the same impedances. The resulting Norton forms of ~i3 _(s+_l)
1 1
--+s+-+--
1 V2(S)
the s-domain equivalent circuits are shown in Fig. 13.8. As with all equivalents, either 2s +6 2s + 6 S 3s + 2
may be freely used. Note that each Thevenin-form s-domain equivalent circuit with a
[ 1- (8 +:;;5) 1 (2)
nonzero initial condition generator adds a node to the circuit, whereas each Norton form 4s ]
viO-) = i 2 (O-) = I, i 3(O-) =-1
adds a mesh. The Thevenin form is often preferred when there are other impedances in vs(O-) = 2, i[(O-) = 4
= 2s+6+:;; 4s+1
series, and the Norton form when other impedances are in parallel. 5) - -
1 - -1+ -3 -
EXERCISE 13.2.3 -1+ ( 8+-
Finally, it is always helpful to combine series impedances before writing nodal s 2s +6 s 3s + 2
analysis equations in the s-domain and parallel impedances if mesh analysis is planned.
528 Chapter 13 Circuit Analysis in the s-Domain Section 13.2 The s-Domain Circuit 529
To determine R(s), all initial conditions are set to zero and the s-domain circuit
13~3TRANSFERFUNCTIONS
drawn. A consequence of the zero initial condition requirement is that there will never
Often our goal in analyzing a circuit is to determine its complete behavior: the values be initial-condition generators in this circuit. The value of the output selected is then
of all its currents and voltages in response to all its sources and initial stored energy. computed. Division by the input, as in (13.17), yields the transfer function R(s),
Most of our work has been oriented toward this completely general goal. The two sys-
tematic circuit analysis approaches emphasized in this book, mesh and nodal analysis, As a first example we compute the transfer function R(s) for the in-
each solve for a set of key variables (the mesh currents and node voltages, respec- put vs(t) and output VI (t). The time-domain and s-domain circuits
tively), from which all remaining circuit variables can be easily derived. Even if in are shown in Fig. 13.9. Note the absence of initial condition gen-
a given problem only a subset of all circuit variables is of interest, our strategy has erators in the s-domain circuit, a consequence of setting all initial
been to first "break the circuit." Once the circuit is broken, in other words once the conditions to zero.
key variables have been found, every other circuit variable of interest becomes easy to We proceed by computing the specified output V I (s). The
compute. equivalent impedance of the three rightmost elements is
Sometimes, however, our interest in studying a given circuit is much more narrowly
ZI (s) = (s)(2s + 4) = 2S2 + 4s
defined than that of finding all responses to all sources. We may wish to understand the
relationship between one particular source or other circuit variable we will call the input
s + (2s + 4) 3s +4
and another circuit variable, the output current or voltage that input produces. The goal Then, by voltage division,
of input-output analysis is to understand their relationship, in other words to be able to V (s) _ 4js
predict the output for each possible value of the input. If only a single input-output I - (4js) + ZI(S) Vs(s)
pair is of interest, it is unlikely that methods oriented to determining all currents and (13.19)
6s + 8
all voltages in response to all sources and initial conditions will be very efficient when
s3+2s 2 +6s+8 Vs (s)
applied to this narrower purpose.
Analysis in the s-domain has already shown its value in speeding calculation of Having solved for the output in terms of the input, R(s) is found by
the complete behavior of a circuit. It turns out that the s-domain approach is also highly division of both sides by the input Vs(s).
suited to input-output analysis, as we consider next. For a circuit with input VieS) and
R(s) = VI(S) = 6s + 8 (13.20)
output V 0 (s), define the transfer function R(s) as the s-domain ratio of the output to the
Vs(s) s3 + 2S2 + 6s + 8
input when all initial conditions are zero:
Vo(s)
(all ICs = 0) (13.17)
R(s) = VieS)
4Q V,(s) 4
Here Vi (s) and Vo (s) may be both currents, both voltages, or one of each. We require
that all initial conditions be set to zero to isolate the selected input as the cause of the
output we are computing.
Rewriting (13.17), we see the significance of this definition:
(a) (b)
FIGURE 13.9 (a) Circuit for Example 13.6; (b) corresponding s-domain
Vo(s) = R(S)Vi(S) (all ICs = 0) (13.18) circuit.
A consequence of the linear relationship between input and output seen in (13.18)
The value of the s-domain output for any input is just the product of the transfer function is that if the input Vi (s) = 0, the output must be zero also. Since this would not be the
and the s-domain input. Thus, once R(s) is known, we may predict the output Vo(s) case if there were other independent sources contributing to the output Vo(s), they must
for any input Vi (s) by simple multiplication of two known s-domain quantities. If the be killed before computing its transfer function R(s) relative to the specified input. The
time-domain output vo(t) is required, the additional step of inverse transformation will transfer function measures the output due to a specified input, with all other sources and
be required. initial conditions set to zero.
530 Chapter 13 Circuit Analysis in the s-Domain Section 13.3 Transfer Functions 531
Find the transfer function HI(s) = VZ(S)!lsl(S) for the circuit in
or
Fig. 13.10. The nodal equations with Vsz(s) killed are, in vector-
matrix form,
Vz(s) = -'-:1------'-I
s +2 2
=
-2
2sz + 7s + 2 Is](s)
responding s-domain inputs.
(13.25a)
In determining a transfer function in a circuit containing more than one independent
-2 s+2
source, the other sources must be killed before the output is computed. Let Vi I (s), ViZ (s), Vz(s) = 2sz + 7s + 2 lsI (s) + 2sz + 7s + 2 Vds) (13.25b)
... , VineS) be the n separate independent sources in a circuit, and Vo(s) be the designated
output variable. Then, with all sources j :f. k killed, From this input-output equation (13.25b), we may now deduce the
output for any input conditions directly, without computing any un-
VokeS) = Hk(s)Vik(S) [all ICs = 0, all other sourcesVij(s) = 0] (13.22) needed circuit variables along the way. For instance, if the inputs
where Hk(s) is the transfer function between output Vo(s) and input Vik(S), and VokeS) are set as isl(t) = u(t) and vsz(t) = e-tu(t), then
is the value of the output produced under these conditions.
Now suppose that we are computing the total output Vo(s) due to all sources Vz(s) = 2s z +-;s + 2 (~) + 2SZ:~: + 2 C~ 1)
Vi! (s), V;z(s), ... , Yin (s), keeping the initial conditions at zero. Using superposition, the
and vz(t) would be computed by partial fractions. (These calcula-
component due to source V ik (s) would be V ok (s), the value of the output computed under
tions assume zero initial conditions.)
precisely the same conditions of (13.22). Thus the total output would be the superposition
Vo(s) = Vol(S) + Voz(s) + ... + Von(S) (l3.23a) Once H(s) is known, it is easy to vary the input and determine the effect on the
output. The transfer function H(s) will prove central to input-output analysis, that is,
532 Chapter 13 Circuit Analysis in the s-Domain
Section 13.3 Transfer Functions
533
analysis focusing on the cause-effect relations between chosen inputs and outputs. We .4
Higher frequency w=Imp
will exploit this capability, for instance, in studying frequency response in Chapter 14,
the dependence of the output on the frequency of the input.
Although useful, transfer functions do not tell the entire story. Transfer functions
cannot tell us about the full range of a circuit's potential behaviors, since they are defined + -4ffiI ~ 1
with initial conditions zero. For nonzero initial conditions, the natural response, that is,
the output when all independent sources are killed, must be superposed with the output
due to the inputs or forced response computed using R(s) if the total output is desired.
The natural response may be computed from the s-domain circuit with initial condition
generators present and all other independent sources killed.
-+
~
-1IID
y--
~
f
1
~a=Rep
Faster decay Faster growth
~ ------':7
EXERCISES
x x x x
1,2(S) 13.3.1. Denote the source in Fig. 13.3 is(t) and find H(s) for input Is(s)
and output V <s).
,--------1<E--l------.
Answer s2/s2 + 3s + 2
x x x x
13.3.2. Find the transfer functions between output I(s) and inputs Vs l(S),
Answer -(7s + 2)/(12s 2 + 13s + 2), (13s + 2)/(12s 2 + 13s + 2) FIGURE 13.12 Dependence of the term e pt in f(t) on the location of the
13.3.3. Find H(s) with input VR(S) and output V<s). [Hint: Find both pole p = a + jw in the complex plane.
4s
in terms of Vs(s).]
Answer -Ls/R term eat decays to zero for a < 0, is constant for a = 0, and grows without limit for
I(s) ~
L -_ _ _ ___ a > O. For w i= 0, the pole is complex, and the corresponding exponential e pt = eat e jwt
~
~
is complex. In this case, Fig. 13.12 shows the term e Pt , combined with the term eP*t,
EXERCISE 13.3.2 which is due to the presence of its complex conjugate pole p* = a - j w. Since
(13.26)
Y,(s)
we see that the rate of growth or decay in each term off(t) is governed by the real part of
1. its pole. If Re (p) = a < 0, the corresponding term inf(t) decays to zero. This conclusion
sC
also holds for repeated poles, since t-multiplied forms of eat eventually decay to zero
with time as long as eat itself does. Note in Fig. 13.12 that only the upper half of the
EXERCISE 13.3.3 complex plane responses are sketched, since each response sketched at the complex-pole
location p actually combines terms due to a complex conjugate pair of poles p and p*.
The responses not sketched in the lower half of Fig. 13.12 are identical to those shown
at the complex conjugate locations in the upper half of the complex plane.
We are particularly interested in the location of the poles of the currents and voltages
in the natural response. For any linear circuit, we may kill all independent sources and
write a complete set of network analysis equations, either mesh or nodal analysis, in
In Chapter 12 we saw that the poles of a transform F(s) determine, up to a multiplying
the s-domain. To be specific, let us assume that mesh analysis has been selected. The
constant, the terms in the partial fraction expansion and therefore the time functions that
equations will be of the form
make up its inverse transform f(t). A simple pole at s = p gives rise to a term Kept
in f(t), while repeated poles at s = p create t-multiplied forms of the same exponential M(s)l(s) = V/c(s) (13.27)
time function.
The dependence of the term Kept in f(t) on the location of the pole p in its where I(s) is the vector of unknown mesh currents, V/c(s) the vector of initial condition
transform F(s) is illustrated in Fig. 13.12. Let p = a + jw , so e pt = eat e jwt . For w = 0, generator source terms, and M(s) the square connection matrix. The values of s for which
we have a purely real pole located on the horizontal axis in the figure. The corresponding the determinant of M (s) equals zero are called the poles of the circuit.
534 Chapter 13 Circuit Analysis in the s-Domain Section 13.4 Poles and Stabi Iity 535
Consider the circuit shown in Fig. 13.13(a) transformed to the The poles of a system dictate the fate of its natural response. As shown in
s-domain in Fig. 13. 13(b). The mesh equations with Vo(s) = 0 Fig. 13.12, each pole whose real part is negative will lead to natural response terms
are that decay to zero. Those with zero real parts will neither grow nor decay, while poles
with positive real parts generate terms that grow without bounds.
[
2s
-4
+4 -4] [Ith(s)
4+~
(s) J = [ ~i (O-=! ]
vc(O )
I We define a circuit to be stable if all natural responses in the circuit decay to zero
with time. Since only circuit poles with negative real parts produce natural responses that
s s
decay to zero, the stability of a circuit is equivalent to Re(p) < 0 for every circuit pole p.
Then the determinant 8 (s) is Stated another way, stability requires that all the poles of the circuit lie in the left half of
Ll(s) = (2s + 4) ( + 8) -
4 ~ 16 =
8(s2+2s+4)
s (13.28)
the complex plane as shown in Fig. 13.12. The circuit of Example 13.9 is stable, since
the complex conjugate system poles have real part -1. The next example illustrates a
circuit that for different ranges of a circuit parameter may be either stable or unstable.
The poles ofthe circuit are the zeros of Ll(s), which are the quadratic
roots PI, P2 = -1j,J3. Solving (13.28) yields Consider the s-domain circuit of Fig. 13.14, which contains a current-
8 controlled voltage source of gain k ohms and a single inductor pro-
II(s) 1 [ 4+- ducing the initial condition generator shown. Determine if this cir-
[12(S) J = Ms) 4 s
(13.29)
cuit is stable.
The mesh equations are
[
2s +2
-2-k
-2J [11(s)J
3 12(s)
= [V (S)+2i0 l (0-)J
I (l3.30a)
From (13.29) we see that the current through each element in the
circuit has as its poles the poles of the circuit PI and P2 determined The determinant is
above. The same is also true of each voltage in the circuit, as can Ll(S) = (2s + 2)(3) - 2(2 + k) = 6s + 2(1 - k) (13.30b)
easily be verified by substituting the mesh currents above into the
element laws for each element in the circuit. So there is a single circuit pole located at p = ~(k - 1). For k < 1,
FIGURE 13.14 Circuit for Example 13.10. this pole has negative real part. We conclude that for k < 1 this
Performing nodal analysis on this same circuit, the only un-
known node voltage is Yes), and summing currents at this node gives circuit is stable; for k ~ 1, it is not stable.
~ [yes) - vc(O-) J + ~ Yes) + ~s[V(s) - 2il (0-)] = 0 Stability has several implications with respect to the behavior of a circuit. Here we
8 s 4 2 point out two that are of immediate importance: the relationship between stability and
V(s) _ sVc(O-) + 8il (0-) steady-state and that between stability and transfer function analysis. We earlier defined
- s2 + 2s +4 steady state as the condition in which all transients had decayed to zero and the overall
response of the circuit was equal to its forced response. But what if the transients in
Mesh and nodal analysis arrive at the same set of poles for the
a circuit never tend to zero? The circuit will never attain steady state, and steady-state
circuit, as they must.
analysis, including phasor analysis of ac steady state, will not determine the long-term
values of currents and voltages. Steady-state analysis is applicable only to stable circuits.
2ir(O-)
2H 2s Yes) The long-term values of currents and voltages in unstable circuits depend strongly on
their initial conditions. There is no steady state for such circuits.
~~
3 3
s Determine the ac steady-state value of i l (t) in the circuit shown in
4>2 !p Vo(s) 4 Fig. 13.14 for the source VI(t) = 40cos2t V. Applying Cramer's
8
+ vc(O-) rule to solve (13.30a) gives
s
1VI (s) +02i l (0-) ~21
II (s) = 1 2S + 2 -21
(a) (b)
-2-k 3
FIGURE 13.13 (a) Circuit for Example 13.9; (b) in the s-domain. (13.31)
536 Chapter 13 Circuit Analysis in the s-Domain Section 13.4 Poles and Stability 537
is unstable, the total response will differ from the forced response determined by transfer
The single pole of the circuit is at p = ~(k - 1). If k ~ 1, the pole
function analysis by a term (the natural response) that may well be going to infinity.
is not in the left half-plane and the circuit is unstable. For instance,
We conclude that the property of stability has important bearing on the usefulness of
with k = 4 the natural response [first term in (13.31)] is i[(O-)e+ t ,
both steady-state and transfer function analysis. In circuits that are not stable, steady-state
which certainly does not decay to zero as t gets large. For k ~ 1, the
analysis does not apply, and transfer function analysis, being limited to determining the
circuit never "settles down" to a steady-state value; steady state does
forced part of a response, may yield results quite different from the actual total response.
not exist in this circuit for k ~ 1. For k < 1, the natural response
Stability may be conveniently determined in the s-domain by examining the real parts of
does go to zero, regardless of the initial condition, and we will have
the circuit poles, which must all be negative for the circuit to be stable. The poles of the
steady state. Using the source V4 (s) = 40S/(S2 + 4), the partial
circuit are the zeros of the determinant of the nodal or mesh analysis matrix.
fraction expansion for the forced response [second term in (13.31)] is
20s A B B*
[s+~(l_k)](s2+4) = s+~(l-k) + s-j2+ s+j2 EXERCISES
For k < 1, the first term on the right will decay to zero with time,
so since only steady state is required, we need not compute A. 13.4.1. Show that a simple two-element parallel LC circuit is not stable.
Answer Circuit poles at j-Jl/LC do not have negative real
B = 20(j2) _ 10
[j2+~(l-k)](j4) - ~(l-k)+j2 [. t
"2 Ie
2F
parts.
13.4.2. Determine the poles of the circuit. Is this system stable?
Answer -1, -4; yes
and the required steady-state value for any k < 1 is
13.4.3. What is the largest value of k for which this circuit is not stable?
i[ (t) = 21BI cos(2t + e) Answer -4
where EXERCISE 13.4.2
6
e= -arctan--
1-k
The inapplicability of steady-state analysis for circuits that are not stable suggests
that we ought to investigate stability before launching a steady-state calculation. This can
be done by determining if all the poles of the circuit have negative real parts as described
above. In practice, circuits lacking controlled sources are almost always stable. Currents
EXERCISE 13.4.3
and voltages due to initial conditions can build indefinitely only if there is an element
supplying power to the rest of the circuit, which cannot be an RLC element (which we
know never supplies average power). Since independent sources are killed in determining
the natural response, the only remaining candidate is a controlled source. The only circuits
lacking controlled sources that may not be stable are those containing isolated resistance-
free LC loops, which correspond to circuit poles exactly on the jw axis [Re(p) = 0].
Finally, consider the implications of stability in the use of transfer functions. The We next develop two useful theorems that permit us to deduce the initial value of a
transfer function H(s) permits calculation of an output Vo (s) for a given input Vi (s) by time function, J(O+), and its final value, J(oo), from its transform F(s). Consider any
the relationship J(t) that has no impulses or higher-order singularities at t = O. Write J(t) in the form
J(t) = get) + duet), where get) is continuous at t = 0, and d = J(O+) - J(O-) is the
Vo(s) = H(S)Vi(S) (all ICs = 0)
size of the jump discontinuity in J(t) at t = O. Differentiating gives
Suppose, however, that the initial conditions in the circuit are not exactly equal to zero.*
To what extent is this relation still applicable? If the circuit is stable, the natural response dJ = dg + do(t)
dt dt
will decay to zero. The forced response determined by the equation above will approach
the actual total response to arbitrary accuracy as t -+ 00. On the other hand, if the circuit By the derivative rule applied to this equation, we obtain
'Which will always be the case for physical circuits, if due only to unavoidable thermal noise voltages
sF(s) - J(O-) = 1
0-
00 dJ st dt
_e-
dt
= 1 00
0-
d
'-!e-
dt
st dt + 100
0-
do(t)e- st dt (13.32)
and currents.
relative to these sources. Find vo(O+) and vo(oo) if oCt) is applied at each
The condition of no impulses or higher-order singularities is satisfied, for instance, by input and all initial conditions at 0- are zero.
any proper rational function F(s). Answer +6; 23
Next consider (13.32) as s -+ O. The last integral is still d, and the integral
summing with it is
lim
s---+o
10-
00 dg
_e- st dt
dt
= 1-
00
0-
dg
dt
dt = g(oo) - g(O-)
The last equality requires that g(oo) exist for the given function. Then if g(oo) exists, The transfer function H(s) permits calculation of an output V 0 (s) for a given input Vi (s)
lim sF(s) - f(O-) = g(oo) - g(O-)
s---+o
+d (13.33) by the relationship
Vo(s) = H(S)Vi(S) (all ICs = 0) (13.34)
Since f(t) = get) +du(t), then g(oo) = f(oo) -du(oo) = f(oo) -d. Substituting into
(13.33) and noting that f(O-) = g(O-), we have the following theorem. If we select the input Vi (t) to be the unit impulse function,
Viet) = 8(t) (13.35)
fiNAL VALUE For any f(t) such that the limit f(oo) exists,
then since the transform of 8(t) is the constant 1, the response under these conditions
THEOREM will be simply
Vo(s) = H(s)(I) = H(s) (13.36a)
or in the time domain
Vo(t) = L- 1 [H(s)] = h(t) (l3.36b)
I(s) = (7s 3 + 2)/(S4 + 3s 3 + 5s 2 + 2s + 2). Find the initial value The time function h(t), being the output caused by a unit impulse input when all initial
i (0+) and the final value i (00). As s -+ 00 the highest power terms conditions equal zero, is reasonably enough called the unit impulse response or sometimes
dominate both numerator and denominator, while the constant terms just the impulse response. By (13.36b) we see that the unit impulse response h(t) and
dominate as s -+ O. Thus the transfer function H(s) are a Laplace transform pair. Since all practical circuits are
lim sICs) = i(O+) = 7, lim sICs) = i(oo) = 0 causal, that is, they cannot begin responding to an input until that input itself begins, we
s---+oo s---+o will require that the impulse response be zero in negative time; h(t) = 0 for t < O.
A principal benefit of these two theorems is that we do not have to compute Find the transfer function H(s) and unit impulse response h(t) for
an inverse transform in order to discover its initial and final values. In the foregoing the input VS1 (t) and output VL(t) for the circuit of Fig. 13.15.
example, for instance, this would require factoring a quartic to discover the poles and The s-domain circuit is shown, prepared for computing H(s)
then computing the partial fraction expansion numerators. These difficulties are bypassed by killing the other independent source and setting initial condition
if only initial and final values of the inverse transform are needed. generators to zero. The equations for nodal analysis in this circuit
540 Chapter 13 Circuit Analysis in the s-Domain Section 13.6 Impulse Response and Convolution 541
by 9, we have
(s + 4)e- 9s
L{e-4(t-9) cos[7(t - 9)]u(t - 9)} = --...,.---
[s +4]2 +49
and since vo(t) = e- 36 {e- 4(t-9) cos[7(t - 9)]u(t - 9)},
(s + 4)e-9(s+4)
V (s) - ..2.-----=_ __
o - (s + 4)2 + 49
The transfer function H (s) is the ratio of output to input under these
conditions:
V ( ) 1 (s + 4)e- 4s e- 36
R(s) = ~ = -"----::----,--
-3
L...:....-----=>e V2(s)
18 H 18 s
VieS) (s + 4)2 + 49
(a)
To get h(t), begin again from the table entry for damped cosine,
(b)
time shifted by 4:
fiGURE 13.15 (a) Circuit for Example 13.13; (b) s-domain circuit with (s + 4)e-4s
the other source killed. L{e-4(t-4) cos[7(t - 4)]u(t - 4)} = ~--=-
(s+4)2+49
are Comparing this with R(s), we must amplitude scale by the factor
VI(S) + [VI(S) - VsI(s)] + [VI(S) - V 2 (s)] =0 (13.37a) e- 36 /3:
h(t) = ~e-4(t+5) cos[7(t - 4)]u(t - 4) (13.39)
IL V 2(s) + [V2(s) - VI (s)] + 2s[V2(s) - VsI(S)] =0 (13.37b)
We see that in this circuit the response h(t) to the unit impulse oCt)
Multiplying (13.37b) by 3 and adding to (13.37a) eliminates VI(S): does not begin until time t = 4, even though oCt) itself "fires" at
Since the impulse response h(t) and the transfer function R(s) are a transform pair,
R(s) = V2 (s) = _s_ we may use knowledge of one to gain the other. As we saw in Example 13.14, an input
VsI(s) s+~ other than the impulse may be used, the transfer function found, and from it the impulse
To determine the impulse response h(t) we need the inverse trans- response. In the case of physical circuits, it is usually more convenient to input a unit
form of the transfer function R(s). After long division, step than a unit impulse. Then from the resulting unit step response set), both R(s) and
h(t) may be computed. We define the unit step response set) to be the output when a
I
R(s) = 1- ~ unit step input u(t) is applied, assuming zero initial conditions and all other independent
s+!6 (13.38) sources killed.
The unit step response set) and its transform S(s) are closely related to the unit
h(t) = oCt) - ~e-tI6u(t) impulse response h(t) and its transform R(s). With unit step input Vi (s) = lis in
Note that we have multiplied by u(t) since we require h(t) = 0 for (13.34), the output Vo(s) = S(s), and we have from that equation
t < O.
When the input Vi (t) = 30(t - 5) is applied to a circuit with zero R(s)
S(s) = - (13.40)
initial conditions and all other independent sources killed, the output S
is measured to be
Vo(t) = e- 4t cos[7(t - 9)]u(t - 9) Using the integration rule on (13.40) gives
Determine the transfer function R(s) and impulse response h(t).
Transforming the input viet) gives
VieS) = 3e- 5s
set) = 1~ her) dr (l3.41a)
Using the Table 12.2 entry for the damped cosine and time shifting
542 Chapter 13 Circuit Analysis in the s-Domain Section 13.6 Impulse Response and Convolution 543
Since h(t) = 0 for t < 0, the lower limit may be extended to -00. From this it is clear Taking G(s) out of the integral, what remains is F(s), and we have the following prop-
that set) = 0 for t < 0 just as h(t) does. The step response cannot begin before the step erty:
input begins. Finally, differentiating both sides, we have
CONVOLUTION
d
h(t) = -set) (13.41b)
L [I: f(i)g(t - i) dtJ = F(s)G(s) (13.43)
dt
The Laplace transform of the convolution of two time junctions is the product of their
The impulse response is the derivative of the step response (the step response is the integral Laplace transforms. Then, comparing (13.34) with (13.43), we have
of the impulse response). This result could have been anticipated by noting that the unit
step is the integral of the unit impulse, and by linearity, the response of a linear circuit
to the integral of an input is the integral of its response to the original input.
L [I: h(i)Vi(t - i) dtJ = H(S)Vi(S) = Vo(s) (all ICs = 0) (13.44)
d(t) = i:
produces a third time function d(t) via the integral
f(i)g(t - i) di
We refer to d(t) as the convolution of f(t) and get). Here we will restrict ourselves to
(13.42)
voltage division:
V 2 (s) [3(15Is)]/[3 + (I5Is)]
H(s) = Vs(s) = {[3(I5Is)]/[3 + (I5Is)]} + 5
3
s +8
So the impulse response is
convolutions between functions f(t) and get) that are both 0 for t < O. Transforming
yields h(t) = 3e- 8t u(t)
where the inner lower limit has been replaced by 0-, since f(i) o for i < O.
Reversing the order of integration, we obtain +
30 v2(t)
D(s) = 1~ f(i)[e-srC(s)]di FIGURE 13.16 (a) Circuit for Example 13.16; (b) input vs(t)
544 Chapter 13 Circuit Analysis in the s-Domain Section 13.6 Impulse Response and Convolution 545
V2(t) = i:
Convolving h(t) and vs(t), the desired output is
h(t - r)vs(r) dr
V2(t) =
o
it 3
3e- 8(t-')e-'dr = _(e- t _ e- 8t )
7
same figure her) is included, identical to h(t) with t replaced by r.
As t increases, the square pulse Vi (t - r) moves from left to
right along the r axis. We may identify several distinct regions,
A sketch of this function is shown as Fig. 13 .17. depending on the type of overlap between viet - r) and her). The
first region corresponds to locations of the square pulse (i.e., values
of t) for which the pulse is entirely to the left of the nonzero range
of values of h (r). There is no overlap for any r, so the product of
these two functions, hence the area under the product curve, is zero.
The convolution vo(t) = 0 for all t in this region, which is marked
by the rightmost edge of the square pulse being to the left of the
origin in Fig. 13.19(b), or
4 --------------------------
FIGURE 13.17 Output V2(t) for Example 13.16.
For each fixed instant of time t, the convolution (13.45) equals the integral over
all r of the product of the two r-functions which make up its integrand. Since the value
of any real integral is just the area under its integrand, we may interpret a convolution
integral as the area under its integrand product curve as a function of the parameter vi(-r)
t. It is often helpful to visualize the process of evaluating this area by fixing t and
sketching (versus r) a graph of each of the two functions in the integrand, Vi (t - r)
and her). Only in r-regions where their nonzero values overlap will the product, and -L--------~----4L------------------~4~'
-3 -1
thus the contribution to the convolution, or area under the product curve, be nonzero. t-3 ~t t-l
(a) (b)
This is sometimes referred to as the graphical method for evaluating a convolution and
is illustrated in the final example in this section. FIGURE 13.19 (a) vi(-r); (b) Vi(t - r) and h(r) vs. r.
546 Chapter 13 Circuit Analysis in the s-Domain Section 13.6 Impulse Response and Convolution 547
In the next region t - 1 > 0, so there is overlap, but t - 3 < 0, Fig. 13.20(c), so for t - 1 > 4 but t - 3 < 4,
so the overlap is partial. In this case the product curve is shown in
Fig. 13.20(a), and its triangular area is given by area = [4 - (t - 3)](t - 3) + H4 - (t - 3)]2
As the right end of the square pulse moves past the triangle h (r ), 13.6.1. A circuit has impulse response h(t) = 2[u(t -1) - u(t - 2)]. Find
the overlap is partial once again and the product is the trapezoid of the zero initial condition response vo(t) to the input Vi (t) = (t - l)u(t) by
convolution.
Answer 0 for t < 1; 2t - 5 for 1 ::: t ::: 2; t 2 - 4t + 3 for t > 2
v/t - r)h(r)
13.6.2. Suppose that two copies of the circuit of Exercise 13.6.1 are con-
nected in cascade so that the output from one becomes the input to the other.
Find the impulse response of the cascaded circuits.
vi(t - r)h( r) t-l ----------------, Answer 4[r(t - 2) - 2r(t - 3) + ret - 4)], where ret) is the unit
:
........... ramp
13.6.3. Verify the result of Example 13.16 by finding V2(S) from the
t-l ----- transfer function H(s) = 3/(s + 8). What is the required source Vs(s)?
t- 3 1- e- 2 (s+l)
Answer Vs(s) = -----
, s+1
--+---~------~----4~r
t- 1 4
(a) (b)
viet - r)h( r) vo(t) In Chapter 8 we saw that the use of phasors greatly simplified electric circuit analysis
4 -------------------- 6 ---------------- where applicable but was restricted to finding the forced response in sinusoidal circuits.
Here we see that the s-domain reaps the same benefits but is applicable to all circuit
t- 3
problems, forced or unforced, steady state or transient.
KVL, KCL, and Ohm's law carry over from the time to the s-domain unchanged.
o The impedance Z(s) of an RLC element or subcircuit is the ratio of its s-domain voltage
and current when all initial conditions are zero.
--~----------L---~r
t- 3 4 3 7 The impedance of the RLC elements is R, sL, and I/sC, respectively.
(e) (d)
The s-domain circuit is the original circuit diagram with the RLC elements labeled by
impedance and independent sources labeled by their transforms.
FIGURE 13.20 (a) Integrand for the region 1 < t < 3; (b) integrand for
region 3 < t < 5; (c) integrand for region 5 < t < 7; (d) convolution The s-domain circuit may be analyzed as if it were a dc resistive circuit, with impedance
vo(t) for all regions. treated as resistance, and s-domain source functions treated as constants.
IiiiiI ~3;'5. Find Ve (t), t > 0, if Ve (0-) = 1 and h (0-) = fiGURE P13.31
fiGURE P13.21 13.34. Find the transfer functions between output vz(t)
IQ and inputs VsI(t) [H1(s)], isZCt) [Hz(s)]. iDl13.39. Repeat Problem 13.38 if VsI(t) = te- t V, vsz(t) =
13.22. Draw the s-domain circuit. Combining parallel el- IH I!mI 0, the initial inductive currents are both zero, and ve(O-) =
ements, write and solve a single mesh equation for I(s). +12 V.
fiGURE P13.25 13.40. Find all the circuit poles. Is this system stable?
sin t u(t) V
ZH ZH
2Q
+
1iiii113.26. Find vej(t) and ve2 (t) for t > O. II~
4 Q v2(t)
IQ
cos t u(t) A
+ fIGURE P13.34
I/LF vC2 2vc, rnA
i(t) + vc(t)_ 13.36. In a certain circuit the zero in initial condition out-
o~
+ I fiGURE P13.27
put vo(t) for t > 0 is measured to be 1 + 3te- Zt - e- Zt
.iF when the input is Vi(t) = 5u(t). What is H(s)? What is
4
v(t) IH the forced response vo(t), t > 0, for Vi (t) = e- 5t u(t)?
13.28. Repeat Problem 13.27 for the transfer function fiGURE P13.41
13.37. In the circuit described in Problem 13.36, there is
VZ(S)!VI(S). a single storage element, a capacitor. When ve(O-) =
13.29. Repeat Problem 13.27 for the transfer function 1 V and the input Vi (t) = 0, the output is found
fiGURE P13.23 VZ(S)!ll (s). to be (t - l)e- Zt V. Using the data given in Prob- 13.42. Find a value for R for which this circuit is
13.30. Find H(s) = Vz(s)!Vs (s) for the circuit of Prob- lem 13.36 also, find vo(t), t > 0, if Vi(t) = 5 sint and stable and another for which it is unstable. Assume
1iiii113.24. Find i (t), t > O. lem 13.4. ve(O-) = 4 V. R~O.
rr,4i','
i(t) 1 rI. 1H f(t) get) 2e-' A
'0
I rI.
2
.-------~~~------~
=2<V ~~" 2 3 0
;;. t FIGURE P13.63
FIGURE P13.54
FIGURE P13.43
+
v
13.55. Show that the convolution of df(t)/dt and get) 13.64. Sketch a circuit whose impulse response is h(t) =
13.44. Find the poles of the circuit of Problem 13.26 and is the same as the derivative of the convolution of f(t) 8' (t) + e-t u(t),
specifying which circuit voltage is the out-
determine if the circuit is stable. and g(t). put, and using an independent voltage source as the input.
FIGURE P13.61 Repeat using an independent current source as the input.
D 13.45. Find the ac steady-state responses il(t), V2(t), of 13.56. Determine the unit step response set) = Ve(t) to
IIillI the circuit of Problem 13.31 if Vs(s) = 1/(s2 + 1). the input vet) = u(t) in the circuit of Problem 13.23. Use
convolution to compute set).
13.46. If a circuit is stable, its impulse response must de-
cay to zero as t --+ 00. Show why this is true. 1Q113.57. Find. the tran.sfer function ":(s) with .V2(t) as
13.47. If a circuit is not stable, its impulse response may
IIillI output and II (t) as mput (Hint: Fmd both m terms
of Vs (s); then divide). Determine the impulse re-
still decay to zero as t --+ 00. Give an example of such a
sponse h(t) and the forced response to the input il (t)
circuit. 2(1 - e- t ).
13.48. F(s) = (3s + 1)/(s2 + 7s + 6). Use the initial and
final value theorems to find f(O+) and f(oo), and verify
by finding f(t) by partial fractions. 2F
13.49. F(s) = _s2 /(s + 1)3. Use the initial and final value
I rI.
theorems to find f(O+) and f(oo), and verify by finding
f(t) by partial fractions. +
13.50. Find v(O+) in Problem 13.21 by use of the initial
IH 2r1. V2
value theorem.
13.51. Assume Z(s) is a rational function. Under what
conditions Z(s) will the current i(t) be continuous at
t = O? FIGURE P13.57
I"
I
Heinrich Hertz
557
Chapter Contents
II1II 14.1 Frequency Response Function III 14.6 Filters
The transfer function H(s) permits us to compute the s-domain output for any input via
l1li 14.2 The Decibel Scale iIIII 14.7 Scaling
the basic input-output equation (13.8), repeated here.
III 14.3 Bode Gain (Amplitude) Plots II1II 14.8 SPICE and Frequency Response
Vo(s) = H(s)Vi(s) (all ICs = 0) (14.1)
III 14.4 Resonance II1II Summary
We apply this simple but general relationship to phasor analysis, which was first in-
III 14.5 Frequency Response of Op II1II Problems ~oduced in Chapter 8 in the context of ac steady state. Suppose that the time-domain
Amps mdependent sources in a circuit are each complex exponentials Viej{J)t at the same fre-
quency w, as required for phasor analysis. Vi is the phasor associated with the source
VieJ{J)t. Since
L(Viej{J)t) = ~
s -JW
then
Vo(s) = H(S):i (14.2)
s - JW
Si.nce the li~ear circuits we consider contain only fixed RLC elements and sources, H(s)
wIll be a ratlOnal function
H(s) = N(s) = N(s)
(14.3)
D(s) (s - pd rl ... (s - Pn)'.
where the Pi are poles of the circuit, as discussed in Section 13.4. Then
Vo(s) = N(S)Vi
Since sinusoidal currents and voltages are easily produced, one convenient way to test (s - PI)'l ... (s - Pn),n(s _ jw) (14.4)
a linear circuit is to inject a sinusoid as input and observe the sinusoidal steady-state
Ass~ming that there are no poles on the jw axis, Re(Pi) =1= 0 for each i = 1, ... , n, by
response. Changing the phase or amplitude of the sinusoid will not test the circuit fur- partial fraction expansion
ther, since the response will predictably scale with the amplitude and phase shift along
with the input. At each possible input frequency, however, the response of the circuit, Vo(s) -_[ A B
+ (S-PI)'l-1 + ... + - Q- ] + -R- (145)
both amplitude and phase, will in general be different. These response variations with (S-PI)'l s-Pn s-jw .
frequency form the frequency response of the circuit. where
We may be acquainted with some aspects of frequency response from the manner in
which audio equipment is often advertised. A speaker whose "3-dB bandwidth" is, say, (14.6)
50 Hz to 20 kHz is better than one of bandwidth 200 Hz to 5 kHz, since it will reproduce
bass and treble frequencies better. In this chapter we introduce the concept of frequency which, by (14.4), evaluates to
response, using it first for analysis and then as the key descriptor of what a linear circuit R = H(jw)Vi (14.7)
"does," the function that it is designed to perform and the quality with which it performs
Examining the (14.5), vo(t) will, in general, contain the term Rej{J)t together with terms
that function. Frequency response is the critical link between the analysis and design of
due t? th~ p~les of cir~uit shown in brackets. Recall that phasor analysis is applicable
linear circuits.
First we introduce the frequency response function H(j w) and show its close rela- only ~n CIrCUIts for WhICh all currents and voltages are of the form Re jwt . This is the
tionship with the transfer function H(s) of Chapter 13. The decibel scale, a convenient case m (14.4) if ~~e cir~uit is stable and if we are considering the steady-state response
vo(t) only. StabIht'y' sIgnaled by negative real parts of all the poles Pi of the circuit,
way to measure nonnegative numbers, is defined and used to produce Bode gain plots,
which describe the changes in the input-output amplification ratio (gain) with frequency. guarantees th~t th~ mvers~ transform of each term in parentheses in (14.4) decays to zero
as t goes to mfimty. This leaves, in steady state, only the final term which by (145)
Using these tools, the frequency response of resonant circuits and of op amps is next
and (14.7) is the forced response , .
investigated, and the very important class of frequency-selective circuits called filters
studied. The chapter concludes with a discussion of frequency response and the use of vo(t) = L -I H(jw)Vi = H(jw)Viej{J)t (14.8)
SPICE. s -jw
558 Chapter 14 Frequency Response Section 14.1 Frequency Response Function 559
By definition, the phasor Vo associated with this output vo(t) is the quantity multiplying The source frequency is w = 1 rad/s, the source phasor is Vs =
ej(j)t, or, by (14.8), 6!S1 V, and, by the frequency response equation, the output phasor
I equals the input phasor scaled by the frequency response function
evaluated at w = 1, or
Vo = H(jw)V i (14.9)
1= H(j1)Vs = j1 6!S1 A
5(jI) + 2
H(jw), the transfer function H(s) with s replaced by jw, scales the input phasor to yield Using a calculator once again, we have
the output phasor. We define H(jw) as the frequency response function. Then (14.9),
1= (0.186(21.8)(6!S1) = 1.12(21.8 A
which we refer to as the frequency response equation, asserts that for stable circuits,
the output phasor is the product of the input phasor and the frequency response function whence
H (jw). Given H(s), or equivalently H(jw), we can determine the sinusoidal steady-state i (t) = 1.12 cos(t + 21.8) A
output of any stable linear circuit at any frequency. Our restriction in (14.5) that there
which is the required steady-state current.
are no j w-axis poles does not further limit this result, since such systems are not stable
and thus have no steady state. Taking magnitudes of both sides of the frequency response equation (14.9),
560 Chapter 14 Frequency Response Section 14.1 Frequency Response Function 561
The circuit has a single pole at s = -lIRC, so it is stable. We
R
may replace s by jw and determine the steady-state behavior of this
EXERCISES
t i(t) circuit via its frequency response function H(jw). The gain as a
14.1.1 For H(s) = (s + 1)/(s2 +s +3), what are the gain and phase shift
+
function of frequency is given by
at w = 2 rad/s? At f = 1 Hz?
C v(t)
. 111RCI 11RC (14.14) Answer 1.00, -53.1; 0.172, -89.3
IH(lW)1 = Ijw + (lIRC)1 = Jw2 + (1IRC)2 14.1.2 For H(s) = -2/(s + 4), find all frequencies w > 0 at which the
which is shown in Fig. 14.3(a). Note that the dc gain is unity and
amplitude of the output is 1 when the input amplitude is 2. Repeat for
phase shift +114 .
declines with increasing frequency. The impedance of the capacitor Answer One w for each: 6.93 rad/s, 8.98 rad/s
FIGURE 14.2 Circuit for Example 14.3.
is infinite at dc, so as the frequency increases the voltage divider will 14.1.3 In the circuit of Fig. 14.2, if the output is defined to be the voltage
completely favor the capacitor at w = O. Its impedance declines to across the resistor, find the gain.
zero with increasing frequency, so as frequency increases the voltage w
Answer -r=======
divider ratio shifts steadily toward the resistor and away from the C. Jw 2 + (1IRC)2
These facts are consistent with the gain shown in Fig. 14.3(a). The
phase shift, shown in Fig. 14.3(b), is given by
IH(jw) = -tan- l wRC (14.15)
To use this scale to measure gain, note that the ratio of output amplitudes A2 and Al
FIGURE 14.4 Gain in dB for Example 14.4.
produced by the same input at two frequencies whose gains are IH(jW2)I and IH(jwI)1 is
A2 IH(jW2)I There are a few numbers R whose decibel equivalents are worth remembering.
= (14.19)
Al IH(jWI)1 The number 1 measured in decibels is clearly 0 dB. The ratio 10:1, or number R = 10,
measured in decibels is
so the ratio of their powers P2 and PI must be as the square of the gains,
10 (dB) = 20 log 10 10 = +20 dB (14.24)
IH(jW2)12
(14.20) Note that when dB is put in parentheses after a number R, it is to be read as "the number
IH(jwd1 2
R, when converted to decibels, is," while dB next to a number without parentheses indi-
Then, applying (14.20) to (14.18), we have cates that the value already has units of decibels, such as +20 dB in (14.23). The implica-
tion of (14.24) is that multiplication by 10 always adds 20 dB. This follows since the prod-
_ 10 IH(j w2)1 2 (14.21)
gain (dB) - 10 glO IH(jWI)12 uct of two numbers when converted to decibels, equals the sum of their decibel equivalents.
RS (dB) = 2010g lO (RS) = 20 log 10 R + 2010g I0 S
. _ 10 IH(jW2)1 (14.22)
gam (dB) - 20 glO IH(jWI)1 RS (dB) = R (dB) + S (dB) (14.25)
The graph is shown in Fig. 14.4. Note that there is 3-dB difference 2 (dB) = 20 10glO 2 = 6.02 dB ;;:: 6 dB (14.27)
between the gains at w = 0 and w = RC and also 3-dB between So multiplication by 2 adds 6 dB and division by 2 subtracts 6 dB. Finally, since raising
w = 100/RC and w = 100-v'2/RC. The latter gains are so small that !
to the power scales the log by !,
multiplication by .j2 adds 3 dB, and division by .j2
they are very hard to make out on a linear scale [see Fig. 14.3(a)]. subtracts 3 dB.
564 Chapter 14 Frequency Response Section 14.2 The Decibel Scale 565
The number 84.4 measured in decibels is 20 10glO 84.4 = 38.5 dB United States in the 1930s, devised a simple but accurate method for graphing gain and
(to three significant digits). Then phase-shift plots. They bear his name, Bode gain plot and Bode phase plot (pronounced
boh'-duh, not the frequently heard bah-dee or bawd).
844 (dB) = 38.5 dB + 20 dB = 58.5 dB Given a transfer function R(s), consider its factored form in which numerator and
844,000 (dB) = 38.5 dB + 4(20 dB) = 118.5 dB denominator of R(s) are factored separately:
14.2.1 Without using a calculator, determine 0.001 (dB), 0.004 (dB), and
500J2 (dB). Then verify with a calculator. Each pair of complex conjugate poles or zeros corresponds to a quadratic factor in R(s).
Answer -60 dB; -48 dB; +57 dB The coefficients are unity (for s2), and we label the s-coefficient a and the constant b.
14.2.2 If an amplifier section has gain of + 18 dB, how many must be For each quadratic factor, define two parameters
connected back to back (in cascade) to lift a I-mV amplitude input to 250-V
amplitude?
Answer 6 (l4.29b)
14.2.3 If the input and output phasors at a certain frequency w are 320/30
and 210/ -41 , respectively, what is the gain in decibels at this frequency?
Answer -3.66 dB
(l4.29c)
or (14.30)
The frequency response function R(jw) reveals the steady-state behavior of a circuit at
any single frequency and, by superposition, due to a sum of inputs at different frequencies. wn is called the natural frequency of the complex root pair factor (14.29a), and ~ is its
It is very useful to be able to visualize the dependence of R(jw) on frequency. Where in dampin~ factor. Th~ resulting form for R(s), after recombination of complex root pair
the frequency domain, for instance, is the gain strong and where is it weak? An input sum factors mto quadratIc factors (14.29a), will be called the standard form. We begin the
of two sinusoids may have its two components equal in amplitude coming into a circuit, process of Bode plotting by putting R(s) in standard form.
but in the output the sinusoid whose gain is high will dominate one whose frequency
corresponds to low gain. One is effectively blocked and the other passed. This type of Put the transfer function H(s) = (2s + 6)/(s3 + 2s2 + 2s) into
behavior can be told at a glance from the gain curve. standard form. First factoring, we find one zero and three distinct
An exact plot of gain versus frequency, whether gain is measured on a linear poles each of order 1,
scale or in decibels, is somewhat tedious to produce, as can be seen from (14.14) and
(14.23). Among his several important contributions to circuit theory and control theory, R(s) = 2(s + 3)
the German-born engineer Hendrick Bode (1905-1982), while working at Bell Labs in the s(s + 1 + j)(s + 1 - j)
566 Chapter 14 Frequency Response Section 14.3 Bode Gain (Ampl itude) Plots 567
Recombining the conjugate terms yields w a, Ijw + al may, with vanishingly small error as wla --* 00, be replaced by liwl,
which is just w itself. So
2(s+ 3)
H(s) = S(S2 + 2s + 2)
INi(jw)1 (dB) ~ 20rlog lO Ijwl = 20rlog lO w, wa (14.37)
We may identify a constant K = 2, a single zero at s = -3, a pole
at s = 0, and a quadratic factor with natural frequency Wn = -Ji The graph for w a will take on a particularly simple form if we choose to plot gain
and damping factor I; = -Ji12. not versus w but versus 10glO w. In this case, (14.37) indicates that we have an equation
of the form y = mx with y the gain in decibels and x the log base 10 of w, that is,
In standard form, the transfer function is a ratio of products, a simple linear relationship between these variables. The graph for w a will be a
straight line with slope m of 20r dB gain change per unit change in the log base 10 of
KNl(s) N",(s) w. A change of one unit in 10glO w corresponds to a change of a factor of 10 in w itself,
H (s) = -~'-------"-'----- (14.31)
Dl (s) ... D1](s) so we refer to this slope as 20r dB per decade.
The two straight-line segments for w a and w a meet at w = lal, as can be
where, after recombination of complex factors, each Ni (s) and Di (s) is either of the form
seen by comparing (14.36) and (14.37). For this reason, w = lal is referred to as the
(s + a Y with a real, a real root factor, or (S2 + 21; wns + w~y, a complex root pair factor.
break frequency. A plot of the piecewise linear approximation to the gain curve that we
Replacing s by j wand taking magnitudes, the gain is
have determined, a pair of straight lines meeting at the break frequency, is called the
IKIINl(jw)I .!N",(jw)! uncorrected Bode gain plot and is shown in Fig. 14.5 along with the exact or corrected
(14.32) Bode gain plot computed by evaluating (14.35). Note that the maximum error occurs
IH(jw)1 = IDl(jw)I" .!D1](jw)!
at the break frequency and is 3r dB. Far from the break frequency the uncorrected and
By (14.25), the decibel value of a product is the product of decibel values, so corrected plots merge smoothly.
The plot of the same factor, (s + aY, when moved to the denominator of H(s),
simply flips around the horizontal axis. This follows, since the decibel value of l/X is
the negative of the decibel value of X. The initial value is then -[Ial (dB)], the negative
+L L [lDi(jw)1 (dB)]
'" 1]
IH(jw)1 (dB) = IKI (dB) [INi(jw)1 (dB)] - (14.33) of the decibel equivalent of lal, and the slope after the break is -20r dB/decade.
i=l i=l Note that there is a special case of the real root factor when the pole or zero
a = O. In this case the form given in (14.37) is not approximate for w a but is
exact for all w. This plot passes through the value 0 dB at w = 1. The plot for
The strategy for plotting the gain IH(jw) I in decibels will be to plot each term on the this factor is shown in Fig. 14.6. The technique for using these component graphs to
right of (14.33) separately and then add these component plots graphically. The result generate the Bode gain plots, both uncorrected and corrected, is illustrated in Exam-
will be the desired gain in decibels. There are only two kinds of terms, real root terms ple 14.7.
and complex root pair terms, to learn how to plot. More precisely, there is a third, the
constant term IKI (dB), but the graph of this term will clearly be a flat straight line at
the level 2010g l0 IKI.
Turning first to the real root case, assume for a moment that this term is in the Gain (dB)
numerator. Then
INi(jw)1 (dB) = 20 log 10 INi(jw)1
(14.34)
= 2010g lO Ijw + air
where r is the multiplicity of the real zero -a. This can be simplified slightly to
For w a, Ijw + al may, with vanishingly small error as wla --* 0, be replaced by la!- lal (dB) I---""""==:::::::=----e--
The gain of this term is then - J - L - - - - - - L - - - - _ L -_ _ _---L_ _ _ _.....L~ ()) (log scale)
lal lal lal 10lal
INi(jw)1 (dB) ~ 20rloglO lal, wa (14.36) 100 10 ~ 1 decade I<E------
Thus, for w a, the graph will be a flat straight line at the level 20r 10glO lal. For FIGURE 14.5 Bode gain plot for the real root numerator factor (s + a)r.
568 Chapter 14 Frequency Response Section 14.3 Bode Gai n (Ampl itude) Plots 569
Gain (dB)
Gain (dB) Gain (dB)
~
OdB/dec o dB ---jl--~l.,-----_ _ _--:;;.. UJ (log scale)
+6 dB
2 w (log scale)
Let us determine the uncorrected and corrected Bode gain plots for (b) _1_ factor (d) Constant factor
H(s) = (-5s + 1O)/(s2 + 20s). In standard form, (s + 20)
There is a zero at s = -2, with poles at s - 20 and s = O. ponent graphs, all values are known at w = 1. Adding, the gain
The real factor (s + 2) has its break frequency at w = 2 and slope in decibels at w = 1 is +6 - 26 + 0 + 14 = -6 dB. The gain at
20 dB/decade after the break (r = 1). The initial level value of this the break frequency w = 2 is then the value at w = 1 reduced by
factor is 121 (dB) = 2010g lO 121 = 6 dB. The real denominator fac- - 20 dB/decade times the number of decades between frequencies
tor (s + 20) has its break frequency at w = 20, slope -20 dB/decade 1 and 2. Since decades measure the factors of 10 separating two
after the break, and initial level value -2010g lO 20 = -26 dB. The frequencies,
factor of s in the denominator has a slope of -20 dB/decade ev-
erywhere and passes through 0 dB at w = 1. Finally, the con-
stant factor K = -5 in (14.38) has a constant graph at the level (14.39)
IKI (dB) = I-51 (dB) = 5 (dB) = 14 dB. The component plots for
each of these factors is shown in Fig. 14.7. Note that no corrections
are required for the sr factor or the constant factor component plots.
Next we add the uncorrected component plots graphically.
First we consider the slopes in each region. Recall that the slope Gain (dB)
of a sum of functions equals the sum of the slopes. Working left ----jl---.:..-i------T-----T-----"'i'---,-? UJ (log scale)
to right, the initial segments all have slopes of zero except the
lis factor, which has slope of -20 dB/decade. Then the slope
of the sum will be -20 dB/decade. This will change, that is, the
slope will break, at the leftmost (lowest) break frequency w = 2.
Adding slopes at w = 2+ results in 0 dB/decade, since the net slope
was -20 dB/decade at w = 2-, one slope increased (or "broke")
+20 dB/decade at 2, and the others did not change. There is a
downward break of +20 dB/decade at the second break frequency
w = 20 and no further breaks in slope. This yields the black curve
in Fig. 14.8.
We are ready to determine the decibel values. Since all slopes
are known, we need only know the decibel value at a single point
FIGURE 14.8 Bode gain plots (corrected and uncorrected) for Exam-
in order to determine all other decibel values. Examining the com-
ple 14.7.
2
D = 10glO 1= 0.3 decade
Having seen how the constant and real root terms are plotted, we tum last to the
complex root pair term (S2 + 2~ Wns + w~y. Taking the same path that we followed in
understanding the real root term, we replace s by j W and take magnitudes. For the case ga = 2r[wn (dB)]
that the term is in the numerator, the gain in decibels is = 40r log to wn
INi(jw)1 (dB) ~ {20rlOgIOW~' W Wn FIGURE 14.9 Bode gain plot for the complex root pair numerator factor
20r 10glO w 2 , W Wn (52 + 2~Wn5 + w~)'. .
Comparing these two, the straight-line segments comprising the uncorrected plot meet at
W = W n , so the break frequency for the complex root pair term is its natural frequency direction inside the knee of the uncorrected plot. Examining Fig. 14.9, the correction at
Wn. The slope before the break is 0, and beyond the break is 40r dB/decade. The graph W = Wn is very close to the desired maximum correction for both lightly and heavily
of the uncorrected Bode plot is shown in Fig. 14.9 in black. damped systems, those with damping factors ~ < 0.3 or ~ > 0.6. For moderately damped
While the uncorrected Bode plot for this term is independent of its damping factor systems, 0.3 :::s ~ :::s 0.6, the calculated correction, while accurate at W = W n , is not a
~, the correction to be applied depends strongly on its damping factor. The gain curves good measure of the maximum correction, which occurs far from W = W n Fortunately,
for various values of ~ ranging from its minimum (for complex roots) absolute value of 1 the maximum correction is not large for moderately damped systems.
upward is shown in Fig. 14.9. The maximum correction occurs near, but not precisely at,
the break frequency W n . To determine the correction at the break frequency W n , evaluate Determine the uncorrected and corrected Bode gain plots for
the exact gain expression at W = Wn: 1
H(s) = 3
INi(jwn)1 (dB) = 20rloglO U2~w~1 = 20rlogI021~lw~ (s2 + ~s + 900)
or (14.41) Checking the radical of the denominator quadratic, we find it is
negative, the roots are complex, and this is a complex root pair term.
Since the second term in (14.41) is the uncorrected value at W n , the correction must be
Its natural frequency and damping factor are, by (14.30), found from
the first term, or 20r 10glO 21~ I. The reference direction for this correction is upward,
or inside the knee of the uncorrected plot. If 1~ 1 < ~, this correction will be negative, W~ = 900
and the graph will be corrected downward. Noting that the correction 20r 10glO 21~ 1 is r
2~wn = ~
times the number 21~ 1measured in decibels, a general rule for correcting the complex pair
term that works for numerator and denominator terms alike is this. The correction to any or Wn = 30, ~ = 3/[(2)(2)(30)] = fa.
The break frequency is Wn =
complex root pair graph at W = Wn is r times the decibel equivalent of2H, with reference 30, and slope is -40r = -120 dB/decade after the break (this is a
572 Chapter 14 Frequency Response Section 14.3 Bode Gain (Ampl itude) Plots 573
Gain (dB) Gain (dB) Gain (dB) 40 dB/dec
{J)n = 30
- - 1 - - - - - - - - - - - . : : . . , . . - - - - - - " - 7 (J) (log scale) 200 (dB) =+ 46 dB
16 (dB) I - - - - - - . J
-99 dB =24dB
-[50(dB)l~
= -34 dB -[2500 ( d B ) l l - - - - - - - " " " ' \
-20 dB/dec = -68 dB
-20 dB/dec
FIGURE 14.10 Bode gain plots for Example 14.8. FIGURE 14.11 Component plots for Example 14.9.
denominator factor). The initial value is, by (14.40), -20r 10glO w~ values,
(or -2r times the decibel equivalent of wn ) with minus sign due to
the denominator location. This initial value is 46 + 24 - 34 - 68 = -32 dB
-20rloglOw~ = (2)(-60) log 10 30 = -177 dB Since there are 10glO 50/4 = 1.1 decades between w = 4 and w =
50, the uncorrected gain at w = 50 is -32 dB +(40 dB/decade)
This completes the uncorrected plot, shown in Fig. 14.10. The cor-
(1.1 decade) = +12 dB, and at w = 2500 it is similarly found to be
rection is r = 3 times the decibel equivalent of 21 ~ 1 = fo'
which is
+12 dB+ (20 dB/decade) (1.7 decade) = 46 dB. The correction due
3 [fo (dB)] = 3(-26 dB) = -78 dB. The reference direction for
to the complex root term at w = 4 is 21~ 1 (dB) = ~ (dB) = -6 dB
this correction is inside the knee (downward in this case), so since
or 6 dB outside the knee (downward). That at the other two break
the correction is negative, it goes outside the knee (upward). The
frequencies is 3 dB inside the knee (downward). The uncorrected
corrected plot is also shown in Fig. 14.10.
and corrected Bode gain plots are shown in Fig. 14.12.
Any number of real root and complex root pair terms may appear in the same
transfer function, as shown in the following example. Their component plots are simply Gain (dB)
added together, as are any other pair of terms in R(s). +46 dB
Let us find the Bode gain plots for a circuit with transfer function
200(S2 + 2s + 16)
+12dB
R(s) = (s + 50) (s + 2500)
I________-L--IIL::--_ _ _-::-:'::-::-_ _ _ _ _ _ _~> (J) (log scale)
In the numerator there is a constant factor and a complex root pair
factor with w~ = 16 and 2~wn = 2, or Wn = 4, ~ = i.
In the
denominator we see real root factors with break frequencies of 50 and
2500. The component plots are shown in Fig. 14.11. Starting from
the left, the uncorrected Bode plot will have a slope that is the sum -32 dB 1=======:::::--
of the slopes of the components; 0 + 0 + 0 + 0 = 0 dB/decade. The
plot breaks upward +40 dB/decade at W = 4, the break frequency -38dB
(natural frequency) of the complex root pair factor in the numerator,
down -20 dB/decade at W = 50, and again at W = 2500. The
dB value at w = 4 in the uncorrected plot is the sum of the initial FIGURE 14.12 Bode gain plots for Example 14.9.
574 Chapter 14 Frequency Response Section 14.3 Bode Gain (Amplitude) Plots 575
rr
i:
This completes our introduction to Bode's gain plotting method. There is a related Answer
method for plotting the phase shift by piecewise linear approximation, graphical addition, Gain (dB)
and correction, producing the Bode phase plot. There are several reasons why the Bode
phase plotting technique is not as widely used as the Bode gain plot described above. +26 dB
The Bode phase plot method is more cumbersome, since there are two, not one, break
frequencies for each real root term. The corrections spread over a much wider frequency
+20 dB
band than in the gain plot and are more difficult to estimate accurately by freehand
sketch. In addition, for many applications the variation of circuit gain with frequency
is more important to its intended function than are phase variations. For these reasons,
while it is not without its uses, we will not discuss the Bode phase plotting method f------1---H---L------~~~---~ (J) (log scale)
further.
In summary, while the process is not entirely without effort, the Bode gain plot
permits relatively quick and accurate freehand plots of gain versus frequency in a circuit, -11.1 dB F-------
and gain is often a circuit's most important single characteristic. We can get considerable
insight into what a circuit "does," the function it performs, by identifying the frequency
bands of high and low gain. If a computer running SPICE is handy, an ac analysis will
get us a more accurate gain plot with less work. But insight into how the Bode procedure EXERCISE 14.3.2
works informs us of how the circuit gain curve will change if we add or delete poles
or zeros or if we shift their locations. When confronted with a design problem, this 14.3.3 Sketch the Bode gain plots for H(s) = (s2 +s + 100)/(s2 +20s +
100).
understanding can be used to guide our selection of a transfer function and ultimately
Answer
a circuit design. Neither SPICE, nor any other computer-based simulation tool, can
substitute for a circuit designer's own understanding of how the frequency response of Gain (dB)
circuits can be manipulated. This is the real payoff for studying the Bode gain plot odB --l~~===="""";;;;:;::__--...:lrO---::::::::======~ (J) (log scale)
method.
EXERCISES
-26 dB
14.3.1 For each component factor in H(s) = 4000(s2_1)/[(s +20)2(s2+
3s + 36)], specify the initial slope, final slope, initial value, and location
of break frequency. Also specify Wn and ~ for complex pole pair fac- EXERCISE 14.3.3
tors.
Answer
4000: 0 dB/decade, 0 dB/decade, +72 dB, no break frequency
(s + 1): 0 dB/decade, +20 dB/decade, 0 dB, 1 rad/s The frequency response function H(jw) reveals the frequency-dependent behavior of a
circuit, and Bode gain plots offer a convenient means to access this information. We
l/(s + 20)2: 0 dB/decade, -40 dB/decade, -52 dB, 20 rad/s
next examine specific circuits to see what sort of frequency-dependent behavior we may
1/(s2 + 3s + 36): 0 dB/decade, -40 dB/decade, -31.1 dB, 6 rad/s, have and why we might want to have it.
Of considerable interest are circuits that treat a narrow range of frequencies very
Wn = 6, ~ = 1/4 differently than all other frequencies. These are referred to as resonant circuits. The
14.3.2 Sketch the uncorrected and corrected Bode gain plots for the H(s) gain of a highly resonant circuit attains a sharp maximum or minimum at its resonant
of Exercise 14.3.1, labeling all slopes and both coordinates at each break- frequency, while its variations outside a small band of frequencies around the resonant
point. frequency are much lower.
*
IY(jw)1 (dB)
for w, since IY(jw)1 is the gain and its maximum value is 1jR. Leaving the algebra to Circuits with high quality factors are very frequency selective, and as we see from (14.52),
the exercises at the end of this section, after subtraction of the two solutions to (14.48), this implies low values of resistance R. Since practical inductors include significant series
the bandwidth for the series RLC is simply resistance included in their own device models, it is difficult and expensive to design
high-Q resonant circuits passively, that is, solely with RLC elements.
FIGURE 14.14 Two gain curves with same bandwidth B but different For high-Q circuits, we will find it more convenient to use active circuits, those
quality factor Q. containing op amps. Their principles will be explored later in this chapter. Thus far we
- - + - - - - - - - ' - - - - - - - - 7 w (log scale) Consider the resonant circuit shown in Fig. 14.16. It is neither series
Wn=~ \
nor parallel RLC, so the formulas in Table 14.1 do not apply. With
(a) (b) the specified input and output, we have H(s) = Z(s) and
FiGURE 14.15 (a) Parallel RLC circuit; (b) Bode gain plot.
Z(s) =
(1O/s)(10s + 1) = 10 (s+ to)
-....:.......,..---=..::!~
(lOis) + lOs + 1 s2 + toS + 1
have considered only series RLC resonant circuits. Many other circuit configurations can
be used to produce resonant behavior. In the case of the parallel RLC circuit (Fig. 14.15), There is a constant factor 10, a real zero at s = - to,
and a complex
taking i(t) as the input and v(t) as the output, the transfer function characterizing the
gain is
root pair factor with Wn = 1, 2r;wn = to;
that is, r; = The cor- fa.
rected and uncorrected Bode gain plots are shown in Fig. 14.16(b).
1 The corrected plot shows that the circuit is resonant near Wn =
H(s) - Z(s) - --,:----:-~-:--_:__----,- (14.53)
- - sC+(1IR) + (1/sL) 1 rad/s. This is not exactly the location Wr of maximum gain, but
We could study this resonant circuit by repeating the steps we took in the series RLC case, it is close. Another estimate of the resonant frequency is that fre-
but there is an easier way. Comparing (14.53) with (l4.42), the transfer functions are quency at which the phase of Z (j w) is zero. In fact, some authors
identical after substitution of C for L, L for C, and llR for R. Since all our results are define this as the resonant frequency for general resonant circuits.
computed from the transfer function, we may simply repeat the results noted for the series In the present case, this results in the equation
RLC case while making these substitutions. The results are summarized in Table 14.1. W
A final observation concerning series and parallel RLC circuits is that at resonance tan- 1 lOw = tan- 1 2)
10(1 - W
the phase shift of their frequency response functions H(jw)lw=wr is zero. In both cases,
IZ(jwr)1 and IY(jwr)1 at resonance are each purely real, and voltage and current are in
phase. The imaginary parts of the impedances and the admittances cancel completely at IZ(jw)1 (dB)
1 1
Wr (resonant frequency)
-lIT -lIT odB 1==::::::=-.../
R 1 -+-----'----....l...----7-W (log scale)
B (bandwidth) I wn = I
L RC
10
*~ R/f
Q (quality factor) (a) (b)
FiGURE 14.16 (a) Circuit for Example 14.11; (b) Bode gain plot.
W Ao(dB)
lOw = -:-:-______:-
10(1 - ( 2 )
and solving, W = (-J99)/1O. These two approximations to Wr are
very close. Neither is the exact frequency Wr at maximum gain, -20 dB/dec
which may be found (by the ambitious) as the exact solution of the
equation
d
dw IZ(jw)1 = 0 (14.54)
584 Chapter 14 Frequency Response Section 14.5 Frequency Response of Op Amps 585
Ao = 105 and Wo = 10 radls used in an inverting amplifier with gain RF IRA = 4. We
sustain a gain reduction from 105 to 4, or a factor of 25,000, in passing from open-loop
op amp gain to circuit gain. The bandwidth, however, has increased from 10 radls to
We = (10 5 )(10)1(1 + 4) = 200,000, a factor of 20,000. Thus we have gotten back in
bandwidth almost all we gave up in gain. This trade-off greatly favors our purpose, since
we need a wider bandwidth and only want the indicated gain of RFIRA = 4. The new
bandwidth, 200,000 rad/s, is almost 40 kHz, large enough to accommodate most audio
circuits tasks.
Finally, note that the gain-bandwidth product of the inverting amp is
FIGURE 14.19 Inverting amplifier.
RF RFIRA
GBW = -We = Aowo-----
RA 1 + (RFIR A)
Factoring out -RFIRA and substituting A(jw), we obtain \ As the gain RF IRA ranges over moderate values, say 1 to 10, the gain-bandwidth product
-RF 1 does not stray far from the open-loop gain-bandwidth product GBW = Aowo. This
H(jw) = RA 1 + [(jw + wo)IAowo] (1 + RFIR A) suggests that gain changes are compensated roughly equally by bandwidth changes in
a manner limited only by the GBW of the op amp itself. If a larger combination of
Separating the real and imaginary parts in the denominator yields gain and bandwidth is needed, the gain must be split between two stages or a different,
probably more expensive, op amp with higher gain-bandwidth product is used.
-RF 1
In this section we have seen that the op amp is highly frequency dependent, with
= ~ jw {[I + (RFIR A)] IAowo} + {1 + [1 + (RFIR A)] lAo} too little bandwidth to be of much practical use open loop. When configured in a
If we agree, as before, to limit our desired circuit gains RF I R~ Ao, no~ aver: negative-feedback circuit, however, such as the building-block circuits introduced ear-
restrictive assumption since Ao is very large, then the real term III the denormnator IS lier (single- and multi-input inverting and noninverting amplifiers and voltage follow-
well approximated by 1, and we have ers), the bandwidth will increase roughly in proportion to the decrease in gain. Thus
if we restrict ourselves to modest circuit gains for single op amp stages, we can se-
. -RF 1 cure the bandwidth we need to build practical circuits despite the very small band-
H(Jw) = ~ jw{[1 + (RFIRA)]IAowo} + 1 width the op amp itself presents us with open loop. Within their bandwidth, the op
amps once again appear to be frequency independent, just as we have been assuming
The Bode plot for this single-pole gain function is shown in Fig. 14.20. It shows thus far.
us the relationship between the gain I-RFIRAI = RFIRA and the bandwidth We for the
inverting amplifier. For any desired gain RF I R A, the circuit will have bandwidth We m~ch
larger than the open-loop bandwidth woo Indeed, the reduction in gain due to negative
feedback, of ratio Ao to R FIR A, has been returned almost completely in expansion of the EXERCISES
bandwidth, ratio wclwo = Aowo/[1 + (RFIR A)]. Consider, for instance, an op amp with
14.5.1 An op amp with I-MHz gain-bandwidth product is used as an
inverting amp. Find the bandwidth of this circuit in hertz if the gain RF / RA
Gain (dB)
is set to 1, 10, 100, 1000, and 104 .
Ao(dB) Answer 500 kHz; 90.9 kHz; 9.9 kHz; 999 Hz; 100 Hz
-20 dB/dec 14.5.2 Suppose that two copies of the circuit of Exercise 14.5.1 are put
Open-loop op amp in cascade, creating a circuit with gain (RF/RA)2. What is the bandwidth
of this new circuit for gains (RF/RA)2 = 100, 104 ? Note from Fig. 14.5
that a corrected gain of +1.5 dB occurs at W = 0.642we .
Answer 58.4 kHz; 6.36 kHz
14.5.3 Starting from (3.19b) and using the approximation Ao [1 +
(R F / R A)], show that the bandwidth We of the noninverting amp is the
-1---L-------L.---~~-"----7 w (log scale) same as that for the inverting amp shown in Fig. 14.20. What is its gain-
bandwidth product?
Answer Aowo
FIGURE 14.20 Bode plot for an inverting amplifier.
In ordinary usage, a filter is a device designed to block some things while letting others
pass through. A car's oil filter stops abrasive metal particles and other contaminants from 1 1
circulating through the engine, while permitting the oil to pass unimpeded. In circuit 12 12
parlance, a filter is a device that impedes the passage of signals whose frequencies fall
within a band called the stopband, while permitting those in another band, the passband,
to pass from input to output relatively unchanged.
Filters are among the most common modules found in general circuit designs. Al-
o +------L..-___....:.:::o,.. w O~----------L---~
most every practical electronic circuit of any complexity contains one or more filters. we
This includes radio, telephone, television and other receivers and transmitters, control (a) (b)
systems for industrial machinery, computer circuits, power supplies, and so on. A com-
mon use of filters is to suppress noise. If we know that the signals in our circuit are FIGURE 14.22 Low- and high-pass filter gain curves.
limited to specific bands, we can put the other bands in the stopband of a filter and block
noise from degrading the performance of our circuit. Another common use is to enhance by (14.58), this occurs at
an important range of frequencies. If we are tuned to a certain radio station, the band 1
of frequencies containing that station's signal can be made the passband of a high-gain We =-
RC
filter so that the desired signal becomes very prominent in the filter output.
The degree to which a signal of a particular frequency W passes from input to output If, instead, we take V R as the output,
is measured by the frequency response function H(jw). An ideal filter is one for which . VR
H(jw) = 0 in the stopband and lLQ in the passband. Idealfilters block stopband signals HHP(jw) = -
Yin
completely while passing signals in the passband without any change. H(jw) = lLQ in
R jwRC
the passband means that the output is the same amplitude as the input and there is no (14.59)
phase shift between them. R + [1/(jwC)] jwRC + 1
Filters are named by the location of their passband. If it consists of all W :s We for RCw
some We we have a low-pass filter with cutoff frequency We' If, instead, the passband is
IHHP(j W) I = -;:::===
JR2C 2w + 1 2
the set of frequencies above a given cutoff frequency We, W 2: We, we have a high-pass
This is a high-pass filter with maximum gain as w --+ 00 equal to
filter. In the case of an ideal filter, the value of the cutoff frequency We is obvious; it is
unity. The -3-dB point is once again We = l/RC. To illustrate
the frequency at which the gain jumps between 0 and 1. For nonideal filters we need to
the degree of nonideality in these RC filters, suppose that we move
define the cutoff frequency We in some reasonable way. We will usually define it as the
1 decade into the passband; for instance, in the low-pass case take
half-power frequency; in other words, the cutoff frequency We is the frequency at which
w = wc/l0 = 1/(lORC). Situated comfortably in the passband, the
the gain is 3 dB below its maximum value. frequency response HLP(jwc/lO) is
The simple RC voltage-divider circuit of Fig. 14.21 can be configured
as either a low- or a high-pass filter. Defining the phasor Ve as HLP lO~C = 0.995/ -5.r
output, we obtain The ideal filter would have gain of unity and phase shift of 0 since
. Ve 1/(jwC) 1
(14.57)
!
we are within the passband. We are within % of unity gain and
HLP(jw) = Yin = R + [1/(jwC)] jwRC +1 within 6 of zero phase shift. For many, but not all, purposes this
performance is sufficient. '
The gain of this circuit is
1
IHLp(jw)I = ,JR2C2w 2 + 1 (14.58) There are other filters designed to pass not all frequencies lower or greater than a
limit, but to pass some intermediate band. A bandpass filter has passband WI :s w :s wu ,
and is plotted in Fig. 14.22(a). Since the high-gain region consists where Wu and WI are the upper and lower half-power frequencies. A bandreject or
of all frequencies below a certain value, this is a low-pass filter. By bandstop filter has a split passband, all w :s WI and all w 2: wu , as shown in Fig. 14.23.
FIGURE 14.21 RC low-pass or high-pass filter (14.58), the maximum gain occurs at dc and equals unity. Three A bandpass filter gain characteristic can be produced by mUltiplying low- and high-
circuit. decibels below that is -3 dB, so the gain at We must be I/J"2. Also pass filter gain curves whose passbands have been set to overlap. Suppose that we have
1
(14.63)
-fi
The condition that the output voltage V2 (s) change significantly when the circuits are
connected is called loading and was first raised in Chapter 3. We see from (14.63) that
-4~~--~L---------~----~--~w in the absence of loading the overall transfer function of a cascade interconnection is the
product of individual transfer functions. If loading does in fact occur, the modified V 2(s)
(a) (b)
will not equal its unloaded value H1(s)V1(s), (14.61) will not be true, and the product
FIGURE 14.23 Bandpass and bandstop filter gain curves.
rule for cascaded circuits (14.63) no longer follows.
Replacing all to the left of the switch by its Thevenin equivalent and all to the right
likewise, assuming no internal sources in H 2 (s), we have Fig. 14.25. The impedance
a circuit for which Zl (s), the Thevenin impedance of the left circuit looking back into its output terminals,
(14.60) is called its output impedance, and Z2(S) is the input impedance of the right circuit. With
the switch open, V2(s) = VT (s). With it closed,
with the cutoff frequency of the high-pass filter IiJcHP set lower than that of the low pass, s - Z2(S) V s
/.,
cvcHP < cvcLp
/., If these two filters are ideal, we will have the ideal bandpass frequency
V2 ( ) - Zl (s)+ Z2(S) T() (14.64)
response, H(j 1iJ) = 1 for all IiJ between IiJI = IiJcHP and liJu = IiJcLP, ~d H(] 1iJ) =.0
elsewhere. For nonideallow- and high-pass filters, the bandpass filter gam (14.60) WI!! V 2 (s) will not change significantly when the circuits are connected together if the
be close to unity within its passband IiJI :s IiJ :s liJ u and have gain lower than 1/-J2 impedance ratio in (14.64) is close to unity. Loading will not occur when two cir-
cuits are cascaded if the input impedance of the load circuit is much greater than the
outside the passband. . . .
How do we produce a circuit satisfying (14.60)? Suppose that two CIrcUIts WIth output impedance of the source circuit. In this case we can use the product rule (14.63);
otherwise, we cannot.
voltage transfer functions HI (s) = V2 (s)fVI (s) and H2(S) = V4(s)fV3~S) are connect~d
in cascade, the output of one to the input of the other, through a sWItch as show~ ill
Fig. 14.24. The transfer function HI (s) is defined for 12(s) = 0; so as long as the sWItch
is open, V 2(s) = H2(S)V 1(s). . . .
Assume for a moment that V2 (s) is exactly the same whether the sWItch m FIg. 14.24
is open or closed. In that case, with the switch closed,
(14.61)
and since V 3 (s) is the input to a circuit with transfer function H2 (s),
(14.62) FIGURE 14.25 Thevenin equivalents of cascaded circuits.
l
s-domain mesh equations:
10
1 + -.;-
-:0 ] [II(S)]= [V1(S)]
-10 11 12(s) 0
FIGURE 14.24 Circuits in cascade. 1+-
s s
- ,,, -,,
,
, Voltage follower '
Low pass --.L = we = 10 High pass RiC = we = I
RC
FIGURE 14.27 Inserting a voltage follower to eliminate loading.
FIGURE 14.26 Cascade of passive low- and high-pass filters.
Thus, if we take care to prevent loading, perhaps by separating stages with a
buffer amplifier with very high input impedance such as the voltage follower, we can
get cascaded circuits whose transfer functions are the products of the individual transfer
By Cramer's rule, functions in cascade. The circuit of Fig. 14.27 has the desired bandpass gain characteristic
shown in Fig. 14.23.
Finally, a bandstop filter characteristic can be produced by subtracting a bandpass
12(s) = 1 + (21/s) + (1O/s2) frequency response from unity:
lOsVI(s) HBS(jw) = 1 - HBP(jw)
s2 + 21s + 10 So wherever the gain of the bandpass is near 1, that of the bandstop will be near 0, and
vice versa, as shown in Fig. 14.23(b).
Since 12 (s) = V4(s), the true overall transfer function is With what physical elements are filters designed? A filter built entirely from pas-
sive (RLC) elements is a passive filter. If op amps or other circuits with electronic
lOs devices requiring power supplies are used, we have an active filter. While having ob-
(14.65)
H(s) = S2 + 21s + 10 vious disadvantages of component count and power consumption, active filters are very
often the choice for practical filter design applications. There are serious limitations to
Assuming no loading, the overall transfer function would be the the use of passive filters, as we now describe.
product of the low- and high-pass transfer functions: Passive filters corresponding to transfer functions containing complex root pair fac-
tors require both inductors and capacitors, and inductors tend to be undesirable elements
HLP (s)HHP (s) = 0. lsl+ 1 C~I) in modem miniaturized circuit design. They are bulky and expensive and have sufficient
series resistance to make them generally useful only in relatively low Q applications. In
(14.66)
lOs addition, inductors cannot be included in thin-film integrated circuits effectively using
current technology. Passive filters are dissipative; in other words, they cannot have gains
S2 + lIs + 10 greater than unity at any frequency. Moreover, as we saw in Example 14.13, cascading
passive filters leads to loading, which makes the design process more difficult, since over-
Clearly, (14.65) and (14.66) are quite different. For instance, wh~n
all transfer functions cannot be predicted accurately from the product of the individual
s = jw = j, the actual gain from (14.65) is IH(j)1 = 0.438, whIle
filter transfer functions. Active filters do not require inductors, may have any desired
the product rule in (14.66) predicts a gain of IHLP(j)IIHHP(j)1 =
gain, and can be easily designed to be highly resistant to loading, as we have seen.
0.704. The effect of loading is to reduce the overall response.
If we insert a voltage follower, introduced in Chapter 3, into
Design a third-order low-pass filter with transfer function
this circuit as a buffer amplifier between a and b to eliminate load-
ing, then in this case we will have the input impedance of the circuit 1
to the right of a essentially infinite (the virtual open principle for op H(s) = (s + 1)(s + 5)(s + 20)
amp analysis), so loading can no longer occur. The overall t:ansfer We can cascade together three low-pass filters with transfer functions
function for Fig. 14.27 will be the product of transfer functlons of 1/(s + 1), l/(s + 5), and 1/(s + 20) and get the overall transfer
the cascaded circuits as given by (14.66). function above, taking care to use a voltage follower to buffer the
C
Rs = aR, Ls = aL, Cs = - (14.69)
EXERCISES a
.-- ... ----------,
, 14.6.1. What kind of filter is this passive RL circuit with H(s) = V L(S)/
Vin(S)? What is its half-power frequency we? Repeat for H(s) = VR(S)/ where the subscript s indicates the values after scaling. To impedance scale by a, multiply
Vin(S). each Rand L value by a, and divide each C value bya. The difference arises because
Answer High pass, R/L; low pass, R/L the impedance of resistors and inductors is proportional to their element parameter R
,, 14.6.2. What kind of filter is H(s) = s2/(s2 + 2"fiwns + w~)? What is or L, while that of a capacitor is inversely proportional to C. To keep Vs the same,
L
,,: VL its half-power frequency? as required by (14.68), independent voltage sources are unchanged, while independent
Answer High-pass; Wn current sources must be scaled by l/a.
,,, - 14.6.3. Loading is frequency dependent. If two identical RC low-pass
,
L _ _ _ _ '_ _ _ _ _ _ _ _ J filters (see Fig. 14.21) with R = C = 1 are cascaded, at what frequency
To impedance scale the circuit of Fig. 14.29(a) by 1000, R and L
will there be no loading at all? What is the ratio of the actual gain at W =
EXERCI SE 14.6.1 1 radls to the product of the two individual filter gains? At W = 100 radls?
are scaled up by 1000 and C down by 1000. The voltage source
Answer dc; ~; 0.9998 is unchanged, and the current source is scaled down by 1000. All
voltages in the new circuit are unchanged from their values in the
old circuit; all currents are reduced by a factor of 1000.
(a) (b) o
FIGURE 14.29 (a) Original circuit; (b) impedance scaled by 1000. FIGURE 14.30 Circuit of Fig. 14.28 impedance scaled to practical
current levels for low-power op amps.
The gains of current-controlled current sources and voltage-controlled voltage \ frequency scale by fJ, divide each Land C by fJ, leaving all other elements unchanged.
sources in the circuit will not change when impedance scaling, since the ratio of controlled
to controlling variable is the same after scaling as before. But voltage-controlled current
sources will have their gain (transresistance r) scaled by a, just as the other elements L C
whose units are ohms, and current-controlled voltage sources, with gain (transconduc- Rs = R, Ls = {3' Cs = - (14.70)
fJ
tance g) measured in siemens, must have these gains scaled by l/a. For instance, if
a circuit with a current-controlled voltage source with source function v(t) = 4i e (t)
is to be impedance scaled by 20, the new controlled source will have source function We earlier impedance scaled the resonant circuit of Fig. 14.29(a),
v(t) = 80ie(t). Then, with the currents in the new circuit reduced by a factor of 20, the
with result shown in Fig. 14.29(b). Let us scale the resonant fre-
increase in the transresistance of the controlled source by a factor of 20 will compensate quency of this circuit upward by a factor of 104 . This is easily
and its voltage will remain unchanged as required. accomplished by replacing the 2-H inductor by a 0.2-mH inductor
A common use of impedance scaling arises in the design of op-amp circuits. Al- and the I-F capacitor by a 100-fLF capacitor. Note that the resonant
though it may be convenient to use values close to unity for Rand C in the design frequency of the circuits of Fig. 14.29(a) and (b) is the same, since
process, this will inevitably result in a circuit whose impedance levels are too low for the
Wr = -Jl/LC and the product LC is the same in these circuits.
use of popular low-power op amps, such as the fLA741. These devices operate best at mil- Impedance scaling does not result in frequency scale changes.
liampere current levels when their voltage levels are an appreciable fraction of the power
supply voltage, typically 15 V. We may use convenient values like R = 1 Q during the
Frequency scaling is often used in the design of filters. Handbooks specifying low-
design process, but if we are not to "fry" the op amp by drawing excessive current, we
pass filter circuits, for instance, will need to list only the values for the low-pass whose
had best impedance scale such values to the order of kilohms before flipping the switch.
cutoff frequency is set to We = 1 rad/s. The user can then frequency scale the circuit to
This scaling also reduces capacitors to more practical values than 1 F or so. Note that
any desired cutoff frequency.
since the scaled circuits have the same voltages as prescaled circuits, there is no change
to any voltage transfer function H(s) = Vout(S)/Vin(S) when a circuit is impedance scaled.
EXERCISES
Consider the circuit of Fig. 14.28. To prevent excess power dissi-
pation from the op amps resulting in circuit failure due to op amp
burnout, we impedance scale by 2000. The resulting circuit, with 14.7.1. Suppose that a circuit is impedance scaled by 100. Consider four
practical values, is shown in Fig. 14.30. transfer functions: HI(s) = V2(S)!VI(S), H2(S) = V2(S)!lI(S), H3(S) =
12 (S)!l1 (s), and H4(S) = 12(S)!VI (s). Which will change? By what factor?
Answer H2(S) will be multiplied by 100, H4(S) by 1/100; others
Given a circuit, frequency scaling by fJ is the specification of a new set of circuit unchanged
parameters such that the scaled circuit responds at frequency fJw exactly as the original 14.7.2. If a parallel RLC passive circuit is frequency scaled by 50 and
one does at w. There is only one way that w enters into the calculation of any response, then impedance scaled by 50, what are new RLC values?
through the impedances jwL and 1/ jwC. If we arrange that the new inductive and Answer 50R, L, C /2500
capacitive impedances at fJw are the same as the old ones at w, all responses must be the 14.7.3. A handbook lists a fourth-order Butterworth filter with We = 1 as
same, as required. Now the product jwL remains unchanged after w is replaced by fJw
H(s) = V2(S) = 1
if we also replace L by L / fJ. The same is true for 1/ j wC if C is replaced by C / fJ. To V I (s) s4 + 2.61s 3 + 3.41s 2 + 2.61s + 1
0.0625
S4 + 1.31s + 0.853s 2 + 0.326s + 0.0625
3 *** ** * * ** *** ** *** ** * ** **** ** ** * ** *** ** **** *** ** * * * ** *** *** ** **** * * *** *** *** ***
FREQ VDB(l)
SPICE may be used to determine frequency response, both gain and phase shift as func- 1. OOOE-02 1. 479E+OO *
1.096E-02 1.728E+DO *
tions of frequency, using the . AC control statement. The gain IH(jaI is specified by 1.202E-02 2. QIOE+OO *
1.318E-02 2. 328E+OO
adding the suffix M(magnitude) to the name of a current or voltage to be output, and the 1.445E-02 2. 684E+OO
1.585E-02 3. 078E+OO
phase shift jH(ja by the suffix P. For instance, the output control statement 1. 738E-02 3. 513E+OO
1.905E-02 3. 987E+OO
2.089E-02 4. 501E+OO
.PLOT AC VM(4) VP(4) 2.291E-02 5. 055E+OO
2.512E-02 5. 647E+OO
2.754E-02 6. 278E+OO
will result in the plotting of both the magnitude and phase of the voltage at node 4 in 3.020E-02 6. 945E+OO
3.311E-02 7. 649E+OO
the frequency range specified on the . AC control statement. If the amplitude of the ac 3.631E-028.389E+OO
3.981E-02 9.166E+OO
source being swept is 1, the magnitude of the output voltage VM (4) is directly equal to 4.365E-02 9. 981E+OO
the gain. If the Bode coordinates, gain in decibels versus log frequency, are desired for 4.786E-02 1.084E+Ol
5.248E-02 1.174E+Ol
the gain curve, the suffix DB is used instead of M. The keyword DEC (decade) is then 5.754E-02 1. 269E+01
6.310E-02 1. 371E+01
used in the AC control statement as shown in Example 14.18. 6.918E-02 1.479E+01
7.586E-02 1.597E+01
8.318E-02 1.727E+01
We will check our earlier calculation of the the Bode gain plot, 9.120E-02 1.872E+01
14.18 Fig. 14.16(b), for the resonant circuit of Fig. 14.16(a). The SPICE
1.000E-01 2. 039E+01
1. 096E-01 2.237E+01
1. 202E-01 2. 485E+01
input file is 1. 318E-01 2. 820E+01
1.445E-01 3.331E+01
1.585E-01 4. 001E+01
EXAMPLE 14.18 1.738E-01 3. 391E+01
1.905E-01 2. 854E+01
* 2.089E-01 2. 506E+01
I 0 1 AC 1 0 2.291E-01 2.250E+01
2.512E-01 2. 046E+01
L 1 2 10 2.754E-01 1. 875E+01
3.020E-01 1.725E+01
R 2 0 1 3.311E-01 1.591E+01
40.01 dB, comparing favorably with our calculated 40 dB. The cor-
FIGURE 14.313 SPICE outputs for Example 14.18.
ner frequency at a> = 0.1 rad/s (f = 0.0159 Hz) has a gain of
+3 dB in Fig. 14. 16(b); this checks well with the gain of 3.08 dB at
598 Chapter 14 Frequency Response Section 14.8 SPICE and Frequency Response 599
**** 04/29/94 11:30:15 ********* Evaluation PSpice (January 1991) ************ f = 0.1585 in the SPICE output. In the example it was mentioned
EXAMPLE 14.18 that the location of the resonant peak could only be approximated
using our Bode plotting method. We may use SPICE to examine
AC ANALYSIS TEMPERATURE = 27.000 DEG
the details of a gain curve in any region by changing the sampling
parameters on the AC control statement. Examining the gain near
******************************************************************************
the peak by using the control statement
FREQ VDB(l)
.AC LIN 25 157 .163
(' )---------- 3.9000E+01 3. 9500E+01 4. OOOOE+01 4. 0500E+01 4.1000E+01
1.570E-01 3.973E+01
we get the output shown in Fig. 14.31(b). To four significant digits,
1. 571E-01 3.977E+01 the peak gain is 40.04 dB, and the peak occurs somewhere between
1.572E-01 3.980E+01
1. 574E-01 3.983E+01 0.1588 and 0.1594 Hz, or 0.998 and 1.002 rad/s.
1.575E-01 3.986E+01
1.576E-01 3.988E+01
1.577E-01 3.991E+01
1.579E-01 3.993E+01
1. 580E-01
1.581E-01
3.995E+Ol
3.997E+01
EXERCISES
1.582E-01 3.998E+01
1.583E-01 4.000E+01
1. 585E-01 4.001E+01 O.I/-iF O.I/-iF 14.8.1. Use SPICE to create plots of the gain and phase-shift curves for
1.586E-01 4.002E+01
1.587E-01 4.003E+01 the circuit of Fig. 14.26 for 0 < w < 10, using linear frequency and output
1.588E-01 4.004E+01 scales with a sample every 0.1 rad/s.
1.590E-01 4.004E+01
1.591E-01 4.004E+01 14.8.2. The bandstop filter shown is known as a twin-T filter. Determine
1.592E-01 4.004E+01
1.593E-01 4.004E+01 the frequency in hertz at the bottom of the notch to four significant digits.
1.594E-01 4.004E+01 Answer 15.92 Hz
1.596E-01 4.003E+01
1.597E-01 4.002E+01
1.598E-01 4.001E+01 + +
1. 599E-01 4.000E+01
1. 601E-01 3.999E+01
1.602E-01 3.997E+01
1.603E-01 3.995E+01
1.604E-01 3.993E+01
1.606E-01 3.991E+01
1.607E-01 3.989E+01
1.608E-01 3.986E+01
1.609E-01 3.984E+01
1.610E-01
1.612E-01
3.981E+01
3.978E+01
EXERCISE 14.8.2
1.613E-01 3.975E+01
1. 614E-01 3.971E+01
1.615E-01 3.968E+01
1. 617E~Ol 3.964E+01
1.618E-01 3.960E+01
1.619E-01 3.956E+01
1.620E-01 3.952E+01
1. 621E-01 3.948E+01
1.623E-01 3.944E+01
1.624E-01 3.939E+01
1.625E-01 3.934E+01
1.626E-01 3.930E+01 The frequency response of a circuit is the combination of gain and phase shift experienced
1. 628E-01
1. 629E-01
3.925E+01
3.920E+01
by an input sinusoid as a function of its frequency. Design criteria for practical circuits
1.630E-01 3.915E+01 such as filters and amplifiers are often expressed in terms of a desired frequency response,
that is, gain and phase shift versus frequency.
JOB CONCLUDED II The frequency response function H(jw) is found from the transfer function H(s) by
TOTAL JOB TIME . 23 replacing s by jw. It exists for stable circuits only .
FIGURE 14.31 b SPICE outputs for Example 14.18. II IH(jw)1 is the gain at frequency w, ;H(jw) is the phase shift.
II The decibel equivalent of a positive real number is 20 times the log base 10 of the
number. Adding 20 dB scales a number by 10, subtracting 20 dB scales by 0.1. The
fiGURE P14.31
!ill!] 103 rad/s and Q = 40. Repeat for parallel RLC.
is 14.40. An op amp with gain-band width product GBW =
fiGURE P14.27
14.32. Repeat Problem 14.28 for this circuit. 106 rad/s is used as an inverting amp with RF / RA = -1.
At what radian frequency W is the actual gain -3 dB?
2Q
14.28. Sketch the Bode gain plots, corrected and uncor- -20 dB? -120 dB?
FIGURE P14.22 rected. Label each slope and label all breakpoints by their I Q I Q 14.41. Starting from (3.19b), derive the single-pole fre-
wand decibel values. The output variable is indicated. quency response of a noninverting amp with A(jw) =
Albert Einstein
609
Chapter Contents In a linear inductor (we shall continue to assume linearity of all inductors and magnetic
media throughout), the flux produced is linear in i, and thus so is A, which may therefore
III 15.1 Mutual Inductance III 15.5 SPICE and Coupled Coils be written
III 15.2 Circuits with Mutual Inductance III Summary
A = Li (1S.2)
III 15.3 Mutual Inductance and III Problems
Transformers The constant of proportionality L is the inductance of the inductor. Standard SI units of
flux <P and flux linkage A are webers (Wb), and inductance L is in henries (H). Faraday's
III 15.4 Ideal Transformers
law asserts that a voltage is induced across the terminals of the coil equal to the time
rate of change of its flux linkage A, leading to the familiar element law for an inductor,
dA di
v = - = L- (1S.3)
dt dt
This previous analysis assumed that the only flux linking the coil was that due to
its own current; that is, the coil is a simple or uncoupled coil. Sometimes the turns of
a coil are threaded by the flux produced by the currents of one or more other coils, in
which case we have coupled coils. Consider Fig. IS.1, in which two coils have fixed
relative locations and orientations. The current il in coil I produces flux <PII, part of
which threads coil 2 and part of which does not:
One way in which elements can be made to interact, to influence one another's current
(1S.4)
and voltage behavior, is by using conducting wire to join their terminals. This is the
I
I only kind of interaction between circuit elements we have considered so far. Indeed, Here <P21 is the component of <PII that threads coH 2, and the remainder, <PLI, is called
we defined an electric circuit as a set of elements and the conducting wire that connects coil I' s leakage flux. Coil 2 also has a current, i2, producing a flux <P22, where
I~ them.
There is another way in which elements may be made to interact, one that does
not require material contact. We saw in Chapter S that a voltage is induced across the
(15.5)
<P12 is the component of <P22 threading the turns of coil 1, and <PL2 is its leakage flux.
r"~, terminals of an inductor in the presence of a time-varying magnetic field. This field
may be produced by the inductor itself, which we called self-inductance, or it may be
The total flux <PI threading coil 1 is the sum of components due to i l and i2,
~,
produced by a second inductor. This is called mutual inductance, which we explore in <PI = <PI I + <P12 (15.6)
this chapter. and the net flux linkage for coil I is, by (1S.1) and the above,
r We begin with a basic physical description of mutual inductance, deriving the
I
[ i-v laws and power exchanges for sets of interacting inductors, or coupled coils. In AI = NI<PI = NI<PII + NI<P12 (15.7)
Section IS.2 the analysis of circuits containing coupled coils is considered in detail. Each term on the right is a flux linkage and, by linearity, is proportional to the current
Useful circuits that do not themselves contain coupled coils, but are equivalent to, and producing it. The first term is due to coil l' s own current,
may be substituted for, those which do are developed in Section IS.3. Next we tum to
the special but practically important case of perfect coupling, the ideal transformer. The (15.8)
use of these devices to transform the levels of currents, voltages, and impedances as well
i as to isolate subcircuits from one another is of great practical importance and is described
next. The chapter closes with a note on the use of SPICE with circuits containing mutual
I inductance and transformers.
+ +
i
! V2
610 Chapter 15 Mutual Inductance and Transformers Section 15.1 Mutual Inductance 611
where the constant of proportionality L[ is called the self-inductance of coil 1, and, as If we hold i2 at its initial value of zero during the entire time interval [0, ttl, the second
in (1S.2), has units of henries. The second term is produced by and is proportional to term in (IS.12a) is zero, as are both terms in (1S.12b), and
the current in the other coil:
(15.9)
(15.14)
The positive constant M12 is called the mutual inductance. This is also measured in
henries, as the dimensional equivalence of (15.8) and (15.9) requires.
The reason for the sign in (15.9) is that the flux component <P12 produced by
the current in coil 2 may be in the same or opposite direction as that produced by coil 1 This is the form derived in Chapter S for energy stored in a uncoupled inductor. For
itself. If the two coils are oriented such that positive i2 produces flux threading coil 1 in a subsequent time interval [t 1, t2], suppose we hold i 1 constant at its value at t1, i 1(td.
the same direction as the flux threading coil 1 due to its own positive iI, then the plus Then W(t2), the energy stored at t2, is given by
sign must be used; otherwise, we use the minus sign.
For a given pair of coupled coils, the correct sign can always be determined from a \
detailed physical description of the coils, how they are wound, and their relative locations
and orientations in space. To avoid the need for specifying the required physical detail,
however, we shall soon introduce a shorthand system called the dot convention, which With il constant between tl and t2, diIfdt = 0, and by (IS.12a) and (IS.12b) and the
will immediately identify the proper sign. above,
Substituting (15.8) and (IS.9) into (15.7) and differentiating, by Faraday's law
dA[ dil di2
VI =- =LI-M12- (15.10)
dt dt dt
Reversing the roles played by the two coils, it follows that
dA2 di2 dil
V2 = -dt = L 2-dt M2I-
dt
(15.11)
Evaluating the integrals and using (lS.14) for w(td with il (tI) = il (t2), the stored energy
where L2 is the self-inductance of coil 2 and M21 is the mutual inductance relating current in the pair of coupled coils at time t2 is
in coil 1 with the resulting voltage induced in coil 2. Note from these equations that
there need be no electric circuit connection between two coils in order for the current in (IS. IS)
one to influence the voltage in the other; a magnetic circuit (flux path) is sufficient. The
ability to produce electrical response yet maintain electrical isolation, that is, no electric Now, starting back at time t = 0 with no stored energy, if we reverse the roles
circuit connection, is one primary use of coupled coils. of coils 1 and 2, that is, hold il at zero until tl and then hold i2 constant thereafter, we
We next determine the energy stored in a pair of coupled coils. Along the way, would have W(t2) given by (1S.1S), but with the coil indexes reversed; that is,
we will discover the relation between the mutual inductances MI2 and M21 and between
the signs to be associated with the mutual inductance terms in the i-v laws (IS.IO) and (1S.16)
(1S.11). The instantaneous power delivered to the coils in Fig. IS.1 is
The last two equations each evaluate exactly the same quantity: the energy stored in
dil di2 the pair of coupled coils at a time t2 when the currents are i 1(t2) in coil 1 and i2 (t2)
PI = vIii = Lli I - M 12i I -
dt dt
(1S.l2a)
in coil 2. Only the manner in which this state was reached is different. Since the
energy stored does not depend on how a final state is reached, only on the value of
(IS.12b)
that state, these two quantities must be identical. This implies that the last terms on the
Assume that at t = 0 all currents and fluxes are zero, so there is no initial energy stored right of (IS. IS) and (IS.16) are equal, and two important facts are revealed. The first
in the coupled coils; w (0) = O. Since inductors are lossless elements, the total energy is that
W(tI) stored in the system at a later time t = tl is simply the sum of all the energy
(1S.17)
delivered to coils 1 and 2 between times 0 and tI:
t' (1 The mutual inductances are identical. We will use the symbol M for their common value,
W(tI) = Jo PI dt + Jo P2 dt (15.l3)
the mutual inductance of the pair of coupled coils. Rewriting (IS.l0) and (IS.11) using
612 Chapter 15 Mutual Inductance and Transformers Section 15.1 Mutual Inductance 613
M, we have the i-v laws for coupled coils: equation. Finally, suppose that at some time t we have instantaneous
coil currents i l (t) = 15 A and i2(t) = -9 A. Then the energy stored
in the pair of coupled coils at t is, by (15.19),
di l di2
vI=LI-M-
dt dt
(lS.lSa) w = !(6)(1S2) + !(5)(-9)2 - 4(15)(-9) = 1417.5 J
di l di2
v2=M-+L2- (IS.18b) The coefficient of coupling k is a measure of the degree to which the flux produced
dt dt
by one coil threads another and is defined by
Simplifying (15.15) or (15.16) by use of M and suppressing the time argument, the energy
stored in a pair of coupled coils is M
k=-- (15.20)
JL I L 2
614 Chapter 15 Mutual Inductance and Transformers Section 15.1 Mutual Inductance 615
has 900 turns. Find the fraction of 1>22 that threads coil 1 and the energy laws are
stored in the pair of coupled coils when i] = +1 rnA and i2 = +5 rnA. dil 3di2
Answer fa;
117.1 /hJ VI= 2 -+ - (l5.24a)
dt dt
15.1.3. Determine the coefficient of coupling for Exercises 15.1.1 and
dil 5di2
15.1.2. V2= 3 -+ - (l5.24b)
Answer 0.671; 0.671 dt dt
For the same coils, suppose that we assign the current and voltage
variables as in Fig. 15.2(b). Again (i3, V3) and (i4, V4) separately
satisfy the passive sign convention. This time one arrow points into
a dotted end and the other into an undotted end, so
di3 di4
V3 = 2 - - 3 - (15.25a)
The i-v laws for a pair of coupled coils, given in (15.18) and repeated here for conve- dt dt
nience, are \ di3 5di4
di] di2 V4=- 3 -+ - (l5.25b)
(15.23a) dt dt
vI=LI-M-
dt dt Examining the figure, if these are the same coils, then V3 = VI,
di l di2 i3 = ii, i4 = -i2, and V4 = -V2. Substituting these into (15.25)
v2=M-+L2- (l5.23b)
dt dt recovers (15.24), so the i-v laws are completely equivalent. It
To analyze a circuit containing coupled coils, we must know which signs to use. In does not matter which way the reference direction pairs are cho-
Section 15.1 we saw that the same sign must be used for both mutual inductance terms in sen. Separate sets of reference directions for the two coils may be
these equations. The shorthand notation we use for indicating which sign should be used freely chosen without regard to the dots. Of course, it will often be
is called the dot convention. Circuit diagrams will include a pair of dots, one for each slightly more convenient to choose the current arrows so that minus
inductor, and values of the two self-inductances and the mutual inductance, as shown in signs in the i-v laws may be avoided, as in Fig. 15.2(a), but not
Fig. 15.2(b).
Fig. 15.2.
To apply the dot convention, first assign current and voltage reference directions
The Laplace transform or s-domain version of the i-v laws (15.23) is very conve-
separately for each inductor so that each satisfies the passive sign convention (arrows for
nient for analyzing circuits involving initial condition responses or switching transients.
i l and i2 point into plus ends of VI and V2, respectively). Then the dot convention is
Using the Laplace transform differentiation property (Table 12.1), the i-v laws for cou-
the following rule: if both current reference arrows point into dotted ends or both into
pled coils in the s-domain are, from (15.23),
undotted ends of the inductors, use the plus sign for both mutual inductance terms in the
i-v laws. Otherwise, use the minus sign.
VI (s) = {sLII I (s) - Lli l (O-)} {sMI2(s) - Mi2(0-)} (l5.26a)
V2(s) = {sMII(s) - Mil(O-)} + {sL2I2(s) - L2i2(0-)} (l5.26b)
2H 5H v2
Vj
In the important special case of zero initial conditions, these simplify to
~ ~
3H 3H
VI (s) = sLIII (s) sMI2(s) (l5.27a)
(a) (b)
V2(s) = sMII(s) + sL2I2(s) (15.27b)
FIGURE 15.2 (a) Coupled coils; (b) same coils but different variables.
Let us find the i-v laws for the coupled coils in Fig. 15.2(a) and Consider the circuit of Fig. 15.3(a) and the desired circuit response
(b). Applying the dot convention for Fig. 15.2(a), we first check V3(t), t > O. The initial conditions at time t = 0- are all zero
that the passive sign convention is satisfied separately for (iI, VI) since the unit step u(t) does not energize the source until t = o. Fo;
and (i2, V2). Then, since the current reference arrows both point t > 0 there is only one unknown mesh current, i m , in Fig. 15.3(b).
into dotted ends, the plus signs in (15.23) are taken, and the i-v Assign reference directions (iI, VI), (i2, V2), which separately satisfy
616 Chapter 15 Mutual Inductance and Transformers Section 15.2 Circuits with Mutual Inductance 617
t=O IV
IH
2V 4H 4H
5Q 5Q
(a) (b)
FIGURE 15.4 Circuit for Example 15.4.
FIGURE 15.3 (a) Circuit for Example 15.2; (b) labeled for analysis.
Using the s-domain i-v laws (IS.26) to eliminate the voltages and
the passive sign convention for the two inductors as shown. The then rewriting in vector-matrix form for convenient solution, we have
mesh equation is 2
411 (s) + [4s1 1 (s) + 2s12(S) - 1] = - (lS.29a)
VI(S) + 3Im(s) + V 2 (s) + Slm(s) = 0 s
~]
1 We wish to find the current transfer ratio H(s) = 12 (s)/11 (s) for the
V 2 (s) + 212 (s) + [SI2(S) - s
(lS.2Sb)
circuit shown in Fig. IS.S and the frequency response H(jw). There
618 Chapter 15 Mutual Inductance and Transformers Section 15.2 Circuits with Mutual Inductance 619
In for coupled coils with zero initial conditions:
2H lp .s (15.33a)
4
(15.33b)
~ ~
(a) (b)
where, since both currents flow into undotted ends of their respective Note the minus sign in front of the term 1js Me. If the dot convention requires that the
inductors, the dot convention requires use of the plus sign for the positive sign be taken in the mutual inductance tenn, the corresponding terms (ljs Me)
mutual inductance term, and in the ifonn of the i-v laws are negative, and vice versa. In other words, if both current
V 2 (s) = SII(S) + 2s12(s) reference directions point into dotted ends or both into undotted ends of their inductors,
then minus signs will be needed in this form of the i-v laws.
Substituting the last into (15.30) and solving for the response 12(s)
Using (15.26), when the initial conditions are not zero, (15.31) becomes
gives
_!S2 [ VI(S)+L l i l (0-)Mi2(0-)J = [SL I SMJ [11(s)J
12(s) =
s2 + 2s + 2
i
II(s) V2(s) Mil(O-) + L2i2(0-) sM sL 2 12(s)
The same matrix is inverted as in (15.31); thus the i form for the i-v law is that shown
The desired transfer function is
in (15.33) but with VI(S) and V 2(s) in (15.33) replaced by VI(S) +LJil(O-) Mi2(0-)
_!S2
H(s) = 2 and V2(s) Mil (0-) + L2i2(0-), respectively.
s2 +!s
2
+2
and the frequency response of this circuit is We wish to determine the impedance Z(s) = V(s)jI(s) of the two-
!w 2 terminal subcircuit shown in Figure 15.6. The nodal equations are
H(jw) = (2 _ w2~ + j(wj2) I(s)
--?>- II (s)+ 2V I (s) - V2(s) = I(s)
+ (15.35)
12(s) + 4V 2(s) - VI (s) = 0
Equations (15.26) and (15.27) are the v forms of the i-v laws for a pair of coupled
coils, in which the voltages are found in terms of the currents. These are convenient for Yes) s 1 We will eliminate II (s) and 12(s) using (15.33). The equivalent
3 3 inductance parameters are
mesh analysis, as the examples have shown. For nodal analysis it is necessary to work
with the i form of these i-v laws. Rewriting (15.27) in vector-matrix form, we have Ij(s) t~ t lis)
(l5.36a)
[~~~~n = [~~~ ~~~J D~~~n (15.31)
Inverting the matrix yields fiGURE 15.6 Circuit for Example 15.6. (l5.36b)
620 Chapter 15 Mutual Inductance and Transformers Section 15.2 Circuits with Mutual Inductance 621
and using the coil i-v laws (1S.33) in (IS.3S), we obtain For instance, the phasor impedance Z(jw) for the circuit of Example IS.6 is, replacing
s by jw in (1S.38b),
[~VI(S) - ~V2(S)J + 2VI(s) - V2(s) = I(s) (lS.37a)
Z 'w _ 4(jw)2 + 9(jw)
(j ) - 7(jw)2 + 30(jw) + 18
[-~VI(S) + ~V2(S)J +4V2(s) - VI(S) = 0 (lS.37b)
-4w 2 + j9w
(1S.40)
= ------~------
(18 - 7( 2 ) + j30w
Solving the second equation for V 2(s) yields
From Fig. IS.6 we see that in dc steady state, the input is shorted by the leftmost inductor
s+6 and the impedance is therefore zero, consistent with setting w = 0 in (IS.40). In the high-
V2(s) = 4s + 9 VI (s) frequency limit, the inductors in Fig. IS.6 behave like open circuits and the impedance
Substituting into (IS.37a) eliminates V 2(s), and
1-
is that of a l-Q resistor in parallel with a Q resistor, or LQ. This checks with the
limiting value in (IS.40) as w ---+ 00. 7
To sum up this section, circuits containing coupled coils may be analyzed without
(~+ 2) VI(S) - (~+ 1) (:s :69 ) VI(S) = I(s) (lS.38a) undue difficulty using either mesh or nodal analysis. The mesh equations are written
first in terms of coil voltage variables, and then the v form of the i-v laws (1S.26), or
V I (s) 4s 2 + 9s (1S.27) if there are zero initial conditions, is used to eliminate these voltages. The dot
Z(s) - - (IS.38b) convention is used to determine the proper sign to be used with the mutual inductance
- I(s) - 7s + 30s + 18
2
terms in the i-v laws. Similarly, nodal analysis equations are first written to contain coil
current variables, which are replaced using the i form of the i-v laws. The i form uses
Compared to the v forms (1S.26) and (IS.27), the i forms of the i-v laws for cou- equivalent inductance parameters Lei, L e2 , and Me defined in (IS.34) and, particularly
pled coils are more cumbersome. This makes mesh analysis preferable to nodal analysis when there are zero initial conditions, is somewhat less convenient to use than the v form
for most problems. It is sometimes the case, however, as in Example IS.6, that there are and mesh analysis.
significantly fewer node voltage variables than mesh current variables, and nodal analysis
is warranted in these cases.
While the s-domain analysis presented in this section may be applied to any EXERCISES
circuit problem, phasor analysis is more efficient for finding the forced or ac steady-
state response to sinusoidal excitation. As shown in Chapter 14, s-domain relation-
ships computed with zero initial conditions may be converted to the phasor domain 15.2.1. Find il for t > a if M = 1/.J2 H and the circuit is in steady state
at t = 0-.
simply by replacement of s by jw. The phasor i-v laws for coupled coils are, from
Answer 6 - 2e- 4 / 3t A
(1S.27),
1=0
EXERCISE 15.2.1
II = _._I_VI _ (_._1_) V 2
]WLel ]wMe 15.2.2. Find V2 in ac steady state if VI = 8e- 2t cos t V.
EXERCISE 15.2.2 Answer ../ie- 2t cos(t + 45) V
12 =- (_._1_) VI + _._I_V2 15.2.3. Find the transfer function V2/V I. Note that each inductor is cou-
] wMe ] wLel
pled with two others, so the i-v laws for each pair of coils should have two
mutual inductance terms.
622 Chapter 15 Mutual Inductance and Transformers Section 15.2 Circuits with Mutual Inductance 623
Answer
3(lOs + 2) impedance levels may be transformed using this device, as we shall see. In addition,
30s 2 + 49s + 11 transformers are often used to electrically isolate the primary circuit from the secondary.
15.2.4. Write the i forms of the s-domain i-v laws for Fig. IS.2(a) with Consider the basic circuit shown in Fig. 15.8. Since transformers are most fre-
zero initial conditions and Fig. lS.2(b) with arbitrary initial conditions.
quently used in ac steady-state circuits, we shall use phasor analysis. The self-inductance
S 3
Answer (a) II = -VI - -V2 of the primary winding is L I , that of the secondary is L 2 , and the mutual inductance of
s s the transformer is M. The external circuit attached to the primary has been simplified
3 2
12 = --VI + -V2 to its Thevenin equivalent, with Thevenin equivalent voltage source Vs and impedance
s s
Zs. The load has also been reduced (if necessary) to its Thevenin equivalent, and the
(b) II = (SjS)[V3 +2i3(O-) - 3i4(O-)] + (3jS)[V4 - 3i](O-) + Si4(O-)]
12 = (3jS)[V3 + 2i3(O-) - 3i4(O-)] + (2jS)[V4 - 3iI(O-) + Si4(O-)] absence of an equivalent source in series with ZL implies that there are no independent
EXERCISE 15.2.3 sources in the load circuit attached to the secondary.
Note carefully the reference directions shown in Fig. 15.8. Following the standard
convention for transformers, the primary current and voltage are defined to satisfy the
passive sign convention, while the secondary current and voltage (12, V 2) do not. Our
statement of the dot convention in Section 15.2 requires that the passive sign convention
be satisfied by both coils separately, so we shall use 12 and -V2 as the current and
voltage variables for the secondary when writing i-v laws.
The most common practical use of the phenomenon of mutual inductance in electric The phasor i-v laws for the transformer are, from (15.39),
circuits is in transformers. A transformer is a two-port circuit containing coupled coils
wound around a common core. A typical transformer is sketched in Fig. 15.7. Each coil, VI = jwLIII - jwMI2 (l5.41a)
or winding, terminates in a pair of nodes called a port, which is available for conn~ction -V2 = - jwMII + jwL2I2 (l5.41b)
to external circuitry. One of these ports is called the primary. Usually, the pnmary
is connected to an external source, as shown in the figure. The other port, called the Since II points into a dotted end and 12 into an undotted end, the dot convention dictates
secondary, is usually connected to a load. In power transformer applications, ~e load minus signs for the mutual inductance terms. The coil voltage variables used in (15.41)
might be a passive impedance, while in electronics applications it might be the lllput to are those that satisfy the passive sign convention together with 11 and 12, that is, VI and
a filter, amplifier, or other active circuit. - V2. Noting from the figure that V 2 = Z L I 2 , we may solve (15.41b) for 12 in terms of 11
The rectangular ring in Figure 15.7 around which both coils are wound is called the jwM
core of the transformer. The core is constructed largely from ferromagnetic material to 12 = II (15.42)
facilitate production of strong magnetic fields in response to the currents flowing in the
ZL + jwL2
coils and to guide these fluxes through both primary and secondary windings. The core Using this expression to eliminate 12 from (l5.41a), we have
can be a simple doughnut of soft iron or a sophisticated layered construction of ~omp~s
ite materials with lamina containing alloys of cobalt, selenium, and other matenals WIth VI = (jWL I + w2~2 ) II (15.43)
desirable magnetic properties. ZL + JwL 2
Transformers are among the most versatile of electric devices and are widely used This expression has an interesting interpretation. It suggests that the impedance Z =
over the full range of circuits applications, including power, industrial, and consumer VI/II looking into the primary port is equivalent to a simple (uncoupled) inductor of LI
electronics. Their main use is to scale, or transform, primary circuit variables to lev-
els better suited to drive the circuit connected to the secondary. Current, voltage, and
:+
,
,
v,
,, -
Primary ~ Secondary
Core jwM
FIGURE 15.7 Transformer. FIGURE 15.8 Basic transformer circuit (phasor domain).
624 Chapter 15 Mutual Inductance and Transformers Section 15.3 Mutual Inductance and Transformers 625
Primary ~
j{J)M
FIGURE 15.9 Secondary winding and load reflected into primary. (a) (b)
12
=
jwM
(15.45) PL = RLIIL12 = 2500(0.031 2 ) = 2.4 W
2
With direct connection, 2.4 W is delivered to the load at 155.4 V
and 3.1 rnA ac steady-state amplitude. Turning to the transformer-
and since coupled circuit Fig. 15.IO(b), the reflected impedance is
(15.46)
then, by (15.43)-(15.46),
Examining Fig. 15.9, the primary current and voltage are
I, = V2 156LQ .
V2 jwMZL
(15.47)
+ ZR + Zs = 38.1 + j80.6 = 0.75 - J1.58
jwLl A
V, jwL,(ZL+jwL2)+w2M2
V, = (jwL, + ZR)I, = (18.1 + j80.6)(0.75 - j1.58)
= 141 + j31.9 V
Equations (15.45) and (15.47) show how the primary current and voltage are transformed
in passage through the transformer, and (15.44) shows how the secondary impedance is Then, by the primary-to-secondary ratios (15.45),
transformed when reflected into the primary circuit. I jwM j377 .
2 = ZL + jWL2 I, = 5000 + j3770 (0.75 - J 1.58)
A 60-Hz household circuit that supplies Vs = 156LQ V at source
impedance Zs = 20 Q is to be used to drive a load ZL = 5 kQ.
= 0.103 - jO.021 A
Determine the current, voltage, and power delivered to the load if it V2 = Z LI 2 = 5000(0.103 - jO.021) = 516 - j106 V
626 Chapter 15 Mutual Inductance and Transformers Section 15.3 Mutual Inductance and Transformers 627
The power delivered to the load in this case is M
II
~ 12 II LI-M L 2 -M 12
...:-::-
PL = T
R 1112
= 250010.103 - jO.0211 2 = 27.6 W
-----'7 ~
+
~
628 Chapter 15 Mutual Inductance and Transformers Section 15.4 Ideal Transformers
That is, the self-inductance of a given inductor is proportional to its turns squared. The
constant fh is determined by the detailed physical properties of the magnetic path. If a
pair of coupled coils 1 and 2 share the same magnetic path, in other words, if they are
unity coupled, then fh = fh and by (IS.48) the self-inductances of unity-coupled coils
are in the ratio of their turns squared,
L2 Ni. 2
-=-=n
LI Nr
where n = N2INI is called the turns ratio of the transformer. (a) (b)
In the case of unity coupling, M = ,JLIL2 and, by (1S.47), the secondary-to-
primary voltage ratio is FIGURE 15.12 (a) Ideal transformer circuit symbol; (b) ideal transformer
with primary source and secondary load.
-V 2 - j wZ2,JL IL 2 - J2
- - n (1S.49) \
VI - jwLI(Z2+jwL2)+w2LIL2 - LI - lines is drawn to suggest the ring of ferromagnetic material comprising the core of this
unity-coupled transformer.
Thus, for any unity-coupled transformer, the secondary-to-primary voltage ratio has the
property we seek: It is a constant independent of load or frequency. The value of the Specify the i-v laws for the two ideal transformers shown. In
voltage ratio is simply the turns ratio n. Fig. IS.13(a), (II, VI) satisfies and (12 , V 2 ) violates the passive
The secondary-to-primary current ratio (IS.4S) in the unity-coupled case is sign convention, as required in the statement of the dot conven-
12 jw~ tion for ideal transformers. Since both currents flow into dotted
= (15.50) ends, we must take minus signs. The i-v laws for Fig. IS.13(a) are
h + jwL 2 11 12 12
~ ~ -<E-"- consequently
Even with unity coupling, the current ratio still generally depends on w and ZL. If,
however, we take L2 to be unboundedly large, ZL may be neglected in the denominator V2 = -nV I
of (IS.S0), and + + - 1
12 = --II
12 = jw~ = {L; = ~ (15.51)
VI
II
VI
II +
V2 n
In Fig. IS.13(b), the given primary variables (II, Vd do not satisfy
II jwL2 YL; n
11 the passive sign convention, but (II, -VI) do, so we will use this
~
Transformers satisfying both (15.49b) and (1S.S1) have current and voltage ratios that 1: n 1: n
pair as our chosen primary variables. (12 , V 2 ) violates the passive
do not depend on load or frequency and are said to be ideal transformers. An ideal (a) (b)
sign convention, as required. Then, using the selected sets of port
transformer is a unity-coupled transformer whose self-inductances LI and L2 are both variables (II, -VI) and (12 , V2 ), one current points into a dotted
unboundedly large but whose ratio L2/ L I is finite. Ideal transformers satisfy the i-v laws FIGURE 15.13 Two circuits for Example 15.8. and the other into an undotted end, so plus signs will be used in the
i-v laws. Writing -VI in place of VI in (IS.S2), the i-v laws for
Fig. IS.13(b) are
V 2 = nV I (15.52a) V 2 = +n(-Vd
1 1
12 = -II (15.52b) 12 = +-11
n n
or V2 = -nV I
1
where the dot convention is applied as follows: Define the primary variables (II, V I) to 12 = +-11
satisfy, and the secondary variables (12, V 2) to violate the passive sign convention. Then if n
II or 12 points into a dotted end while the other points into an undotted end, use plus signs Figure IS.12(b) shows an ideal transformer connected to a source and load. The
in both i-v laws (IS.S2); otherwise, use minus signs in both. This is an awkward but com- impedance looking into the primary port is
pletely equivalent restatement of the original dot convention, which is necessitated by the
standard transformer practice of reversing the current reference direction in the secondary.
The circuit symbol for an ideal transformer, shown in Fig. IS.12(a), is similar to (15.53)
that for a nonideal transformer except that the turns ratio is specified rather than the
inductances L I , L 2 , and M (all three of which are infinitely large), and a pair of parallel
630 Chapter 15 Mutual Inductance and Transformers Section 15.4 Ideal Transformers 631
+ +
II
I: n
(a) (b)
FIGURE 15.14 (a) Ideal transformer and secondary load; (b) equivalent ac generator transformer I long lines transformer 2 load
when reflected into the primary circuit. (a)
\
Thus an ideal transformer together with its secondary load ZL is equivalent to an im- 0.1 a
pedance of value Z L /n 2 reflected into the primary circuit, as shown in Fig. 15.14. It is
often convenient to use this reflected impedance to eliminate the transformer and one
loop before analyzing a circuit containing an ideal transformer.
Vs 4
The simple relationships (15.52) and (15.53) reveal most of the main uses of trans-
formers. Equation (15.52a) shows that an ideal transformer can be used to step up a
voltage level if the secondary winding has more turns than the primary, n > 1, to step
down the voltage level if n < 1, which results in an equivalent increase in the current a': I: 10 b', I: n
level as seen in (15.52b), and to impedance scale a load. Each of these uses is illustrated
in the next example.
(b)
Figure 15.15 shows a power delivery circuit. A hydropower-driven FIGURE 15.15 Circuit for Example 15.9: (a) pictorial; (b) schematic.
ac generator produces the Thevenin equivalent voltage phasor Vs
through a O.I-Q Thevenin equivalent impedance. Its output is in- real. Thus, for maximum power supplied to the rest of the circuit
creased using a 10: 1 step-up transformer for more efficient trans- by the source,
mission by long lines. The transmission lines have an equivalent
1 + (4/n2) 1
impedance of 1 Q. The power is to be delivered to a 4-Q load Z2 -- ---:-::--
102 10
through a second ideal transformer. We wish to determine the turns
ratio n of the second ideal transformer to maximize the power sup- Solving for the required turns ratio,
plied by the source. The equivalent of this circuit to the right of the
terminals b-b', the primary of the right transformer, is an impedance n=!i=~
of value Thus a 10: 1 step-up transformer is used to generate higher voltage
h 4 for efficient transmission, and then a 2:3 step-down transformer is
ZI=-=-
n2 n2 used to impedance match to the given load.
The left transformer has an equivalent secondary load impedance of
Just as we can reflect the load into the primary, we may also reflect the source into
1 Q in series with Z2, or 1 +4/n 2 Q. Reflecting this into its primary
the secondary circuit. The open-circuit voltage V 2 in Fig. 15.16(a) is
circuit, the impedance to the right of a-a' is
1 + (4/n2)
Z2 = 102 VT = nVI = nVs (15.54)
632 Chapter 15 Mutual Inductance and Transformers Section 15.4 Ideal Transformers 633
10
4 4
1 :n
1: n
(a) (b)
(a) (b)
fiGURE 15.17 (a) Circuit of Fig. 15.15 reflected into secondary of left
FIGURE 15.16 (a) Source and ideal transformer; (b) equivalent when re-
transformer; (b) reflected again into secondary of right transformer.
flected into secondary circuit.
step-down transformer draws more power out of the source, but less
when the primary and secondary are reversed, has turns ratio l/n. By (15.53), the of this power is dissipated by the 4-Q load and more by the l-Q
impedance looking into the secondary is transmission lines than using the turns ratio n = 2/-JTI.
(15.56)
leading to the Thevenin equivalent circuit shown in Fig. 15.16(b). An ideal transformer, !
Recalling from Chapter 10 that S = VI* is interpreted as power delivered to the element
together with its primary voltage source Vs of source impedance Zs, reflects into the sec- if the terminal variables satisfy the passive sign convention, the complex power delivered
ondary circuit as a source nVs of source impedance n 2 Zs. The sign follows from the to the secondary is
dot convention. Both reflecting the load into the primary (15.53) and the source into
the secondary (15.54)-(15.55) result in an equivalent circuit with one less loop and no (15.57)
transformer. Note that reflection to the secondary eliminates the primary subcircuit, so
it is most useful when the primary variables are not required. Similarly, reflection into
Thus the sum of the power delivered to the primary and secondary equals zero, and the
the primary, as in (15.53), obliterates the secondary variables, so it is best suited where
ideal transformer is lossless. The source supplies SI to the primary, and the secondary
primary variables are of interest.
dissipates -SI; that is, it supplies SI to the rest of the circuit. Power flows from primary
to secondary without losses. Of course, perfectly lossless flows of power are not possible,
This example continues Example 15.9. For the same circuit, suppose and the ideal transformer is just that, an idealization. But transformers with efficiencies
that we change the goal to maximizing power delivered to the 4-Q beyond 95% (less than 5% power loss) are common, and the ideal transformer is a very
load. Reflecting the source and 10: 1 step-up transformer into its useful idealization that is quite well approximated in practice.
secondary circuit yields the equivalent shown in Fig. 15.17(a). Then
reflecting the primary circuit of the remaining transformer into its
secondary yields the circuit of Fig. 15.17(b). For maximum power
into the 4-Q load, we need Iln 2 = 4, or EXERCISES
2
n=-- 15.4.1. Find the phasors VI and V2 in Fig. 15.6 if a current source I = Is
ffi is connected across the input and and the transformer is replaced by an ideal
So the maximum power delivery to the 4-Q load occurs if we use one with turns ratio n = 10.
a step-down transformer with turns ratio 2:ffi. It is interesting Answer Is/382; 10 Is/382
to note that the turns ratio for maximum power supplied by the 15.4.2. Find VI, V2, II, and h.
source computed in Example 15.9, 2:3, is not the same. The 2:3 Answer 10/-36.9 V; 50/143.1 V; 2~ A; 0.4/180 A
634 Chapter 15 Mutual I nductance and Transformers Section 15.4 Ideal Transformers 635
The coupling coefficient for the coupled coils is
k =
M
JL 1 L z = If
! fi
= V"3
II V2
which is entered in the k statement of the SPICE input file .
Example 15.11
EXERCISE 15.4.2 *IG 0 1 AC 1
R1 1 0 1
15.4.3. Find the nTIS values of the primary and secondary currents in
Fig. lS.12(b) if the phasor Vg = 16 V, Zg = 2 + ) Q, Z2 = 8 - 4) Q, and L1 1 0 0.5
n = 2. L2 2 0 0.33333
Answer 2./2 A; ./2 A K L1 L2 0.8165
R2 1 2 1
R3 2 0 033333
.AC DEC 1 1 1K
.PRINT AC VM(1) VP(l)
.END
Mutual inductance is signaled in SPICE by the inclusion of an element statement of the The output file contains the print table
form
FREQ VM(l) VP(l)
K LXXXXX LYYYYY KVAL
1.ooE+00 5.24E-01 1.64E+01
L XXXXX and L YYYYYare the two inductors, and KVAL is the value of the coupling 1.ooE+01 5.71E-01 1.85E+00
coefficient. k = M/,JL 1 L 2 , where Ll and L z are the self-inductances and M is the 1ooE+02 5.71E-01 1.86E-01
mutual inductance as in (15.20). The value of k is required to be in the range 0 < k < 1, 1.ooE+03 5.71E-01 1.86E-02
so ideal transformers must be approximated using k set just below 1.000. For purposes
of the dot convention, the dots are taken to be located at the first nodes specified on the Our use of a unit current source phasor makes the terminal volt-
L XXXXX and L YYYY element statements. age phasor V( 1) identical to the impedance of this circuit. Thus
VM( 1) and VP ( 1) in the table are the magnitude and angle of the
impedance at each frequency. From our previous result (I5.38b),
15.11 We will use SPICE to determine the impedance at the frequencies
1 Hz, 10 Hz, 100 Hz, and 1 kHz of the circuit analyzed in Exam- . -4wz + j9w
ple 15.6 and redrawn as Fig. 15.18. Z(jw) = (I8 -7wZ) + j30w
i(t)
Evaluating at w = 2n, we obtain
--?-
-158 + j56.5
+ Z(j2n) = = 0.524/16.4
-258 + j188
which agrees with the first line of the SPICE output table. The other
vet) lQ
values may be checked similarly.
SPICE does not permit "floating" nodes; that is, there must always be a dc path
between every pair of nodes so that the voltage of each node relative to the reference
node (node 0) is well defined. If we wish to use SPICE to analyze a circuit containing a
fiGURE 15.18 Circuit for Example 15.11. transformer that splits the circuit into two electrically isolated subcircuits, we must take
636 Chapter 15 Mutual Inductance and Transformers Section 15.5 SPICE and Coupled Coils 637
special steps to prevent SPICE from complaining. A shorting wire can be introduced The load voltage at which the power is reduced by 90% is VM(3)2/
between the two inductors without affecting any element current or voltage. If there is 10 = (0.10)(2693) or VM(3) = 51.9 V. Rerunning the input file
~ .connection .between subcircuit~ containing the two inductors, but it is a purely capac- with the value of k decreased by cut-and-try, we find that this con-
ItIve connectIOn, one whose dc Impedance is infinite, a very large resistance should be dition is met for k = 0.994 (to three significant digits). Note that
introduced into the connection added in order to prevent significant current flow in this only a small leakage flux is required to sharply reduce the efficiency
dummy connection. of a transformer-coupled power delivery circuit.
Determine the ac steady-state power delivered to the 5-Q load re- Finally, an ideal transformer may be modeled for SPICE as a pair of controlled
sistor if the transformer is unity coupled (Figure 15.19), and de- sources satisfying the i-v laws (15.52) for this type of device. The model is shown in
termine the coupling coefficient at which the power is reduced to Fig. 15.20.
10% of this value. The SPICE input file for the unity-coupling
evaluation is shown below. Note that the lower ends of the in-
~ ~ ~ ~
ductors have been tied together to add a dc path between the pri- 0
+ + + +
mary and secondary subcircuits and that the coupling coefficient
has been set at 0.99999 to comply with the permissible range
O<k<1. VI Vz VI nVI Vz
l:n
Example 15.12 0
*V1 1 0 AC 156
(a) (b)
RS1 2 1 FIGURE 15.20 Ideal transformer model for SPICE: (a) circuit device; (b) de-
LPRI 2 0 1 pendent source model.
LSEC 0 3 10
K LPRI LSEC 0.99999
C1 3 0 50U
RL 3 0 5 EXERCISES
.AC LIN 1 60 60
.PRINT AC VM(3)
15.5.1. Find V2(t) for VI (t) = Vg cos(lOOOt + g) using SPICE.
.END
Answer Use a pair of dependent sources to model the ideal trans-
former. V2(t) = 10.45Vg cos(lOOOt + g - 145.3).
lQ
2 p,F
~ EXERCISE 15.5.1
M
638 Chapter 15 Mutual Inductance and Transformers Section 15.5 SPICE and Coupled Coils 639
15.2. A pair of coupled coils have self-inductances Ll = 15.9. Find i for t > 0 if i(O) = 0 and v(O) = 4 V.
EXE15-5.2
1 H, L2 = 5 H. If il(t) = 5cos2t A, i2(t) = 2cos2t A,
*
V1 1 0 DC 2
what is the maximum amplitude A of the sinusoidal flux
lH
linkage Al (t) = A cos(2t + ) threading coil I? The mini- i 2
R1 1 2 4 mum amplitude A?
---?>
L1 0 2 4 IC=O 15.3. When a' and b' in the figure for Problem 15.1 are e~1
'\
L2 0 3 4 IC=-0.5 +
shorted the inductance between a and b is 1 H, and when
KL 1 L2 0.5 a' is sh'orted to b, the inductance between a and b' is 7 H. v ~F IH
V2 3
4 3 DC -1 Find the mutual inductance M. e
L3 4 5 1 IC=-0.5 15.4. Ll = L2 = 2 H, M = 1 H and positive dirfdt leads 2Q
R2 5 0 2 to negative V2. Find il and i2 for t > O.
.TRAN .1 10 UIC
.PRINT TRAN I(V2) FIGURE P15.9
.END
15.10. Determine Z(s), the impedance looking into this
circuit.
6V 12 V
5H
Elements may interact electrically, through the wires that connect them, or magnetically,
when the magnetic field spawned by one element threads the other. Mutual inductance 2H
FIGURE P15.4
causes the terminal current and voltage of one inductor to depend on those of another
through their electromagnetic interaction.
The voltage induced in a coupled coil consists of a term proportional to the derivative 15.5. If w(O) = 0, determine the power pet) and energy FIGURE P15.10
of its own current plus terms proportional to the derivatives of the currents flowing w(t) for the coupled coils of Problem 15.1 above, with the
through each of its coupled coil partners. indicated currents.
1mi115.n. Find i2(t) in ac steady state.
15.6. Suppose Ll = 1 H, L2 = 5 H, M = 2 H in the
The dot convention requires positive coupling terms in the i-v laws if both inductive
figure for Problem 15.1. Find the peak instantaneous p~wer
currents flow into, or both out of, the dotted ends.
in watts supplied to the inductors if i 1 (t) = 3 cos t A, l2 = lQ
The coupling coefficient between two coupled coils is their mutual inductance divided
by the geometric mean of their self-inductances.
A transformer is a device for changing voltage, current, or impedance levels in a circuit.
It consists of a pair of coupled coils on a common core.
sin(t + 45) A. Is this power dissipated in the form of heat?
Explain.
15.7. When two identical coils are each made to simul-
taneously carry a dc current of 1 A, they each have flux
linkage i 1 = i2 = 10 Wb. When they are shielded from
5COS7tv[3+ ~H e(J
~
lH
e H
t
4Q
li2(t)
An ideal transformer steps up the voltage by the turns ratio and steps down the current 4
one another so no flux from one threads the other, the flux
by the negative inverse turns ratio.
linkage rises to 12 Wb. Find M in both cases. FIGURE P15.11
An ideal transformer dissipates no electrical power. 15.8. Write the i-v laws in the time and s domains. In
both cases il (0-) = 1 A, i2(0-) = O.
1mi115.12. Find the ac steady state value of VR(t).
PROBLEMS 2H
15.1. Suppose Ll = L2 = 0.1 H, M = 0.05 H, Nl =
N2 = 1000 turns. Find the leakage fluxes Ll, L2 and
the total fluxes 1 and 2 threading the inductors if i 1 = 3 cos 4tV
3t + 2 A, i2 = 3 A.
15.17. (a) Find v for t > 0 if Vg = 4u(t) V and (b) find fiGURE 1"15.21
fiGURE 1"15.13
\
=
the steady-state value of v if Vg = 4 cos 8t V and the output
terminals are loaded with a resistor of 8 Q.
~ 15.22. Find the voltage V across the source. fiGURE 1'15.27
15.14. Find the steady-state currents il and i2'
-j2 lOQ
42sin8tV 8
=
b fiGURE 1"15.17
fiGURE 1"15.22
+
fiGURE 1"15.14 15.18. Find II and 13 using mesh analysis.
01 0
0.2H 5Q v
015.23. Determine the reflected impedance ZR, current
I~
of
IIilll and voltage ratios h/ll, V2/V 1 for the circuit of Prob-
lem 15.21.
15.24. Determine the reflected impedance Z R, current
and voltage ratios 12/11, V2/V I for the circuit of Prob-
lem 15.22.
fiGURE 1"15.28
=
015.25. Solve 15.21 using the uncoupled coil model of
fiGURE 1'15.18 IIilll Fig. 15.11 and nodal analysis. 15.29. Find i(t) for t > 0 if i(O+) = o.
15.26. Find I L .
15.19. Find v for t > 0 if ig = 2u(t) A. Use mesh analysis 2Q i(t)
IH 1Q and assume no initial stored energy at t = O. ,..----'1.11/\,----, 1:5 ....-----="-----,
4
'11 ~
50Q
fiGURE 1"15.15 100LO
0.01 P
,--_2_Q_,--_2_H--, H
15.16. Find the steady-state current i2' fiGURE 1"15.19 fiGURE 1"15.26 fiGURE 1'15.29
15.30. Find the average power delivered to the 8-Q 15.40. Plot i for 0 < t < 5 s in the circuit Problem 15.9. 15.43. Show that a real transfonner, with 0 :s k :s 1,
resistor. 15.41. In the circuit of Problem 15.28, replace the lO-Q satisfies the passivity condition wet) ::: 0 for all t.
resistor bridging the transfonner with a series connection g 15.44. Find il (t) in ac steady state. The 4 Hand 6 H
24n
4 of a lO-Q resistor and a 1000-rnF capacitor. Use SPICE to IillII coils are coupled with Ma = 2 H (filled dots). The 6 H
plot the frequency response for the voltage v of the resulting and 5 H coils are also coupled with Mb = 4 H (open dots).
circuit in the interval 1 < f < 50 Hz. The 4 Hand 5 H coils are not coupled.
2cos4tA
More Challenging Problems
FIGURE P15.34 r---~ ~)--,----....,
15.42. When b is shorted to b' , the other inductor satisfies
g ~5.35.. Fi~d II i~ Problem 15.18 by reflecting impedances VI = 4diI/dt. When a is shorted to a', the other inductor IF 30
IIlliJ mto pnmanes tWIce. satisfies V2 = 2di2/dt. If LI = 5 H, find M and indicate
o
whether the dot should be at b or b' .
8n
g 15.36. Find the real average power delivered by the 5H
IillII source. 15 cos 41 V ~
4H
\v ~
t'"
1jt12
-j 2 +. +
v~ 'LI\ , ' \Ai L2 ~2
FIGURE P15.30 FIGURE P15.44
a' h'
1l0Lov rms +
15.45. Find 13 in Problem 15.18 by reflecting sources into
15.31. Write the i-v laws in the time and s domains. Do secondaries twice.
FIGURE P15.42
not assume initial currents are zero.
1:3
FIGURE P15.36
+ SPICE Problems
15.37. Use SPICE to determine the ac steady state values
VI
II V2 VI
II V2
for the four port variables.
+ + -~
1:8
(a)
FIGURE P15.31
8:3
(b)
50 cos 3771 V
[j
~
il+
-
VI
In.
2H
~
2H
6n
FIGURE P15.32
Bell was. born ill Edinburgh. Scotland~.His father, .Alexander M61ville Bell,
was a wel1..,knownspeechteacher and his grandfather; Alexander Bell, was also
a speech teacher. Young. Bell, after attending the University of Edinburgh and
the University of London, also qec.ame a speech teacher. In 1866Bell became
interested in trying to transmit speech electrically after reading a book describ-
ing how vowel sounds could be made with tuning forks. Shortly afterwards,
Bell's two brothers died of tuberculosis and Melville Bell moved his family to
Canada for health reasons. In 1873 young Graham became a professor at Boston
University and began his electrical experiments in his spare time. It was there
that he formed his partnership with Watson and went on to his great invention.
Bell's telephone patent was the most valuable one ever issued. and the telephone
opened a new age in the development of civilization.
647
Chapter Contents
II 16.1 Two-Port Circuits II 16.5 SPICE and Two-Port Circuits +
circuits may be anchored. The requirement that the current into the circuit at one terminal
of the port equal that flowing out the other terminal of the port is called the port condition.
A two-port circuit has two ports available for external connection, each of which
is assumed to satisfy the port condition. Reference directions for the port variables will
usually be assigned as shown in Fig. 16.1(b). Most often the currents flowing out of the
lower terminals will not be explicitly labeled but are understood to be equal and opposite
Two-port circuits are those in which two pairs of terminals, or ports, are identified as to the currents of the corresponding upper terminals. The left port will be referred to
attachment points for sources, loads, or other circuitry. Two-port analysis is most efficient somewhat arbitrarily as the input port and the right as the output port. Our goal in this
not when all the currents and voltages in a circuit are of equal interest, but rather, it is the discussion is to characterize two-port circuits, that is, to learn how to write down the
relationship between a designated input circuit variable (current or voltage) and output relationships linking their port variables, and explore their uses.
variable that matters. For instance, while there are many currents and voltages present Why might we wish to represent a circuit as shown in Fig. 16.1(b), a featureless
within the chassis of a graphics equalizer we may purchase for audio applications, it is "black box" out of which four wires protrude? One answer touches what might be called
specifically the relationship between the voltage routed to the input port and that which the engineering method of problem solving. When confronted with a challenging problem,
emerges at the output port which determines the usefulness of this device. Presumably, the break it up into a set of manageable subproblems, solve each separately, and then link
output has superior frequency response characteristics, and the other voltages and currents the subproblem solutions together. Remove the case from a radio receiver sometime and
are of interest only to the extent that they permit the desired input-output behavior. Two- take a close look inside. There is considerable complexity in there: circuit boards, wires,
port analysis allows us to focus on this relationship, collapsing the many equations of and switches forming intricate patterns. No designer could have, or did, visualize that
nodal or mesh analysis to a single pair of equations governing the port variables. circuit in its detailed totality as a unified solution to the design problem. Most likely, the
We begin by defining the relationships and properties of two-port circuits. In designer started by considering the first subtask needing to be done and then proceeded
Section 16.2 the variety of two-port equations is catalogued and the method for writing toward a solution by breaking the problem into several simple, manageable steps. To get
these equations from circuit diagrams is developed. Equivalent models for two-ports are started, an antenna is needed to capture the radio signal. We route the output from the
described next, and the effect of interconnecting two-ports together in series, parallel, antenna into an amplifier to strengthen it before proceeding. Then we must narrowband
and cascade is considered in Section 16.4. The chapter concludes with a look at SPICE filter the signal to the station we are interested in receiving. Then ... and so on. Each task
support for the derivation and use of two-port circuits. defines a simple transformation, some desired relationship between the module's input
signal and its output. The overall design (Fig. 16.2) is executed as an interconnection
of two-port circuits, with the output from one two-port becoming the input to the next.
To apply this general and very useful strategy in the context of linear circuits, we need
to understand its basic building block, the two-port circuit, how it is described, and how
these descriptions may be combined. This is the main work of the remainder of this
chapter.
A one-port circuit [Fig. 16.1(a)] contains exactly two terminals at which connections to A second reason for our interest in two-port circuits is that they focus attention
external elements are permitted. Since no other external connections are permitted, if where it often matters the most. In many circumstances we are interested in the input-
current i 1 (t) flows into one terminal, by Kirchhoff's current law the same current must output behavior of a circuit, not in its detailed interior currents and voltages. Requiring
flow out the other. This pair of terminals forms a port, or terminal pair, to which other that only the two sets of port variables be involved in a two-port description effectively
These two equations are the desired two-port description. They are not the only two-port
description possible, as we shall see shortly, but they form a complete characterization
of the two-port in the following sense: given the two-port description (16.4), then for any
Input port Output port specific pair of port termination circuits we can determine its input-output behavior, that
termination termination
circuit circuit is, all the port variables, as will be illustrated in Example 16.1. The only requirement is
that the two-port circuit satisfy the port condition at its input and output ports.
FIGURE 16.3 Two-port in s-domain with port terminations shown. (De- The two-port description (16.4) is called the impedance, or z-parameter, two-port
pendence of all quantities on 5 has been suppressed.) description. The name follows from the observation that each parameter Zl1, Z12, Z2J,
Z22 in these equations has units of ohms. The impedance two-port description may be ~
written compactly as +
4 V2 1 ~ 1..
2 s
(16.5)
(a) (b)
where the square matrix is called the impedance matrix or the z-parameter matrix.
FIGURE 16.6 (a) One pair of port terminations; (b) another pair.
We wish to find the impedance two-port description o~ the circuit
shown in Fig. 16.4 and use this description to determ~ne .the port \ To find them we perform the experiment shown in Fig. 16.5(b), re-
currents and voltages under the two different port terrmnatlOn con- locating the open circuit to the input port (II = 0) and the current
ditions shown in Fig. 16.6. . source 12 to the output port. There is no current through the leftmost
To find (16.5), we perform the experiment. shown III two resistors, and the port voltages under these conditions are
+ + Fig. 16.5(a), open-circuiting the output port to estabhs~ the port
VI =12
condition 12 = 0, applying an 11 current source to the lllput port,
and determining the resulting port voltages. By (16.4) V 2 = (s + 1)12
Thus
(16.6)
Zll = IVII '
1 12 =0 Z12 = 1, z22=s+1
I , "I
~ _________ ~ __ ~ __ ~_J
Examining Fig. 16.5(a), by series-parallel combinations, and the z-parameter two-port description is
Two-portN
VI =41 1
(16.7)
FIGURE 16.4 Two-port circuit for
V2 = 11
Example 16.1.
To find the specific values of the port variables for the given port
Since VI and V2 have been computed under the open-circuit condi-
termination circuits in Fig. 16.6, combine the two-port description
tion 12 = 0, then by (16.6), (16.7) with the two separate i-v laws for the port termination cir-
ZII = 4, Z21 =1 cuits. In the case of Fig. 16.6(a), these are
The remaining two z-parameters are specified by (16.4) to be VI =4-11
V 2 = -!I2 (16.8)
Together, (16.7) and (16.8) comprise four equations in the four port
variables. Eliminating the currents by solving (16.8) for 11, 12 and
substituting into (16.7) gives
s+1
1 ][4-Vl]-2V2
Combining like terms yields
5V 1 +2V2 = 16 (16.9a)
VI + (2s + 3)V2 = 4 (l6.9b)
Two-portN
Two-portN Multiplying the last by 5 and subtracting from the equation above
(a)
(b) it, we have
fiGURE 16.5 Two experiments for determining z-parameters. -(lOs + 15)V2 + 2V2 = -4
-8 FIGURE 16.7 (a) Two-port N satisfying port conditions; (b) port conditions
12 = lOs + 13 (l6.lOb)
violated. N is not being operated as a two-port circuit.
which completes the evaluation of the port variables for the termi-
nation shown in Fig. 16.6(a). For different port termination circuits
[Fig. 16.6(b)], we have the same two-port, so we will still use the
same two-port description (16.7). The i-v laws for the new port ~ 16.1.1.
11 S
12
Find the z-parameters for the two-port shown in part (a) of the
termination circuits are --3>
0
<E-"--- figure.
0
+ + Answer
3s +1
Zll = - - ; ZI2 = Z2I = 1; Z22 = - -
+2
3s
(l6.lla) S s
4 4
VI V2 16.1.2. Find the i-v laws for the one-ports shown in parts (b) and (c) of
1 the figure.
V 2 =- (l6.11b) -s 2 -1 2
S Answer VI = - - 1 1 + - - ; V2 = -12 + -
2s+1 2s+1 s s
Using (16.11) to eliminate VI and V 2 from (16.7) and solving as (a) 16.1.3. Find 11 and 12 if the two-port of part (a) of the figure is terminated
above yields the port variables and completes the example. by the one-ports shown in parts (b) and (c).
1 Answer
12s - 5 1 ~
2 11
VI - -,---- V2 =- ---'?- 11 = 2s(s + 2)
- s2 + 5s + 3' S
+ + 19s 3 + 35s 2 + 18s
+ 3'
2S2 + 2s - 1 -s+4 1.. 12 = 2(6s 2 + 5s + 1)
11 = , 12 = --::----- VI V2
S(S2 + 5s + 3) s(s2 + 5s + 3) S
(s + 1)(19s 3 + 35s 2 + 18s + 3)
The example illustrates that the two-port description (16.5) may be used to deter- (b) (e)
mine the port variables for any pair of one-port termination circuits. We say that it is
a complete characterization of the input-output or port behavior of the two-port circuit. EXERCISE 16.1.1 (a) Two-port; (b) input port
termination circuit; (c) output port termination
Note that this characterization is not only complete but very concise. Regardless of the circuit.
complexity of the two-port, its count of meshes, nodes, and elements, the two-port de-
scription summarizes the port behavior in just two equations containing four parameters.
These are the z-parameters in the impedance two-port description (16.5), but may be
defined in various other ways, as we shall see in Section 16.2.
If the port condition is violated, the assumptions under which the two-port descrip-
tion was derived are not valid and equations such as (16.4) are not valid. We must insist
that the current that flows into the upper terminal of each port also flow out the lower in ~election of the voltages as dependent variables in the two-port equations led to the
order that the circuit be operating as a two-port circuit. As long as each port termination Impedance two-port description
circuit is a one-port, we are safe. An example of a four-terminal circuit that does not
behave as a two-port is shown in Fig. 16.7(b).
(16.12)
v2 =1
Since each parameter Yij in this equation has units of siemens, we will call this the
admittance, or y-parameter, two-port description. Two-port Two-port
The manner in which the y-parameters may be determined for a specific circuit is (a) (b)
very similar to the manner in which the z-parameters were computed in Section 16.1. Two
FIGURE 16.8 (a) Two-port with V1 1, V2 = 0; (b) with V1 = 0,
experiments are performed in which first one, then the other, independent port variable is
V2 = 1.
set to zero. In the present case the independent variables are the port voltages, and setting
a port voltage to zero is equivalent to short-circuiting that port. The other independent and, by current division,
port variable is considered to be a source (voltage source in the present case). In each
experiment both dependent variables, port currents in this case, are determined. The -4I (
1 3 -1
y-parameters in the first column of the admittance matrix in (16.12) then follow from the 12 = - 1111= --)(-)=-=Y21
4+2 3 4 4
results of the first experiment, Reversing these port termination conditions as shown in
(16.13) Fig. 16.8(b), by KCL, we obtain
II = 2(1) - ~ = ~ = Y12
and the second column from the results of the second experiment,
12 = (1 + D(1) - 2 = - ~ = Y22
Y22= - 121 (16.14)
Thus the y-parameter two-port description is
V 2 v1 =o
The calculation of these parameters may be simplified even a bit further. In each
experiment there is a single independent source, the voltage source V I in the first exper- D~J = U~~
iment (16.13) and V2 in the second (16.14). Suppose that its s-domain value is set to _
3
4
[
unity. Then by (16.13), - 41
Y1l = Illvl=1,V2=O' Y21 = 12I vl=I,V2=O
The first column of the y matrix is the value of the port currents with the output port short- There are six ways in which two variables can be selected from a set of four
circuited and a unit voltage source V I (s) = 1 put across the input port. Then shorting and si~ corresponding two-port descriptions. Along with the impedance and admittanc~
the input port and setting a unit source in the output port, by (16.14), forms mtroduced thus far, they are the hybrid, inverse hybrid transmission and inverse
Y12 = Ill v l=o,V2=1' Y22 = 12Ivl=o,v2=1
transmission two-port descriptions. The definitions are shown' in Table 16. I.
1?~ simil~ty of form among the various two-port descriptions guarantees that
The second column of the y matrix is the value of the port currents with the input port deterrnmmg theIr parameter matrices proceeds similarly. The steps are summarized as
short-circuited and a unit voltage source V2(S) = 1 put across the output port. follows:
To find the y-parameters for the given two-port circuit, first short- 1. ~x?eriment 1: Set the second independent variable to zero. Open circuit its port if
circuit the output port and put a unit voltage source in the input port ~t IS a. current. and short its port if a voltage. Set the first independent variable to 1,
as in Fig. 16.8(a). The controlled source carries no current in this Insertmg a umty voltage source in its port if it is a voltage; otherwise, insert a unity
case, and the conductances & S and ~ S are in parallel, current source.
2. Determine the two dependent variables. This is the first column of the parameter
II = (& + D(1) = ~ = Yll matrix.
-1 -!s-2
[V1]=[Zll
V2 Z21 Z12][I1]
Z22 12 [12I1]=[Yll
Y21 Y12][V1]
Y22 V2 12 = 1 + [s/(s +4)](1) = s+2
la. Impedance (z-parameter) lb. Admittance (y-parameter)
V 1 and 12 form the first column of the hybrid matrix (entry 2a in
[VI]
12 =[hll
h21 h12]
h22 [11]
V2 11]2 =[gll
[V g21 g12]
g22 [VI]
12 Table 16.1). To get the second column, we open port 1 and put a
unity voltage source across port 2, as shown in Fig. 16.9(b). Since
2a. Hybrid 2b. Inverse hybrid V = I = 0 across the parallel RL,
[VI]
11 =[tllt21 t12][
t22 -12V2] [ -12V2] =[SllS21 S12]
S22 [VI]
11 I _ V2 s
3a. Transmission (ABeD) 3b. Inverse transmission 2 - 2 + (4/s) 2s +4
VI = (1 +~)s
12 = _s_+_4_
2(s + 2)
3. Experiment 2: As in experiment 1, but set the first independent variable to zero and
the second to unity. Thus the hybrid matrix is
4. Solve for the two dependent variables. This is the second column of the parameter
matrix. 5s +4
s+4 ]
h12 ] = [ 2(s + 2) 2(s: 2)
Let us find the hybrid and transmission parameters for the two-port h22 -(s + 4)
,Example 16.3 circuit shown in Fig. 16.9. The independent variables in the hybrid 2(s + 2) 2(s + 2)
description are 11 and V2; the dependent variables are VI and h. The
experiments for finding the hybrid description are shown in Fig. 16.9. Turning to the transmission parameters, in Fig. 16.1O(a) we
The impedance Zin looking into the circuit of experiment 1 is have set -12 = 0 and V 2 = 1. By KVL around the right loop,
Va = 1 and
2s 1[1 + (4/s)] _ ~s + 2
Zin = S + 2 + 1 + 1 + (4/s) - s + 2 I - I _ Va = s
~s - 2
1- a - a + (4/s) s +4
VI = ZinI I = _2_ _ 2
s+2 V1 = ( -2s- + 1 + -4) II = -3s-+-6s-+ -
8
s+2 s (s +2)(s +4)
v2 =o -1
FIGURE 16.9 (a) First experiment for determining hybrid parameters; FIGURE 16.10 (a) First experiment for determining transmission
(b) second experiment. parameters; (b) second experiment.
Va s
To: Y
Ia=----
1 + (4/s) s+4 Z [Zl1 Z12 ]
[ Y22
-Y21
-Y12 ]
Y22 [ -~21 hi 2 ]
[g~1 -~12 ] [ t~1 t~2 ]
[ -S22
-8
-1 ]
-Sl1
Z21 Z22 Y h22 gl1 t21 S21
2s +4
II =Ia +1 = - -
2+4 [ Z22
-Z21
-Z12 ]
Zl1
Y12 ]
[h~1 -~12 ]
[ -~21 gi 2 ] [ t22
-1
-T]
t11
[ -~11
-!22]
[Y11
2s 5s + 4 Y
Z Y21 Y22 h11 g22 t12 S12
VI = - - I I +Va = - -
-~21
s+2 s+4
h
[
zi 2 ]
[Y;2
-?12 ]
[h11 h12 ]
[ g22
-g21
-g12 ]
g22
[t12
-1 t~ ] [ -S12
-8 -;2J
and the transmission, or ABeD, parameter matrix is \ Z22 Y11 h21 h22 G t22 Sl1
3S2 + 6s + 8 [Z;1
-~12 ]
[ -~21 Yi 2 ] [ h22
-h21
-h12 ]
h11
g12 ]
[ tIl -T] t12
[ -~21 -1 ]
-S12
g [gl1
5s +4] H
(s + 2)(s + 4)
Zl1 Y22 g21 g22 t11 S22
tl1 t12 ] = [ s+4
[
t21 t22 s
s+2
2s +4
s+4
[ zi 1
Z~2] [ -Y22
-y
-1 ]
-Y11
[-H
-h22
-h11 ]
-1 [g~1 g3]
[t11 t12 ]
[ S22
-S21
-S12 ]
S22
Z21 Y21 h21 g21 t21 t22 8
which completes the example. [ Z22
-1
-Z]
Zl1
[ -?11
Y;2 ] [ -~22 -~11 ] [-G
-gl1
g22 ]
-1
[ t22
-t21
-t12 ]
t11
S
Z12 Y12 h12 g12 T [Sl1
S21
S12 ]
S22
follows from the observation that the dependent and independent variables are reversed conversion factors, we obtain
in the two representations across a row. If MI and M2 are two square matrices such that
for all vectors A and B we have MIA = B and M2B = A, as we do across a row of [ 3s
2 + 83sJ
h12 ] = s 1 [ ~s2 + 4
Table 16.1, then inverting the second, B = M 2 IA, and from the first, MIA = M2IA; h22 2 ~
that is, the matrices are inverses of one another. The impedance and admittance matrices
are a pair of matrix inverses, as are the hybrid and inverse hybrid matrices and the The derivation of each of these conversion rules consists simply of solving the orig-
transmission and inverse transmission matrices (which explains the use of inverse in inal two-port description for the dependent variables of the description we are converting
their names). to. For instance, in converting from impedance to hybrid, solve the second impedance
Rules for conversion between all pairs of two-port descriptions are gathered in equation (16.4b) for 12 , substitute into the first (16.4a), and gather terms. This yields VI
Table 16.2. Each entry shows the conversion in terms of a matrix divided by a scalar and 12 as the dependent variables and II and V2 as the independent variables, which is
that is common to the four elements of the matrix. Examples of the use of this table the desired hybrid description. The formulas in the corresponding Table 16.2 entry are
follow. the hybrid parameters in terms of the original (z) parameters.
[ tll t12] =[
3s2 + 6s + 8
(s + 2)(s + 4)
5s +4]
s+4
variable, and computing the dependent variables in the resulting circuit. Each experiment
produces one column of the parameter matrix. The six descriptions are related by the
conversion factors listed in Table 16.2.
t21 h2 s 2s +4
s+4 s+4
Applying the transmission to the hybrid conversion cell from Ta-
ble 16.2 with these transmission parameters and
T _ 3s 2 + 6s + 8 2s + 4 _ 5s + 4 _s_ = 1 16.2.1. Find the y-parameters. Which, if any, of the two-port descriptions
- (s+2)(s+4) s+4 s+4 s+4 fail to exist for this circuit?
Answer Yll = Y22 = 1, Y12 = 0, Y21 = -1; inverse transmission
yields does not exist
16.2.2. Determine all six of the two-port descriptions.
,~J
[5' +4 Answer
s+4
-1 [ 5>+4 ,+4 ] z = ~ [3S + 4 S + 4 J . Y _ _1_ [ 3s + 4 -(S+4)J'
h12] =
= 2(s + 2) 2(S: 2) Two-port 4s s+4 3s + 4 ' - 2s + 4 -(s + 4) 3s +4 '
[hll
-(s + 4)
h21 h22 (2s + 4)/(s + 4)
2(s + 2) 2s +4
EXERCISE 16.2.1 h=_I_[ 2s+4
3s + 4 -(s + 4)
S+4J.
4s
__ 1_[ 4s -(s +4)J.
' g - 3s + 4 s + 4 2s +4 '
2
which agrees with the hybrid matrix calculated independently in t=_I_[3S+4 2S+4J.s _ _l_[3S+4 -2(S+2)J
Example 16.3. S+ 4 4s 3s + 4 ' - s + 4 -4s 3s + 4
16.2.3. A bilateral circuit has Z12 = Z21. What equalities does a bilateral
+ + circuit satisfy in terms of the other two-port parameters?
Being able to convert from one to another shows vividly that t~er~ is much r~dun
dancy in the information conveyed by the six separate two-port descnptIons. What ~s ~he Answer Y12 = Y21, h12 = -h21' g12 = -g21; S = T = 1 (S and
1
~~
T are the determinants of the s and t matrices as in Table 16.2)
point of having six when one will do? One reason is that not every two-port descrzptlOn
exists for every two-port circuit. Examining Table 16.2, note that to convert betwee~ :_l________][L-_S______
descriptions, division by a parameter is required. If that parameter happens to be z~r~, It
is not possible to perform the conversion because the corresponding two-port descnptIOn EXERCISE 16.2.2
does not exist.
A second practical reason for having several different two-port descriptions is that
it is sometimes easier to determine one set of two-port parameters than another. For
instance, if the two-port description of a transistor circuit is to be determined by bench
measurement, the four parameters comprising the hybrid description are easily measurable
for this type of circuit. Others, requiring experiments with an open-circuited output port,
are more difficult to arrange without driving the circuit out of linear operation. Or we
may be calculating the two-port description of a circuit using mesh analysis, and. the Thus far in our discussion of two-port circuits we have focused on the uses and the
open-circuited port conditions of the impedance description may lead to fewer equatIOns procedures for converting a circuit diagram to a set of four parameters, the process of
than the short-circuited admittance description. two-port analysis. The reverse process, in which we find a circuit that agrees with a
Finally, when two-ports are interconnected, it is far easier to determine the overall given set of two-port parameters, is called two-port synthesis or modeling.
two-port description of the interconnected circuit if we are using the right t~o-port param- Two-port models are useful in several contexts. They permit us to visualize a circuit
eters. Which is right depends on the manner in which the two-ports are mterconnected, described by a set of two-port parameters, to consider the effects of interconnecting them
as we shall see in Section 16.3. with other elements or subcircuits, and to use SPICE to analyze them. In the case of
To summarize, in this section we have defined the six two-port descriptions: impe- coupled coils, they permit us to model a circuit with mutual inductance by one that does
dance, admittance, hybrid and inverse hybrid, and transmission and inverse transmission. not contain mutual inductance.
Ij 12
----'?>- ~
+
+
V2
Vj V2 Y12 V 2 Y21 Vj V2
(a) (b)
(al (b)
FIGURE! 16.11 Two-port circuit models: (a) z-parameter; (b) hybrid
parameter. FIGURE 16.12 (a) Three-terminal two-port; (b) model.
v'2 VI
l~ 3 V2
a'
(a) (b)
FIGURE 16.13 Circuit model for three-terminal two-port.
FIGURE 16.14 (a) Circuit for Example 16.6; (b) two-port model with
no internal loops.
easily verified that this circuit obeys the two-port description \
z-matrix is
+ Z1212
VI = Zllll (l6.19a) 3s + 11
V; = Z1211 + Z2212 (l6.19b) Z12 ] = [ S +4 s+43 ]
Z22 3 3s 2 +7s + 16
Next observe from the figure that V; = V2 + (Z12 - Z21)11. Combining this with (16.19), s+4 s(s + 4)
we are back with the general form (16.15). Thus Fig. 16.13 is a general circuit model for Comparing this with Fig. 16.13, a two-port model for this three-
a three-terminal two-port with impedance parameters Zll, Z12, Z21, and Z22. In contrast terminal bilateral circuit is shown in Fig. 16. 14(b) with
to the previous models, it contains a single controlled source.
3s + 8
In the important special case Z12 = Z21, even this one remaining controlled source ZA = Zll - ZI2 = - -
disappears, resulting in an impedance-only model. Two-port circuits for which ZI2 = Z21 s+4
or, equivalently, in terms of y parameters YI2 = Y21, are called bilateral. It can be shown 3s 2 +4s + 16
that all circuits containing only impedances are bilateral and hence can be modeled by ZB = Z22 - Z12 = s(s +4)
three impedances in the Y configuration shown in Fig. 16.13. 3
ZC =Z12 = - -
s+4
Let us find a compact two-port model for the circuit of Fig. 16.14(a), Any pair of coupled coils forms a bilateral two-port circuit, as can be seen from
a model with no internal loops. Begin by finding the z-parameters (16.17). If the two coils can be tied together to form a three-terminal two-port circuit,
for this circuit. Open-circuiting the output and placing a unit current the impedance-only model of Fig. 16.13 with Z12 = Z21 is a very useful circuit model for
source in the input port, the loop equation is the coupled coils. The model, previously introduced in Fig. 15.11(b), replaces the pair
of coupled coils, with their magnetic coupling, by three simple (uncoupled) coils. Note
l(la - 1) + (s + 3)la = 0 that even if the pair of coupled coils does not have a common reference node, as long
whence la = l/(s + 4). as there is no external electrical connection between them, a shorting wire can always be
installed between the coils without affecting any element's current or voltage drop. This
s +3 3s + 11 is due to the generalized KCL, which guarantees that a single wire between otherwise
VI = 211 + 1(1 - la) = 2 + -- = ---
s+4 s+4 electrically isolated subcircuits carries no current.
and
Find an equivalent circuit for Fig. 16.15(a) that has no coupled coils
3 or controlled sources. The i-v laws for the coupled coils are
V 2 = 31a = - -
s+4
VI = 2sl 1 - 2s12 (l6.20a)
VI and V2 form the first column of the z-matrix. Next, open-
V 2 = -2sll + 3s12 (l6.20b)
circuiting the input and putting the unit current source on the output
yields Connecting a and a', we may use the three-terminal model shown
in Fig. 16.13, which, for the bilateral case Z12 = Z2lo is impedance-
3(la + 1) + (s + l)la = 0
only. With Zll - Z12 = 4s, Z22 - ZI2 = 5s, and Z12 = -2s, the result
Solving for la and then V I and V2 as in the first experiment, the is shown in Fig. 16.15(b).
Two-port Two-port
(a) (b)
transformation may be used to convert these models to equivalent models having the same
number of elements, but fewer loops (Y configuration) or fewer nodes (Ll configuration).
The Y-Ll and Ll-Y conversion formulas are given in Section 11.5 as (11.31) and (11.32).
~v
l
A certain three-terminal two-port circuit has z-parameters, 4s + 1 2
2(4s + 1)
Z12 [ 4s + 3 +3
4s4s (b)
Z22] = -2(2s + 3) 4s(s + 2)
EXERCISE 16.3.1
4s + 3 4s +3
Following Fig. 16.13, a circuit model for this circuit is shown in 16.3.2. What are the requirements for a three-terminal two-port to have
Fig. 16.16(a). Using the Y-Ll conversion formulas (11.31) with a circuit model containing only impedances? State in terms of the various
two-port parameters.
4s + 3 Yb = 4s + 3 4s +3 Answer Z12 = Z21, Y12 = Y21, h12 = -h21, g12 = -g21 , S =
Y -.,---,--- Yc= - - -
a - 2(2s + 1) 4s(s + 1) 4s T = 1 (S and T are the determinants of the sand t matrices as in Table 16.2)
we arrive at the equivalent two-port model shown in Fig. 16.16(b). 16.3.3. Apply the model of Fig. 16.13 to a three-terminal pair of coupled
coils. Compare the result to Fig. 15.11(b).
In both models, the values inside the boxes are impedances.
Answer They are identical.
We began our study of electrical circuits in Chapters 1 and 2 by considering the properties
of basic one-port (two-terminal) subcircuits: resistors, sources, series-parallel combina-
tions, and so on. The power and versatility of these simple one-ports was not obvious
in their limited behaviors as individual one-ports but became evident when two or more
(a) (b)
were interconnected. From RC pairs emerged low-pass and high-pass filters, and from
FIGURE 16.16 (a) Two-port model for Example 16.8; (b) equivalent RLC combinations we derived resonance, interconnections of op amp building blocks
two-port model using Y-Ll transformation. yielded circuits of simultaneously high gain and bandwidth, and so on.
668 Chapter 16 Two-Port Circuits Section 16.4 Interconnection of Two-Port Circuits 669
What is true of one-ports is also true of two-port circuits: Their real usefulness Now (16.21) is in the form of a transmission two-port description
emerges only with their interconnection. Indeed, our main motivation for studying two-
ports in the first place was so that we could implement an approach to circuit design based t 12
t22 J[-14V J
4 (16.22)
on their interconnection. We proposed to solve complex design problems by breaking
of the overall interconnected circuit, where the tij' s are the overall transmission parame-
them up into a set of narrowly defined input-output tasks, each solved by a manageably
ters for the cascaded pair of of circuits. Comparing (16.21) and (16.22), it must be the
simple two-port circuit. The solution to the overall problem lay in the interconnection of
case that
these two-ports.
In this section we consider several practical ways to interconnect two-ports. Our CASCADE
emphasis will be on finding the two-port description of the overall circuit from the de-
scriptions of its interconnected two-port components. Two two-port circuits are said to (16.23)
be connected in cascade if the output port of one is tied directly to the input port of
the other, as shown in Fig. 16.17. Together they form a single overall two-port whose
input port is A's input port, and whose output port coincides with B's. We refer to A The overall transmission matrix for a cascaded pair of two-ports is the product of their
as the input two-port and B as the output two-port of this cascade interconnection. The transmission matrices, with the input two-port premultiplying.
individual two-ports A and B have transmission parameter matrices, which we notate as To find the overall transmission description, we need only mUltiply the individual
transmission matrices together in the natural order (input two-port matrix first). Recall
A: B: that matrices do not in general commute; that is, the matrix product AB is not in general
equal to BA, so order matters.
The simplicity with which the overall transmission parameters of the cascaded in-
Examining the port at which they are interconnected, clearly V3 terconnection can be derived from the individual transmission parameters does not carry
12 , or over to other two-port descriptions. For instance, there is no simple relationship between
the impedance parameters of two circuits and the impedance parameters of their cascade
interconnection. For this reason it is useful to convert to transmission parameters before
computing the two-port description of a cascade.
(16.21)
B:
[ -Ha -h~1 J
[ t~1 t~2 J = -hZ2 -1
tZI tZ2 hZl
670 Chapter 16 Two-Port Circuits Section 16.4 Interconnection of Two-Port Circuits 671
The overall transmission matrix is their product, with the input two-
port A's transmission matrix written first (premultiplying).
+
S
VI v2
Carrying out the matrix multiplication, the transmission parameters
of the cascade of A and B are
t11 = s(4s + 5) + l(s + 1) = 4s + 6s + 1 2
Overall two-port
s+3 1 4s+13
h2 = --(4)
s
+ -(1) = - -
s s FIGURE 16.18 Parallel interconnection of two-ports.
Table 16.1 cites the conversion from transmission to hybrid param-
then, using the admittance matrices for the two circuits,
eters as
1
12 J = [Y!l
[I
Y21 Y!2J
Y22 [V!J
V2 + [Y!l
Y21 Y!2J
Y22 [V!J
V2 (16.24)
But the figure also makes clear that the voltages are equal:
where T = t11 t22 - t12t21' The final result is (16.25)
4S +1 -3 ]
[ And combining the last two equations yields
-1
4s 2 + t8s + 16
(4s + 13)/s
s
D~J = {[~:
This is an equation of the form
s(4s + 1)
_ [ 4s + 13
-s
4s -3
+ 13 ]
4s 2 + 18s + 16 D~J = [;~: ;~~J [~~J
4s + 13 4s + 13 so it must be the admittance (y-parameter) description of the parallel combination.
672 Chapter 16 Two-Port Circuits Section 16.4 Interconnection of Two-Port Circuits 673
---------------------------1
~
Ia Ia 12 We wish to know the transfer function H(s) = V2(S)/V 1 (s)II2 (s)=O,
4 ~ <E-"-
, called the open-circuit voltage transfer ratio, for the circuit in
+ + + ,, + Fig. 16.20. Since any two-port description is a complete characteri-
Two-port
Vf A
V z In In zation of the port behavior of a circuit, a transfer function involving
only port variables such as this H(s) should be easily computable
~ IL
<E-"-
once a two-port description is known. Examining Fig. 16.20, the
+ + overall circuit consist of two T-shaped two-ports in parallel, all in
VI
1F 1n 1F 1F
V2 4 2 2 4 series with a two-port consisting of just a capacitor, as shown in
Ib Ib
4 ~ VI VL Fig. 16.20. Since parallel admittances add, let us begin by finding
the admittance matrices for Nl and N 2 . For Nl we first set VI = 1
+
vf Two~po~
+
Vb2
TIF and V2 = O. Then
,,, B 0
\
0 1 s+2
- 11 = - - - - - - -
,, FIGURE 16.20 Circuit for Example 16.10. 1 + (2/s)/(1 + 2/s) s+4
- - - - - - - - - - - - - - - - - - - - - - - _____ 1
-I _ s/4 s(s + 8)
2- (s/4)+28(s+4) 8(s + 4)
When substituted into (16.27), this is
N2 has the same symmetry as NI did, so
[
Yll Y12] 1 [s2 + 8s (16.29)
Y21 Y22 = 8(s + 4) -s2
[~ ~ J
Y12J- 1
Yzz
[ Zll
ZZI
ZIZ J =
ZZZ
1 1
= 8(s + 4) [SZ + 16s + 16 _(sz + 16) J- 1 s s
-(sz + 16) sZ + 16s + 16
Adding the z-parameters for the series two-ports N12 and N 3 , the
where we have used the fact that if P = a Q, where P and Q z-parameters for the overall circuit are
are square matrices and a is a scalar, then p- 1 = (l/a) Q-l. The Zll
determinant is [ ZIZJ
ZZI ZZZ
(SZ + 16s + 16)z - (sz + 16)z = 32s 3 + 256s z + 512s (16.31)
(16.30) 1 [sZ + 20s + 32 s2 + 4s + 32 J
= 32s(s + 4)z = 4s(s + 4) s2 + 4s + 32 s2 + 20s + 32
and We stated earlier that any transfer function, such as V 2!V 111z=0 re-
quired here, can easily be determined from a two-port description.
Zll ZIZJ 1 [SZ+16S+16 sZ+16 J The z-parameters satisfy
[
ZZI ZZZ = 4s(s + 4) sZ + 16 SZ + 16s + 16
VI = Zll l i + z12lz
We need the z-matrix for N3 as well. Inspection of Fig. 16.21(b)
shows that with a unit current source in either port and the other
Vz = Z2111 + Z2212
Setting 12 = 0 and dividing the second equation by the first gives
11 12
r---------------------- ~ *-"-- H(s) = V 2 (s) I Z21 (16.32)
11 12 + + V 1 (s) 12 (s)=0 Zll
~ *-"--
and in light of (16.31),
r
+ + Z
N 1: VI V2
sZ +4s + 32
H(s) - -:::---::-c:-----::-::-
- s2 + 20s + 32
0 0
which completes the example.
11 12
VI V2 ~ *-"--0 A final cautionary note on the use of these results is required. We saw earlier that
0
+ + if we are not careful, it is possible to terminate a four-terminal circuit in such a way
.s .s that the port condition is violaterl and the circuit is not being operated as a two-port
N 2: VI V2 circuit. If the port condition fails to be true, our basic two-port representations and all
results that flow from them fail to apply. It is also possible, even likely, that certain
interconnections between two-ports will cause port condition violations for one or more
of the interconnected two-ports. To use the results of this chapter, a port condition
violation must be prevented.
11 12
0
~ *-"--0 One way to guarantee that the port condition is satisfied everywhere is to introduce
N3:
+
VI
11TS +
V2
a 1: 1 ideal isolation transformer at either one of the ports of each two-port being inter-
connected, as shown in Fig. l6.22(a). Since the primary is just a solid length of wire
connecting upper and lower terminals, the port condition will clearly be satisfied there.
And since it is satisfied there, by KCL applied to the two-port as a whole, it is also satisfied
at the other port as well. If, say, five two-ports are placed in parallel, four transformers
0 0
(a)
will be sufficient, since the fifth two-port must satisfy the port condition if all the rest do.
(b)
If we do not wish to add isolation transformers, we must take care that the port
FIGURE 16.21 (a) Circuit of Fig. 16.20 as the interconnection of three conditions are satisfied at the overall input and output port of the interconnection, that
two-ports; (b) Njf N2f N3. is, no cross connections as in Fig. 16.22(b). Even so, the individual circuits making up
676 Chapter 16 Two-Port Circuits Section 16.4 Interconnection of Two-Port Circuits 677
,
~--------------------------I,
I 1 1 I then do we truly have an interconnection of two-port circuits as they have been defined
~: ?
and studied. If we are not working with three-terminal two-ports, this may necessitate
+ +
adding a 1: 1 isolation transformer.
-<E-- --7
~ ,--------------------------. 12 EXERCISES
(a)
12 0 16.4.1. Find the t matrix and determine the voltage across the input port
1---------------------------------------
, 11 ~---------------'
-----:;;"1 I~
12 if four of these circuits are put in cascade, the output is open-circuited, and
+ +
a +2-A dc source placed in the input port.
240 120 V2
Answer tll = 2, t12 = 12, t21 = t22 = ~; 20.2 V !,
16.4.2. Show that when the two-ports Ca) and (b) are put in parallel and a
unit current source is put in the input with shorted output as shown in (c),
+
the port condition is violated for both two-ports.
+
Answer For port condition to be satisfied, need lal = I~l' la2 =
EXERCISE 16.4.1
1~2' Ibl = I~I' Ib2 = 1~2' But lal = la2 = 1~2 = I~I = 0, while
Ib1 = I~I = 1, Ib2 = 1~2 = -1, so none of the port conditions is satisfied.
,
---------------------------------------- 0 - - -I -_; ------;- --?
0
Two-port Two-port ~ _ _ _ _ _ _ _ _ .J
~a2
the interconnection may be in violation of the port condition. Only in cascade is the port
condition of each circuit satisfied as long as the overall circuit's port condition is satisfied.
For series-parallel circuits there is a useful special case. The port condition is always triv-
ially satisfied for two-ports in parallel or in series if each two-port is of the three-terminal
type, as shown in Fig. 16.22(b) and (c), and the common terminals of both circuits are
tied together as shown. For instance, in Fig. 16.22(c), since Ia + Ib = II + 12 and Ia 1
I~ bl " 1
- - :.. - \,,'. _<./. _.' - - - _. - - - - - - - - - -
and Ib flow through parallel shorts, we may assign Ia = II and Ib = II. thus satisfying
I
678 Chapter 16 Two-Port Circuits Section 16.4 Interconnection of Two-Port Circuits 679
A portion of the output follows:
680 Chapter 16 Two-Port Circuits Section 16.5 SPICE and Two-Port Circuits 681
A portion of the output follows:
SPICE may be used to determine two-port descriptions, but only at specific frequencies.
Replacing s by j w in any of the two-port descriptions results in phasor-domain equations, FREQ VR(l) VI (1) VR(5) VIeS)
which can be evaluated using SPICE in .AC mode.
1.00E+00 -3.94E-01 5.44E-01 -1.05E-01 9.18E-02
1.00E+01 -2.52E-01 6.35E-02 -2.48E-01 1.28E-04
Let us check the results of Example 16.10, in which the z-parameters
for the circuit of Fig. 16.20 were found to be
These values correspond to the first column of the z-matrix for w =
[
Zll Z12 ] 1 [s2 + 20s + 32 s2 + 4s + 32 ] 2n: and w = 20n:. For experiment 2 the input file is amended by
Z21 Z22 = 4s(s + 4) s2 + 4s + 32 s2 + 20s + 32
editing the II source statement to reference node 5 rather than node 1,
Replacing s by jw, the z-parameters at the angular frequency ware thus moving the unit source to the output port and the open circuit
to the input port. The corresponding SPICE output is the second
Zll (jw) zdjw) ] column of the z-matrix for these two frequencies.
[
Z21(jW) Z22(jW)
1 [(32 - ( 2) + 20jw
= 4jw(jw + 4) (32 - w2 + 4jw
FREQ VR(1) VI(l) VR(5) VI(5)
1.00E+00 -1.05E-01 9.18E-02 -3.94E-01 5.44E-01
We will use SPICE to check the value of the z-parameters at the 1.00E+01 -2.48E-01 1.28E-04 -2.52E-01 6.35E-02
frequencies of 1 and 10 Hz. Two experiments are required to find
the z-parameters for a two-port circuit. The SPICE input file for
experiment 1, in which a unit current source is located in the input
The values do indeed check. For instance, at f = 1 Hz or w =
port and the output port is open-circuited, is shown next. Note that 2n: rad/s,
100-MQ resistors have been put in parallel with the capacitors to
eliminate floating nodes because SPICE insists on dc paths between
all nodes. Since the impedance of each capacitor is many orders of Zll (j2n:) = (32 - ( 2) + 20jw
" I = -0.394 + jO.0545
magnitude less, the error introduced is negligible. 4Jw(jw + 4) w=2:n:
which agrees with the results of the first experiment for V(l). At
EXAMPLE 16.11 f= 10 Hz or w = 20n: rad/s,
*I1 1 0 A( 1 Z22(j20n:) =
(32 - ( 2) + 20jw
"
I = -0.252 + jO.0635
(1 1 2 0.25 4Jw(jw + 4) w=20:n:
RD1 1 2 100MEG in accord with the second experiment's value for V(5).
R1 2 3 0.50
(2 2 5 025
RD2 2 5 100MEG
(3 3 0 1 SPICE may also be used to determine circuit variables for circuits containing two-
RD3 3 0 100MEG ports. In this case the two-port descriptions must first be converted to circuit models as
R2 described in Section 16.3.
1 4 1
R3 4 5 1
(4 4 3 050
.A( LIN 2 1 10 16.12 Let us determine the Bode gain plot (gain in dB versus log frequency)
.PRINT A( VR(l) VI(1) VR(5) VI< 5) in the range 1 Hz to 1 kHz for the circuit of Fig. 16.23. Replacing
.END the two-port by its hybrid model of Fig. 15.28(b), we have the circuit
of Fig. 16.24.
1~-I]
[10 3 + 1~5 C2 2 3 lOU
-104
R2 3 4 lK
VD 0 4 AC 0
Fl 0 5 VD 10K
FIGURE 16.23 FIGURE 16.24 R3 5 0 10
C3 5 6 looU
RL 6 0 30
Note that a dummy voltage source has been installed so that the AC DEC 10 1 lK
controlling current h can be referenced. The resulting output for .PLOT AC VDB(6)
the following input file is shown in Fig. 16.25. .END
FREQ VDB(6)
1.OOOE+OO -2.140E+Ol 16.5.1. Find the admittance matrix Y for this circuit. Note that there are
1. 259E+OO -1.848E+Ol no storage elements, so Y is independent of frequency.
1.585E+OO -1.579E+Ol
1.995E+OO -1.329E+Ol
2.512E+OO -1.096E+Ol 1Q IQ
3.162E+OO -8.758E+OO
3.981E+OO -6.657E+OO
5.012E+OO -4.645E+OO
6.310E+OO -2.719E+OO *
7.943E+OO -8.949E-Ol
1.OOOE+Ol 7.994E-Ol *
1. 259E+Ol
1.585E+Ol
2.318E+OO
3.601E+OO
~ ~
*
1.995E+Ol 4.579E+OO + Two-port +
2.512E+Ol 5.187E+OO 8Q 3Q
[ Zll Zl2 ]
3.162E+Ol 5.379E+OO Z21 Z22
3.981E+Ol 5.139E+OO * VI 50 Q
EXERCISE 16.5.1
V2
5.012E+Ol 4.488E+OO [4S + 104 104 ]
6.310E+Ol 3.476E+OO + 10
10 4s- 6 0.353 -0.368 ]
10 4
7.943E+Ol 2.172E+OO
-
s
-
Answer y = [ -0.641 1.05
1.OOOE+02 6.472E-Ol
1. 259E+02 -1. 037E+OO
16.5.2. Find the frequency f at which the magnitude of the look-in imped-
* ance is minimized.
1.585E+02 -2.830E+OO
1.995E+02 -4.696E+OO
EXERCISE 16.5.2 Answer 79.2 Hz
2.512E+02 -6.611E+OO
3.162E+02 -8.556E+OO *
3.981E+02 -1.052E+Ol
5.012E+02 -1. 250E+Ol *
6.310E+02 -1. 449E+Ol
7.943E+02 -1. 648E+Ol
1.OOOE+03 -1. 847E+Ol
The current and voltage at any pair of wires, or input port, affect those at a second pair, or
output port, with which it is connected electrically or magnetically. Two-port circuit rep-
FIGURE 16.25 SPICE plot for Example 16.12.
resentations are sets of equations that show the input-output relationships at the two ports.
FIGURE 1'16.1
IH FIGURE 1'16.14
fiGURE 1'16.9
16.2. Find the transmission and inverse transmission pa- fiGURE 1'16.4
16.15. Find the inverse hybrid and inverse transmission
rameters. 16.10. Let hll = 1 kQ, h12 = 10-4 , h21 = 100, and parameters for the circuit of Problem 16.14. (Do not use
16.5. Show that the hybrid parameters may be obtained
3Q from the z-parameters by h22 = 10-4 S in Fig. 16.11 (b) and find the network function Table 16.2.)
V2fVI if port 2 is open-circuited. 16.16. Find all six of the two-port parameter descriptions
+ + ZI2 16.11. Find the z-parameters for this two-port circuit. for the circuit of Problem 16.11, computing the g-parameter
gll = - , gl2 = - -
Zll Zll descriptions from two experiments and the rest from Ta-
6Q Z21 ~z 2Q ble 16.2.
g21 = - , g22 = -
Zll Zll 16.17. Find the z-parameters, and find the impedance
16.6. Find the h- and g-parameters. looking into the input port with the output port open-
circuited VIlli lIFO and the impedance looking into the
4mF + + output port with the input open-circuited V2!12I i l=O.
IQ
v
V
--7
~
(s)
1 1 nnr'\
vv
I F ! F V2+(S)
~o
[ YII
Y21
Yl2 ]
Y22 =
[ I +~
- f
s
3+l
- ~]
s
FIGURE 1'16.11 ~ T ~r-_TL--_2__ 0
Find the hybrid parameters by using Table 16.2. FIGURE 1'16.6 16.12. Find the z-parameters for this two-port circuit. FIGURE 1'16.17
(a)
of Problem 16.18 in parallel. [
s+-
1000 s l2kll V2(t)
+ +
g 16.34. Find the inverse transmission parameters of three s+ 2~00 s2+_1_s
1000
Iill!I copies of the circuit of Problem 16.18 in series.
and
1
Yll = Y22 = 2(Yb + Y a)
1
YI2 = Y21 = 2(Yb - Y a)
VI Vz
i2
~ 2H IF
+
+
il
Vz
FIGURE P16.44
, ' ,
fiGURE 1'16.41 16.45. Use superposition to show that if a two-port con- ciencYiii' , , ", ""
16.42. Show that for the circuit of Problem 16.4, in terms
tains independent sources, its two-port equations will have Lll<:e nibstF(enobtnerihl~ ~',-'-' "'lc'~""""Il,"!'i(:~:W~~P~l~t'i)'it;t(~E,,oJ~.,'fi"'!,cs,~f,theFreneh'.Rey;; i '
of the transmission parameters we have
an extra source term. For instance, the impedance descrip-
tion is
olutlon anc;l its ~~nri~tll 'toi~singhislife; H~
V2 was one of the first teache(s' in, the', ,','," ", """'."." ,"', ",', ,,', ,"', ", ',P()l.Ytechtriqu~ and later "
VI A+B Z12][II]+[VIS]
Z21 12 V2s became its professor of mathematical analySis. Atage,3o.F~urier was appointed
16.43. The circuit of Problem 16.22 (no termination) is a scientific advisor by Napoleon on an eXl?~diti(jtito Egypt and for four years Was
lattice with series arms both equal to Za and cross arms How would the new term [~~:] be found? secretary of the Institute d'Egypte, the work of which 1,llarked EgyptologY as
a separate discipline. He was prefect ofthedep~ment of Isere from. 1801 to
1814, where he wrote his famollstreatise ()~heatconducti(jn. fie completed a
book on algebraic equations just before his death,in 1.830.
FIGURE 1>16.40
where Y a = IjZa and Yb = IjZb.
16.44. Find the hybrid and inverse hybrid descriptions for
1mil16.41. Find the h-matrix for the two-port. this circuit. Jean Baptiste Joseph Fourier
1768-1830
lOkQ
2Q
2H IF
fiGURE 1>16.44
FIGURE P16.41
dency.ih '< i ' . ' .. i" 4~:nieiria''n'~c's' ."... '.' . "... '.' .
. '. '.' .tea,j:;he(jil1lle,sch<:>(JL.. '
16.45. Use superposition to show that if a two-port con- u
tains independent sources, its two-port equations will have Like mostFrenc~~n"'" oithe FrenehRev-
16.42. Show that for the circuit of Problem 16.4, in terms
.. ,. " '. , .. '.'.' .. .' , tol?smg hi~
an extra source term. For instance, the impedance descrip-
of the transmission parameters we have . " ..... ' life. l:Ie
V2
tion is
and
was one of the. first teacher~)nthe .....................,. . . . ...',I3cQlePplYtcp.rlqe later
VI A+B [ VI] = [Zll Z12][II]+[VIS] became its professor of matherrtati;calanalysis.At age' 30 Fourier was appointed
V2 Z2I Z2I 12 V2s
16.43. The circuit of Problem 16.22 (no termination) is a scientific advisor by Napole{)llon' an expedition to Egypt and for four years was
lattice with series arms both equal to Za and cross arms How would the new term [~~:] be found? secretary of the Institute d'Egypte, the work of which marked Egyptology as
a separate discipline. He was prefect of thed~Partment ofIsere from 1801 to
1814, where he wrote his famous treatise onheatconduction; He completed a
book on algebraic equations just before hisdeafu in 1830.
Chapter Contents
II1II 17.1 Periodic Functions II1II 17.6 The Fourier Transform
As first noted in our study of the properties of sinusoids in Chapter 8, a function f(t) is
l1li 17.2 Trigonometric Fourier Series II1II 17.7 Fourier Transform Properties
said to be periodic if there is a real constant T > 0 such that
l1li 17.3 The Exponential Fourier Series l1li 17.8 SPICE and Fourier Analysis
f(t + T) = f(t) for all -00 < t < +00 (17.1)
l1li 17.4 Response to Periodic Inputs IIlIII Summary
The smallest T satisfying (17.1) is called the period of f(t) and has units of time.
ill 17.5 Discrete Spectra and Phase III Problems Equation (17.1) implies that
Plots
f(t + nT) = f(t) for all -00 < t < +00, all integers n (17.2)
The period T is the smallest amount we may shift the graph of f(t) and have it coincide
with its unshifted form, while (17.2) identifies other values of shift that have the same
effect: 2T, 3T, -T, and so on.
Periodic functions are completely specified by their values over any period, as can
be seen from (17.2). Let
fret) = f(t)[u(t - to) - u(t - to - T)] (17.3)
be a function that may be nonzero only over a finite interval of duration T starting at to.
Then, adding together shifted copies of fret), we define
physicist Jean Baptiste Joseph Fourier (1768-1830), has important implications for circuit f(t) = L fr(t + nT) (17.4)
n=-OQ
analysis. Since linear circuits satisfy the superposition principle, it mean~ that. we can
write down the response to any input if we know the response merely to smusOlds. The
response to sinusoids is the frequency response, which was introduced in Chapter 14. In as the periodic extension of fret). fret) is, in tum, called the generator of its periodic
the clear light of Fourier's observation, the idea of frequency response takes on expanded extension f(t). to and T are the initial time and duration of the generator fret). f(t) is
meaning. It is a tool for predicting the response of a linear circuit to any input whatsoever, clearly periodic, since by (17.4)
not just sinusoidal inputs. The ability to express an arbitrary signal as a superposition of 00
sinusoids also implies that phasors are a far more general tool than we may have suspected
thus far. Upon Fourier's expansion of an arbitrary input into sinusoids, phasor analy~is
f(t + mT) = L fr (t + (n + m)T)
n=-oo
may be applied to each component sinusoid in the circuit. Superposition may then be ~n and with i = n + m,
voked to determine the overall response. Thus in linear circuits phasors may be used qUIte 00
generally to determine outputs driven by nonsinusoidal inputs as well as sinusoidal ones. L fr(t + iT) = f(t)
The form taken by the superposition into sinusoidal components hinges on whether i=-oo
the signal is periodic. For periodic signals, we will get a straightforward sum ~f sinu- While this verifies that f(t) is periodic, note that the duration T of the generator fret)
soids called a Fourier series. Otherwise, the superposition takes the form of an mtegral may not always coincide with the period of f(t), since there may be a smaller number
rather than a sum, and the resulting representation is called a Fourier transform. We first satisfying (17.1) for f (t). For instance, if f (t) is a periodic function of period T and
discuss some of the basic properties of periodic signals and then the process by which the we consider any fret) in (17.3) with duration 2T, then its periodic extension f(t) just
Fourier series for periodic functions may be found. Two forms for the Fourier series are reproduces f(t) itself. The period of f(t) is still T, not the duration 2T of its generator
introduced, the trigonometric and complex exponential forms, which are closely related fret). These ideas are illustrated in Example 17.1.
through the Euler identity. We use this result and the principle of superpositi?n ~o fi.nd the
response of a linear circuit to any periodic signal. Next we tum to nonpenodlc SIgnals, Consider the function f (t) = e- 1tl It is clearly not a periodic
defining the Fourier transform and showing its relation to both nonsinusoidal signals and function. The periodic extensions of two generator functions of
to the Laplace transform introduced in Chapter 12. Some useful properties of the Fourier different duration and initial time, both taken from f(t), are shown
transform are developed, and the chapter concludes with a brief discussion of SPICE and in Fig. 17.1. One period of the periodic extension has been tinted
the Fourier domain. green to aid visualization.
690 Chapter 17 Fourier Series and Transform Section 17.1 Periodic Functions 691
f(t) = e-III is also even, since
-~-)it
h(-t) = KdeC-t) + K2ge(-t)
= KdeCt) + K 2ge(t) = h(t)
(a)
while by similar reasoning any linear combination of odd functions is odd. Thus the
sums and differences of even functions are even and of odd functions are odd. Products
hT(t) =f(t) [u(t) - u(t - 1)] of even and odd functions inherit symmetry properties in the same fashion as products
of even and odd integers do. For fe(t) and ge(t) even,
op )it
implies that
h(t) = fe(t)ge(t)
:~
so h(t) is even. The product of an even function fe(t) and an odd function go(t) is
h(t) = fe(t)go(t)
_~~~tt
so
__L-__L-__L -__L -__L -__
-2 -1 0 +1 +2 +3 h( -t) = fee -t)go( -t) = - fe(t)go(t) = -h(t)
and the product is odd. The product of even functions is even, of odd functions is also
(b) (c)
even, and of an even and an odd function is odd.
FIGURE 17.1 (a) f(t); (b) generator hT(t) and its periodic extension; While arbitrary functions are neither even nor odd, they can be rewritten as the
(c) different generator gT(t) and its periodic extension. sum of two components each possessing symmetry. Any f(t) may be expressed as the
sum of an even function and an odd function. Define
The minimum shift of the function h(t) in Fig. 17.I(b) for
which the graph coincides with itself is clearly T = I, and that of
Fig. 17.1(c) is T = ~. Consider next a generator h~(t) of duration fe(t) = f(t) + f(-t)
3 taken from the function h(t) of Fig. 17.1(b) starting at t = 0: (17.5a)
2
h~(t) = h(t)[u(t) - u(t - 3)] fo(t) = f(t) - f(-t) (l7.5b)
This generator contains exactly three periods of h(t). Shifting by 3 2
and adding as in (17.4), we see that the periodic extension h'(t) of
h~(t) simply replicates h(t) three periods at a time. So h'(t) = h(t), as the even part fe(t) and the odd part fo(t) of f(t). Clearly, fe(t) = fee -t) and
and the period of this function is T = 1, as noted above. This is fo(t) = - foe -t) as required of even and odd functions. Adding (17.5a) and (17.5b) gives
smaller than the duration 3 of the generator h~(t). fe(t) + faCt) = f(t)
verifies that their sum is f(t).
Periodic functions may possess symmetry properties that facilitate their analysis.
A function f( -t) = f(t) for all t is said to be an even function, and f(t) = - f( -t)
Find the even and odd parts of the periodic function f(t) = 2cos(3t+
for all t is called an odd function. Periodic functions need only be tested for evenness
30). Because of the 30 phase shift, this function has neither even
or oddness over the symmetrically located single period - T /2 ::: t < +T /2. Given nor odd symmetry. By (17.5a) its even part is
any t ' = t + nT with -T /2 ::: t < +T /2, then by periodicity f(t ' ) = f(t), and if
f(t) = f(-t), then f(-t ' ) = f(-t - nT) = f(-t) = f(t ' ) and the function is even. fe(t) = ![2 cos(3t + 30) + 2 cos( -3t + 30)]
A similar argument holds for f(t) periodic and odd. = cos(3t + 30) + cos(3t - 30)
The cosine function A cos wt is even since Acosw(-t) = Acoswt, and the sine
function is odd, since Asinw(-t) = -Asinwt. The complex exponential function e jwt U sing the identities for cosine of sum and difference angles,
is composed of the sum of an even real part, cos wt, and an odd imaginary part, sin wt .
If fe(t) and ge(t) are both even functions, the general linear combination of them, = ( Ii cos 3t - ! sin 3t) + ( Ii cos 3t + ! sin 3t)
(17.6)
h(t) = Kde(t) + K 2ge(t) fe(t) = .J3 cos 3t
692 Chapter 17 Fourier Series and Transform Section 17.1 Periodic Functions 693
and, by (17.Sb), Changing the variable of integration from t to -x in the first integral on the far right,
we have
+ 30) - + 30)]
fo(t) = H2cos(3t
TI2
fe(-x)dx + -1 lTI2 fe(t)dt
21TI2
= -
ToT 0
J. (t)dt
e
(17.11)
Thus the average. value of an even periodic function is its average over the half-period
or fo(t) = - sin 3t (17.7)
[0, T12l. If f(t) IS odd, the two integrals in the sum above cancel,
fe(t) and fo(t) are clearly even and odd, respectively. The result
f(t) = fAt) + fo(t) with fe(t) in (17.6) and fo(t) in (17.7) in fact T1 lTI2
-T12
fo(t)dt = -T
-11 0
TI2
fo(-x)dx + -1 lTI2 fo(t)dt
T 0
= 0 (17.12)
follows directly from the identity for the cosine of a sum angle (these
trigonometric identities are tabulated on the inside front cover). The and the average value of an odd periodic function is zero.
even part of a sinusoid is a cosine function, and the odd part is a sine . ~ach?f these averages is of interest in the decomposition of periodic functions into
function. theIr smusOldal components, to which we tum next.
fay = -1
T
l to
to T
+ f(t) dt (17.8)
17.1.1. Show that the general linear combination of odd functions is odd
and that the product of odd functions is even.
17.1.2. Find and sketch the even and odd parts of h(t) shown in
and mean square value Fig.17.I(b).
Ifl~v = -1
T
l to
tO
+T If(t)1 2 dt (17.9)
Answer For 0 < t < I, fe(t) =
!(e- t -et - I ).
!
(e- t + et - I ), fo(t)
where the integrals range over one period T, and to is arbitrary. The mean square value lit) 1 + e-1
-2-
is also called the average power in f(t). The square root of the mean square value is
the rms value of f(t):
frms = Jlfliv = -1
T
l
to
to T
+ If(t)1 2 dt (17.10)
-1
For instance, the average value of the function f (t) shown in Fig. 17.1 (b) is (a)
fay = 11 e- t dt = 1 - e- I 1 - e- 1 lo(t)
2
Ifl;v = 11 e- 2t dt = !(1 - e- 2 )
EXERCISE 17.1.2
The calculation of these averages is simplified somewhat for even or odd functions. For
an even function fe(t), 17.1.3. Is f(t) = sint + coswt periodic for w = ~? If so, what is the
fe av = -1 lTI2
T -T12
fe(t) dt = -1
T
1 -T12
0
fe(t) dt + -1 lTI2 feet) dt
T 0
period of f(t)? Repeat for w = ../i.
Answer Yes; T = 6JT; no
694 Chapter 17 Fourier Series and Transform Section 17.1 Periodic Functions 695
receivers and in power supplies. Its generator is a half-period from a sine wave. The
next, or half-rectified sine wave, is also found in many power supplies, and the square
Our earlier work on ac steady state emphasized the fact that many practical sources are wave in the clock circuits of digital computers. None of these periodic functions is
sinusoidal. By no means, however, are all periodic source functions that are of practical sinusoidal. Here we consider how functions such as these may be expressed as sums of
interest sinusoidal. Consider, for instance, the four periodic functions shown in Fig. 17.2. sinusoids.
The first, or sawtooth, wave is used to sweep a beam of electrons across the face of a As Fourier showed, if a periodic function J (t) with period T satisfies a set of rather
cathode ray tube to generate an image. Its generator is a ramp with initial time to = 0 general conditions, it may be represented by the infinite series of sinusoids
and duration T. The second, fully rectified sine wave, is common in communications ao
J(t) = 2" + al cos wot + a2 cos 2wot + ...
+ b l sin wot + b2 sin 2wot + ...
J(t) \ or, more compactly,
/~""'4T J(t) = ;
a 00
+ I::(ancosnwot+bnsinnwot) (17.13)
n=l
(a)
J(t)
where Wo = 2n IT. This series is called the trigonometric Fourier series, or simply the
Fourier series, of J(t). The an's and bn's are called the Fourier coefficients and depend
brvYv
on J(t).
The Fourier coefficients may be determined rather easily. Let us begin by obtaining
ao, which may be done by integrating both sides of (17.13) over a full period; that is,
10 T
(b)
IT IT ~
iT
o
J(t)dt =
iT 0
a
~
2
dt
+ I:: Jo
00 t + bn sin nwot) dt
AV\ !\ /
J(t) (an cos nWot
n=l 0
Each term in the last summation is an integral of a sinusoid over n of its periods Tn = Tin.
Since the average value of a sinusoid over a period is zero, the summation is zero, and
therefore we have simply
lOT T 3T 2T
ao = ~ t J(t)dt (17.14)
T Jo
2 (c)
2
Next, let us multiply (17.13) through by cos mWot, where m is a nonzero integer,
J(t)
and integrate. This yields
jJ n n n T
(d)
2T 3T
I
4T
loT J(t)cosmwotdt = loT ~ cosmwotdt
+ I:: an
00 iT cos mWot cos nWot d t (17.15)
n=l 0
FIGURE 17.2 (a) Sawtooth wave; (b) full-wave rectified sine wave; (c) half-
00 T
wave rectified sine wave; (d) square wave. One period has been tinted
green.
+ ~ bn 10 cos mWot sin nWot dt
696 Chapter 17 Fourier Series and Transform Section 17.2 Trigonometric Fourier Series 697
Using trigonometric identities found on the front cover, we have any function we are likely to encounter in engineering. Sufficient conditions called the
Dirichlet conditions are:
cos mWot cos nWot = ! (cos(n + m )Wot + cos(n - m )Wot)
cos mWot sin nWot = ! (sin(n + m )wot + sin(n - m )wot) 1. J(t) is piecewise continuous.
2. J(t) has isolated maxima and minima.
Applying these to (17.15), each term in the sum is once again zero, since it is the integral 3. J(t) is absolutely integrable over a period,
of a sinusoid over an integral of periods, except the term where n = m in the first
summation. This term is given by !aT IJ(t)ldt < 00 (17.18)
Recall that piecewise continuity was defined in Section 12.1, and isolated maxima
so that and minima means every finite time interval contains only a finite number of such points.
Note that in the Dirichlet conditions for Laplace transformability in Section 12.1 condition
am = ~ [T J(t)cosmwotdt, m=I,2,3, ... (17.16a) 3 is replaced by J(t) being of exponential order. Functions satisfying the Dirichlet
T 10 conditions for Fourier series (17.18) are equal everywhere to their Fourier series (17.13),
(17.17), except perhaps at points of discontinuity t = r where the series converges to the
Finally, multiplying (17.13) by sin mWot, integrating, and applying the same argument,
we have . average value HJ(r-) + J(r+)] on either side of the discontinuity.
bm = ~ [T J(t) sinmwot dt, m=I,2,3, ... (17.16b) As an example, let us find the Fourier series for the sawtooth wave
T 10 of Fig. 17.3, given by
We note that (17.14) is the special case, m = 0, of (17.16a) (which is why we used J(t) = t, -n: < t < n:
ao/2 instead of ao for the constant term). Also, since J(t) and cosmwot have periods of (17.19)
T and Tim respectively, we may integrate their product over any interval of length T, J(t + 2n:) = J(t)
such as to to to + T, for arbitrary to, and the results will be the same. The same argument Since T = 2n:, we have Wo = 2n: IT = 1. If we choose to = -n:,
applies to J(t) and sinmwot as in (17.16b). Therefore, we may summarize by giving the
Fourier coefficients in the form
the first equation of (17.17) for n = yields
ao = -1
n:
In
-n
tdt =
2 [to+T For n = 1,2,3, ... , we have
an = - J(t)cosnwotdt,
T to
2 [to+T
n = 0,1,2, ...
n = 1,2,3, ...
(17.17) an = -1
n:
In
-n
tcosntdt
bn = - J(t)sinnwotdt,
T to 1
= -2-(cosnt + nt sinnt) In
n n: -n
698 Chapter 17 Fourier Series and Transform Section 17.2 Trigonometric Fourier Series 699
f(t)
The case n =
had to be considered separately because of the
appearance of n 2 in the denominator in the general case. Also, since
ao/2 is the average value of the sawtooth wave over a period, by
inspection of Fig. 17.3 we see that ao = 0.
From our results, the Fourier series for the sawtooth wave is
The fundamental and the second, third, and fifth harmonics are
As another example, suppose that we have the generator function
sketched for one period in Fig. 17.4. If a sufficient number of terms
in (17.20) are taken, the series can be made to approximate f(t) 0, -2 < t < -1
very nearly. For example, the first 10 harmonics are added and the fT(t) = { 6, -l<t< 1
result sketched in Fig. 17.5. 0, l<t< 2
... ...
I I I I
-5 -4 -3 -2 -1 0 1 2 3 4 5
FIGURE 17.4 Four harmonics of (17.20).
FIGURE 17.6 Periodic function.
ao 211
= 4:
-I
6dt 21
+-
4 1
3
Odt =6
an = 0, n=0,1,2, ...
Also, we have (17.22)
41T/2
211
an = -
4 -I
n:rrt
6 cos - dt
2
+- 21
4 1
3 n:rrt
Ocos -
2
dt
bn = -
T
J(t)sinnwotdt,
0
n = 1,2,3, ...
12 . n:rr In either case one entire set of coefficients is zero, and the other set is obtained by taking
=-sm-
n:rr 2 twice the integral over half the period.
In summary, an even Junction has no sine terms, and an oddJunction has no constant
and finally,
or cosine terms in its Fourier series. For instance, the sawtooth function of Example 17.3
211
bn = -
4 -I
n:rrt
6 sin - d t
2
+- 21
4 1
3 n:rrt
sin - d t
2
is odd, and its Fourier series has no cosine terms, while the square wave of Example 17.4
is even and there are no sine terms in its Fourier series.
Now let us see how calculation of the Fourier coefficients an and bn may be bn = 1~,n:rr
n odd
simplified if J(t) has symmetry. In (17.17) we need to integrate the functions
= { ~4,
0< t < 1
J(t) 1<t<2
41T/2
an = - J(t)cosnwotdt, n=0,1,2, ... J(t + 2) = J(t)
T 0 (17.21)
n = 1,2,3, ... A nswer -16 Loo sin(2n - l)Jit
If 2n - 1
n=1
702 Chapter 17 Fourier Series and Transform Section 17.2 Trigonometric Fourier Series 703
17.2.2. Find the Fourier coefficients for
J(t) = {3,
-1,
O<t<l
1<t<4 We may obtain yet another form of the Fourier series by replacing the sine and cosine
functions by their exponential equivalents, using Euler's formula. This form is called the
J(t + 4) = J(t) exponential Fourier series and is extremely useful, particularly in considering frequency
responses, one of the most important applications of Fourier series. In this section we
Answer ao = 0, an = ~ sin mr , bn = ~ (1 - cos mr:) n = consider the exponential Fourier series, and in the following section we investigate the
mr 2 mr 2'
1,2,3, ... frequency responses or frequency spectra.
17.2.3. Find the Fourier series for By the Euler identity,
2,
J(t) ={ \
0,
17.2.4. Find the Fourier series for the sawtooth wave of Example 17.3 If we define a new coefficient Cn by
using symmetry properties.
17.2.5. Find the Fourier series for the half-wave rectified sinusoid with (17.24)
generator
iT (t) = 4 cos 2t [ u (t + i) - u (t - i) ] and substitute for an and bn from (17.17) with to = - T /2, we have
4
1 jT/2 .
Cn =- J(t)e- 1nwot dt (17.25)
T -T/2
-n n n o 7!. n 3n We observe also that for real J(t), an and bn are both real, so c~ (the conjugate of cn) is
2 4 4 2 4
f(t) given by
EXERCISE 17.2.5
an
c* = ---'---
+ jbn
17.2.6. Find the Fourier series for the full-wave rectified sinusoid n 2
= 14sin2tl 1 jT/2
J(t) = - J(t)(cos nWot + j sinnwot) dt
T -T/2
shown in the accompanying figure.
which is evidently with n replaced by -n). That is,
EXERCISE 17.2.6
Answer - 16 (1 + L 1
JT
-
2
oo
---2
1-4n
cos 4nt ) en (c n
an + jbn
n=l C- n =
2
= C: (17.26)
704 Chapter 17 Fourier Series and Transform Section 17.3 The Exponential Fourier Series 705
Finally, let us observe that Also, we have
ao
- = -
1 lTI2
f(t)dt Co = 111
2"
-1
f(t)dt
2 T -T12
which by (17.25) is = --
2
11 -1
0
4dt +-111
2 0
4dt
ao
- =Co (17.27) =0
2
Since Cn = 0 for n even and Cn = 81 jnn for n odd, we may
Summing up, (17.24), (17.26), and (17.27) enable us to write (17.23) in the fonn write the exponential series in the fonn
f(t)
00
= Co + LCnejnwot + LC_ne-jnWQt
n=l
00
n=l
f(t) = ~ f= __
} n n=-oo 2n - 1
I_ ej (2n-1):n:t (17.29)
00 -00
The reader may verify that this result is equivalent to that obtained
= L cnejnwot + L cnejnwot in Example 17.5.
n=O n=-l
We have combined Co with the first summation and replaced the dummy summation index It is straightforward to convert between the trigonometric and exponential Fourier
n by -n in the second summation. The result is more compactly written as series. By (17.24), the exponential series coefficients may be derived from the trigono-
metric by
00
f(t) = L cnejnwot (17.28) and by (17.24) and (17.26), the trigonometric coefficients are
n=-oo
an = C-n + Cn = 2 Re(cn) (17.30a)
bn = C- n - Cn = -2 Im(cn ) (l7.30b)
where Cn is given by (17.25). This version of the Fourier series is called the exponential
Fourier series.
EXERCISES
As an example, let us find the exponential series for the square wave
of Example 17.5 given by 17.3.1. Find the exponential Fourier series for f(t) periodic with period
T =4.
= { ~4,
O<t<1 -1 < t < 1
f(t) l<t<2 f(t) = { 1,
0, 1 < It I < 2
00
with T = 2. We have (Vo = 2n IT = n, and thus by (17.25) 1 1 sin(mr 12) jnwt 12
Answer - +-
L e
Cn = - 111
2 -1
.
f(t)e- Jn7ct dt
2 Jr
n=-oo
n
-1
(-4)e- jmrt dt +~
210
t 4e- jnnt dt called the sampling function, occurs frequently in communication theory. [A
closely related function is the sinc function, defined by sinc x = (sin Jr x) IJr X
4 = Sa(Jrx).] Show that if we define
- -.-[1- (-It]
}nn Co = n-+O
lim Cn
706 Chapter 17 Fourier Series and Transform Section 17.3 The Exponential Fourier Series 707
then the result of Exercise 17.3.1 may be given by 2
=~ f: c;)
~1
f(t) Sa ejmrt/2
n=-oo
17.3.3. Generalize the results of Exercise 17.3.2 by finding the exponential IF
4
4Q Vn
jw
4
series of a train of pulses of width 8 < T, described by
8 8
1, -- < t < -
f(t) = 1 0,
2
8
2
FIGURE 17.7 (a) Circuit for Example 17.7; (b) phasor circuit.
f(t + T) = f(t)
frequency w. The source phasor has been labeled as Vn , the phasor
Answer ~ nf;oo sa( n;8) e j2n:rrt/T associated with the nth term vn(t), which will depend on the fre-
quency w = nwo of that term. The output when this phasor source
17.3.4. Show that if f(t) is even, the exponential Fourier coefficients are is used is labeled In. The impedance of the circuit is
Cn
an
= "2 = T 10
2 rT/2 f(t) cosnwot dt Z = 2+ (4/ jw)4 = 6 + jw2
(4/ jw) + 4 1 + jw
and if f(t) is odd, they are
Then the current In is
Cn
b
= 2~ = -:-
IT/2 f(t)sinnwotdt
2
J JT 0 In = ' ; 2(13 :jj:) Vn (17.31)
Use the results to obtain (17.29).
17.3.5. Find the exponential Fourier series for the function of Exer- In Example 17.4 the Fourier series expansion of the square wave
cise 17.2.3. v(t) of period T = 4 (wo = x /2) was found to be of the pure cosine
Answer 1 + ~ [ - j2 e- j(2n-I)2t]
type since v(t) is an even function.
n=-oo (2n - l)Jr 00
708 Chapter 17 Fourier Series and Transform Section 17.4 Response to Periodic Inputs 709
and to the nth term in the Fourier series of the excitation (n = we have input phasors at each ev = nevo given by
1,2, ... )
I n= 1 + jev I V (17.36a)
2(3+jev) w=(mr)/2 n (17.40)
where, as usual in phasor analysis, the amplitude of the sinusoidal Then the output phasor at ev = nevo is the input scaled by the frequency response evaluated
response is the magnitude of the phasor, and the phase shift is the at that frequency,
angle of the phasor: (17.42)
(17.38)
and the overall sinusoidal output is the sum of the individual sinusoidal output compo-
Equation (17.37b) is the trigonometric Fourier series for the forced nents:
response i (t) in which sine and cosine terms at the same frequency
have been combined. Using (17.36b), we can evaluate the Fourier
coefficients in this combined form, 00
-tan- -
1 nn
+n
eo(nevo) = /H(nevo) + ei(nevo) (17.44)
n=3,7,1l, ...
2 6'
These last remind us that the gain experienced by the component at frequency nevo is the
This completes the example.
magnitude of the frequency response at that frequency, and its phase shift is the angle
of the frequency response. Each component of the Fourier series expansion of the input
The reader may notice a close similarly between this analysis and use of the fre-
has gain and phase shift given by H(jnevo), the frequency response function of the circuit
quency response function H(jev) introduced in Chapter 14. Indeed, the response to
evaluated at the component's frequency nevo.
periodic inputs may be computed directly in terms of H(jev). Recall that the forced
response Vo(ev) to the phasor input Vi(ev) at frequency ev is
We wish to determine the steady-state output vet) when the input i(t)
Vo(ev) = H(jev)Vi(ev) (17.39) is the periodic impulse train shown in Fig. 17.8. The circuit is clearly
Given the trigonometric Fourier series for a general periodic input stable (there are no controlled sources or isolated LC loops), so the
steady-state response exists and can be computed from the frequency
response H(jev). The frequency response is derived from the transfer
function H(s) by replacing s by jev as shown in Chapter 14.
710 Chapter 17 Fourier Series and Transform Section 17.4 Response to Periodic Inputs 711
r:. I
i(t)
411" 411" ,411" 411" 411" so the output phasors are
. . t t t t 1.
-411" -211" 0 211" 411"
t (17.48)
or
(a)
n=-oo
H(jnwo)cnejnwot (17.52)
The phasors for each component of the input are therefore simply
I(nwo) = In = 4LQ, n = 0,1,2, ... (17.47) While this is an appealingly direct route to Vo (t) compared to using the formulas (17.17)
and (17.40) required for the trigonometric Fourier series, in most cases it is desirable
has amplitude
EXERCISES
n = 1,2, ...
17.4.1. The periodic signal J(t) of period T = 2 which equals
cosht [cosht = 1(e+t + e- t )] for -1 < t < +1, is passed through an
ideal low-pass filter H(jw) = u(w + 4) - u(w - 4). Find the forced output and phase
get).
Answer (e 1 - e- 1) [1 + _1-2 cos(Jrt + 1800
1+Jr
)J
(a) -1 -bn -1 bn
17.4.2. The current into a parallel combination of a 4-&1 resistor and a <Pn=tan --=-tan - (17.54)
1-H inductor is the square wave J(t) of Exercise 17.2.3. What is the current an an
1Q into the 1-H inductor [defined to satisfy the passive sign convention with
J(t)]. Express in terms of exponential Fourier series.
+ Answer Wo = 2; Cn = 0 for n even; Cn = 4/([jnJr][2 + jn]) for With Ao = aol2 and An for n = 1, 2, ... as above, the variation of An with n, n = 0,
n odd 1, 2, ... , reveals how the amplitude of the harmonically related frequency components
1F v(t) 17.4.3. Find the steady-state response vet) in the circuit to the periodic in J(t) changes with frequency, and <Pn versus n shows how the phase changes with
impulse train vet) shown. Express as a trigonometric Fourier series. frequency. This same information may be extracted from the exponential Fourier series.
Answer In light of (17.24),
(b)
vet) = Loo [2nJr(1 - n 2 Jr2) sin(nJr/2)
2 2 2 2 2
Jcos mrt (17 .55)
(1 - n Jr ) + n Jr
n=l
EXERCISE 17.4.3 (a) vs(t); (b) circuit for
Exercise 17.4.3. / ~ -1 -bn
~ = tan - - =<Pn (17.56)
an
Thus a plot of ICn I versus n reveals the variation of amplitude with frequency and that of
~, the variation of phase with frequency. The plot of ICn I versus n for all n is called
the discrete amplitude spectrum or line spectrum of J(t) and that of ~ versus n, the
discrete phase spectrum.
n:rT,-Ttlrr-~,
-6-5-4-3-2-1 1 2 3 4 5 6
n
We wish to compare the discrete amplitude and phase spectra of the FIGURE 17.10 (a) Discrete amplitude spectra for input vet) and output
periodic signals vet) and i(t). By the results of Example 17.5, itt); (b) discrete phase spectra.
n~' n odd among its frequency components. For instance, the purpose of rectifying a sinusoid as
vet): Icnl = { (17.61) in Fig. 17 .2(b) may be to produce dc power. The resulting Fourier series exhibits a dc
0, n even component with power, but has power at many other frequencies as well. Or we may
pass a signal through a filter and need to know how much of the power in the output lies
and the phase spectrum is within the passband of the filter and how much in the stopband.
Multiplying J(t) by its conjugate f*(t) in (17.28) and integrating over one period
vet): ~ = -90, n odd (17.62)
yields
With Wo = n the exponential Fourier series for i(t) has coefficients
i (t):
1/1/2
cn = - [-88(t + 1) + 88 (t)]e-jmrtdt = .
4(1- e lmr )
lto
to + T
IJ(t)1 2 dt =
lto+T 00
to
00
n~oom~oo cnc~ej(n-m)uV dt
2 -3/2
Integrating term by term, we have zero for all terms for which n - m is not zero, since
Here we have selected _l as the initial time for the integration we are integrating the complex exponential over an integral number of periods, and both
so that the impulses do n~t coincide with either of the limits of its real part cos(n - m)wot and imaginary part sin(n - m)wot have zero average value.
integration. Then Thus, with the remaining terms n = m having constant integrands,
n=-oo
Icnl2
n=-oo
Icnl2 (1765) I
716 Chapter 17 Fourier Series and Transform Section 17.5 Discrete Spectra and Phase Plots 717
We may identify each term on the right as the power of a single frequency compo-
nent of jet). Equation (17.65) is the Parseval relation, which asserts that the average So, for instance: the dc power (n = 0) in jet) is (1 - e- I )2. Since
the total power 10 jet) is
power in a periodic signal is the sum over all components n oj the power Icn l2 in its
exponential Fourier series components. A plot of Ic n l 2 versus n is called the discrete
power spectrum of the signal jet). It consists of a set of equally spaced spectral lines
that show us how the power in a periodic signal is distributed in the frequency do-
Ijl~v = ~ faT j2(t)dt = 11 e- 2t dt
main. = !(1 - e- 2 )
.: .....
,
.. "' ....
.
.. ' '"
.........
-2 -1 o 2
(a)
FIGURE 17.11 f(t) for Example 17.10. One period has been tinted
green.
With Wo = 2rr, Cn is given by 17.5.1. S~ppose that the waveform J(t) of Example 17.10 is passed
through a hl~h-pass .filter with frequency response H(jw) = jw/(l + jw).
cn =
10
t e-(l+j2nn)t dt = ~
1 + ]2rrn
[1 _ e-(1+ j 2nn)]
Sketch the dIscrete hne spectrum of the output get).
Answer lenl = (1 - e- I )2Jrlnl
1 + (2nn)2
1- e- I
Cn = ----
1+ j2rrn en for get)
. ,, 0.1
,
., , .,
, ,
The amplitude spectrum is given by
.
,, ,
., ,,.
,
,,, ,
,
, 0.05
.. , , . ,
..
'
and the power spectrum by "
,,
"
-~-~~--~2~-_71--!0-~--2~-1--J4~'~'~n
EXERCISE 17.5.1
718 Chapter 17 Fourier Series and Transform Section 16.5 Discrete Spectra and Phase Plots
719
17.5.2. Rewrite the Parseval relation (17.65) in terms of the trigonometric !pet)
Fourier series coefficients. 1 [2 00 ]
Answer Ifl;v = 2: a; + ?; (a~ + b~)
17.5.3. How are the discrete amplitude spectrum ICn I and discrete phase
spectrum en of f(t) related to the discrete amplitude spectrum Ic~ I and
discrete phase spectrum e~ of its derivative df(t)/dt?
Answer Ic~1 = Inlwolcnl; e~ = en +900
(a)
FIGURE 17.13 (a) Function fr(t); (b) its periodic extension fp(t).
The discrete spectra show us that the energy in periodic functions is highly concentrated We ha:e replaced .wo by 27T IT and are using the dummy variable x instead of t in the
in the frequency domain over just a discrete set of frequencies. These functions have no coefficle~t ~xpr~sslOn. Our intention is to let T -+ 00, in which case fp(t) -+ f(t). The
energy whatsoever at the vast majority of frequencies, those that happen not to satisfy energy distribut1?n over frequ.ency in the periodic fp(t) (17.67) will, in the limit, reveal
how the energy m the nonpenodic f(t) is distributed
w = nwo for integer n. If we were to pass a periodic signal through any filter whose
gain at the frequencies nwo was zero, then regardless of its gain elsewhere in between Since the limiting process requires that Wo = 2; IT -+ 0, for emphasis we re lace
27TIT by b..w. Therefore, substihlting (17.68) into (17.67), we have p
these frequencies, nothing would come out at all.
Nonperiodic functions are more even handed, scattering their energy continuously
throughout the frequency domain. The Fourier transform is a function of continuous w,
which reveals how the energy in a nonperiodic function is distributed over the continuous
fp(t) = f:n=-oo
[b..W jT/2 !T(x)e-jxnt.w dX] ejtnt.w
27T - T /2
frequency domain, just as the Fourier series showed how periodic functions distribute their (17.69)
energy only over their discrete harmonic frequencies nwo. = f: [_1 jT/2 !T(x)e-j(x-t)nt.w dX] b..w
We begin the development of the Fourier transform by considering a function n=-oo 27T - T /2
f(t) defined on an infinite interval, but that is not periodic and therefore cannot be
represented by a Fourier series. It may be possible, however, to consider the function to If we define the function
be periodic with an infinite period and extend our previous results to include this case.
The development we will give is nonrigorous, but the results may be obtained rigorously g(w, t) = -1 jT/2
!T(x)e-jw(x-t) dx (17.70)
if f (t) satisfies the Dirichlet conditions of Section 17.2 with the absolute integrability 27T -T /2
i:
condition over a period replaced by full-scale absolute integrability.
If(t)1 dt < 00
Our strategy is to consider the periodic extension fp(t) of the generator !T(t)
(17.66)
then clearly the limit of (17.69) is given by
f(t) = T-+oo
lim
00
~ g(nb..w , t) b..w
~ 07.71)
given by t.w-+o n=-oo
!T(t) = f(t) [u (t +~) - u (t - ~) ] By the fundamental theorem of integral calculus, the last result appears to be
= 1~-00 g(w, t) dw
as illustrated in Fig. 17.13. The exponential Fourier series for fp(t) is
00 f(t) (17.72)
fp(t) = L cnej27rnt/T (17.67)
n=-oo B(lu7t6i9n)t~e limit,.iT -:*.f an~ T -+ 00 in (17.69), so that what appears to be g(w t) in
where . IS really ItS hffilt, WhIch by (17.70) is '
Cn = -
1 jT/2
T -T/2
.
fT(x)e-J27rnx/T dx (17.68)
lim g(w, t) = - 1
T-+oo 27T
100
-00
f(x)e-jw(x-t) dx
720 Chapter 17 Fourier Series and Transform Section 17.6 The Fourier Transform
721
Therefore, (17.71) is actually the superposition is a sum. In the transform case every frequency w contributes an
incremental amplitude F(jw)dw, so an integral over all w is needed.
f(t) = _1 /00 [/00 f(x)e-jw(x-t) dX] dw (17.73)
2n -00 -00 As an example, let us find the transform of
We may write this result in the form
f(t) = e-atu(t)
f(t) = - 1 /00 [/00 f(x)e- Jwx
. dx] e Jwt. dw (17.74)
2n -00 -00
i:
where a > 0. By definition (17.75), we have
Defining the expression in brackets to be the function
F[e-atu(t)] = e-atu(t)e-jwt dt
or F[e-atu(t)] =
= 100 e-(a- jw)t dt
. e-(a-jw)t 1 00
-(a +1 JW) 0
where we have changed the variable of integration from x to t, we have
Because a > 0, the upper limit is
1
The function F(jw) is called the Fourier transform of f(t), and f(t) is called the F[e-atu(t)] = - - . - (17.79)
inverse Fourier transform of F(jw). These facts are often stated symbolically as a+Jw
i:
n=-oo -8/2
oo 2A e jw8 /2 _ e jw8 /2
Fourier transform: f(t) = F(jw)e jwt dw (l7.78b) =
w p
Both express a time function as a superposition of frequency components. In the series . 2A. w8
or F{jw) = - s m - (l7.80b)
case the time function is periodic and only a discrete set of frequencies is needed, so w 2
722 Chapter 17 Fourier Series and Transform Section 17.6 The Fourier Transform 723
f(t)
transformability are more stringent. To complete the justification of (17.83), write the
i:
definitions of the transforms together:
The reader may have noticed a close similarity between the first example, which Find the Fourier transform of f(t) = e- t sin2tu(t). This is a one-
led to the Fourier transform sided t :::: 0 function. From Table 12.2,
2
1 L[e- t sin2tu(t)] = 2
F(e- at u(t)) = - . - - (17.81) (s + 1) + 4
jw+a
By (17.68), we have the desired Fourier transform simply by replac-
and the Laplace transform of the same time function from Chapter 12, ing s by jw.
1 . 2
L(e- at u(t = - - (17.82) F(jw) = (jw + 1)2 + 4
s+a
Even the Fourier transforms of general two-sided functions
Even the steps in the derivation of these results, (12.3) and (12.4) for the Laplace trans- f(t), those not zero for t < 0 or for t > 0, may be found from
form and Example 17.11 for the Fourier transform, echo one another almost exactly. the Laplace transforms of one-sided t :::: 0 functions such as those
This is no coincidence. For any Fourier-transformable function f(t)u(t) which is
listed in the Table 12.2. This technique is demonstrated in the next
zero for t < 0, the two transforms are simply related by example.
Let us call a function f(t) that is zero for t < 0 a one-sided t :::: Ofunction. Then (17.83) Then
asserts that whenever it exists, the Fourier transform of anyone-sided t :::: 0 function is F(jw) = F(e-tu(t)) + F[e+t(u(t + 1) - u(t))] (17.85)
identical to the Laplace transform of the same function with s evaluated at j w. This is
certainly the case for the one-sided t :::: 0 function e- at u(t)(a > 0), which was the subject The first term, the transform of the one-sided t :::: 0 function e-tu(t)
of Example 17.11, as can be seen from (17.79). is given by (17.79) with a = 1.
To see that it is true in general, we first note that the above claim implies that as
long as f(t)u(t) is Fourier transformable, it will also be Laplace transformable. Looking
back at the requirements for transformability, anyone-sided t :::: 0 function satisfying the
FIGURE 17.15 f(t) for Example 17.14.
F(e-\u(t + 1) - u(t))) = 1
-1
ete- jaJt dt = - fO e-'e jaJ , dr:
1
Dirichlet conditions must be absolutely integrable and is therefore of exponential order The second term may be evaluated by changing variables r: = -t in
(with (J' = 0) and hence Laplace transformable. Indeed, the set of Fourier transformable the integral. Now if we replace jw by -s, this integral is identified
functions is a subset of the Laplace transformable functions. Conditions for Fourier as the Laplace transform of the time function e-t(u(t) - u(t - 1.
724 Chapter 17 Fourier Series and Transform Section 17.6 The Fourier Transform 725
Then we can compute this Laplace transform using the techniques One of the most often used properties is linearity, which is a consequence of the
of Chapter 12: fact that the Fourier transform is an integral transform, and thus it is a linear operation.
That is, the transform of the combination
EXERCISES 2 2
F(jw) = - - - - -
2+ jw 3 + jw
17.6.1. Confirm the result of Example 17.14. Find F(jw) by evaluating
2
the integral definition of the Fourier transform (17.75) directly.
17.6.2. Find F[e- a1tl ], where a > O.
(2 + jw)(3 + jw)
2a
Answer -z--z
w +a
17.6.3. Find F[-~(t - 3)e-(t-3) sin2(t - 3)u(t - 3)].
Another operation involving Fourier transforms that we shall find useful is that of
time differentiation. Suppose that we wish to find the Fourier transform of the derivative
e- 3jw (jw + 1) of a function f(t). By definition
Answer --..:.::....,::--~
[jw + l]z +4z
17.6.4. Which of these functions satisfies the Dirichlet conditions? sin t,
t
l/t, u(t), u(t + 1) - u(t - 1), cos (u(t + 1) - u(t - 1. f(t) = - 1
21T
1 00
-00
F(jw)e Jwt
. dw
Answer No; no; no; yes; no
17.6.5. Find the Fourier transform of J(t) = e- Zt , t > 0, e+ 3t , t < O. from which we obtain
5
Answer --=-z- - -
w +6+jw df(t)
-- = -
dt 21T
1 100
-00
d
dt
.
-[F(jw)eJwt]dw
= -1
21T
100
-00
[jwF(jw)]e Jwt
. dw
Therefore, we have
Just as Laplace transform properties summarized in Table 12.1 proved useful in working DIFFERENTIATION
in the s-domain, we will derive and apply Fourier transform properties in this section.
Indeed, the close relation between these transforms suggests that Fourier properties may df(t))
F ( ---;jf = jwF(jw) (17.89)
be derived from Laplace properties. Here we derive the desired Fourier properties from
the integral definition (17.75) of this transform.
726 Chapter 17 Fourier Series and Transform Section 17.7 Fourier Transform Properties 727
That is, the transform of the derivative of J is found by simply multiplying the transform and
of J by jw. This result may be readily extended to the general case:
8
Iwl < 2"
(17.90) 8
Iwl > 2"
where n = 0, 1, 2, .... We are assuming that the derivatives involved exist and that the We have tabulated the Fourier transform operations derived in this section, along
interchange of operations of differentiation and integration is valid. The conditions under with others, in Table 17.1. The reader is asked to derive the others as exercises and
which this is true are quite general and hold for almost any function we are likely to problems.
encounter.
= -1 /00 ..
[F(jw)e-1W']elwt dw
2. Time/frequency scaling
3. Time shift
f(D
f(t - r)
laIF(jwa)
e-jUJrF(jw)
2:rc -00
4. Frequency shift e jUJot f(t) F(jw - jwo)
from which we conclude, by (17.76), that 5. Duality property F(jt) 2n:f(-w)
TIME SHIFT :F(f(t - i = F(jw)e- jw , (17.91) 6. Time/frequency reversal f(-t) F(-jw)
dn
The physical significance of this result is that a delay in the time 7. n-Fold differentiation dt n f (t) (jw)nF(jw), n = 0, 1,2, ...
domain [the function J(t-i) is J(t) delayed i seconds] corresponds dn
8. n-Fold t-multiplication tnf(t) (-I)n-.-F(jw), n = 0, 1,2, ...
to a phase shift (an addition of -Wi to the phase) in the frequency d(Jw)n
domain. Note that i may be positive or negative.
The similarity of the integrals for J(t) and its transform in
(17.75) and (17.76) suggests that interchanging t and w in some We may also use Fourier transforms to extend the concept of frequency response,
way would lead to a new set of transforms. Indeed, if in (17.75), we considered in Chapter 14, to circuits with nonsinusoidal excitations. As we will see,
replace w by x and t by -wand multiply both sides of the equation the response functions are exactly the same as in the case of circuits with sinusoidal
by 2:rc, we have
:F (_1_.
a+
)Jt
= 2:rce aW u(-w), a> 0 FIGURE 17.16 General circuit.
728 Chapter 17 Fourier Series and Transform Section 17.7 Fourier Transform Properties 729
of the circuit to be Also by (17.79), we have
dnvo d n- 1vo dvo
an - -
d n + a - 1 - - - + ... + a l -
n d n 1 dt + aovo 1
t t - dm Vi dm-1 Vi d Vi Vi(jw) = 3 + jw
=bm--+b
dtm -1--++bl-
m dtm- 1 dt +bovI
so that the transform of the output is
Taking the Fourier transform of both sides, making use of entry 7 in Table 17.1 and
linearity, we have Vo(jw) = H(jw)V;(jw)
2
[an (jwt + an-l (jwt- 1 + ... + aljw + aO]Vo(jw)
(2 + jw)(3 + jw)
= [bm(jw)m + bm_ 1(jw)m-l + ... + bijw + bO]Vi(jw)
This can be rewritten, after partial fraction expansion, as
The functions V 0 and Vi are the Fourier transforms of the output and input functions, Vo
Vi (t) = e- 3t u(t)
730 Chapter 17 Fourier Series and Transform Section 1 7.7 Fourier Transform Properties 731
17.7.5. Verify in Exercise 17.7.4 that Assuming that a transient analysis has been specified, the control statement that
2 4 2 directs SPICE to subsequently perfonn a Fourier series analysis is the F 0 UR statement,
F[y(t)] = jw + 1 - jw + 2 + jw + 3 whose fonnat is
and find yet) for the case yeO) = y' (0) = O. .FOUR TINV Fl < F2 F3 .. >
Answer (2e- t - 4e- 2t + 2e- 3t )u(t)
The first parameter TINV is the inverse of the duration T of fr(t), the generator for the
Fourier series. TINV is therefore also the fundamental frequency fa = wo/2n measured
in hertz. The initial time for the generator is taken to be TSTOP - T, where TSTOP is
the end time of the TRAN transient analysis and T the duration of fret), that is, the
inverse of TINV. Fl is the current or voltage containing fret), and F2, F3, ... are
optional additional currents or voltages whose Fourier series is desired. If more than one
\ current or voltage is specified, the same generator duration and initial time are assumed
to apply to all. Only the first 10 components of the Fourier series are computed and
SPICE can be used to find the Fourier series for a periodic function f(t). The fonn output by SPICE, the dc component, the fundamental at fa = TIN V, and the next eight
directly supported by SPICE is harmonic components.
In Section 17.1 it was pointed out that if a generator fret) consists of an exact
integer number of repetitions of a wavefonn, the period of its periodic extension will not
00
coincide with the specified duration of the generator (it will be shorter). SPICE simply
f(t) = do +L Dn sin(nwot + o/n) (17.94) assumes that the generator does not have this property, usually a good assumption since
n=l
SPICE computes approximate sample values, not exact continuous values, for the gen-
erator and could not tell if there is an exact repetition in any event. Therefore, TIN V
will always be output as the fundamental frequency fa of the Fourier series that SPICE
Comparing (17.53) and (17.94), the latter can be easily converted to the more familiar
computes and be listed as component 1 (fundamental component) of the output, as shown
phase-shifted cosine form (17.53) by the change of variables next.
732 Chapter 17 Fourier Series and Transform Section 17.8 SPICE and Fourier Analysis 733
fundamental frequency of 0.159 Hz is used in the F 0 UR command. Let us find some samples from the Fourier transform F(jw) of the
The resulting solution for the decomposition is time function f(t)(u(t + rr) - u(t - rr, for the f(t) sketched in
Fig. 17.3. For T = 2rr, we have WI = 1, and the Cn in (17.95)
are the exponential Fourier series coefficients corresponding to the
FOURIER COMPONENTS OF TRANSIENT RESPONSE V(l)
coefficients found in Example 17.19. For instance, the fundamental,
DC COMPONENT = -2497780E-05
or harmonic component 1, found in that example was found to be
HARMONIC FREQUENCY FOURIER NORMALIZED PHASE NORMALIZED
(DEG) DI = 2.00 and 1ft1 = 0.000584. Let us round this result to 2sint.
NO (HZ) COMPONENT COMPONENT PHASE (DEG)
In terms of the trigonometric Fourier series coefficients, this implies
1
2
1.5~OE-Ol 2.000E+00
3.180E-Ol 1.001E+00
1.000E-00
5.002E-Ol
5.805E-04
-1.800E-02
O.OOOE+OO
-1.800E+02 al = and hI = 2. Thus
3 4.770E-Ol 6.676E-Ol 3.338E-Ol 1.936E-04 -3.869E-04 CI = ~(al - jh l ) =-j
4 6.360-01 5.013E-Ol 2.506E-Ol 1.800E-02 1.800E+02
5 7.950E-Ol 4.016-01 2.008E-Ol 1.156E-04 -4.649E-04 By (17.95),
\
6 9.540E-Ol 3.353E-Ol 1.676E-Ol 1.800E-02 1.800E+02
7 1.113E+00 2.880E-Ol 1.440E-Ol 8.192E-05 -4.986E-04 F(jW)/ = TCn = - j2rr
w=1
8 1.272E+00 2.526E-Ol 1.263E-Ol -1.800E-02 -1.800E+02
9 1.431E+00 2.252E-Ol 1.126E-Ol 6.606E-05 -5.144E-04 In other words, the Fourier transform of the time-limited function
f(t)(u(t + rr) - u(t -rr has the value - j2rr at w = 1. Using the
remaining coefficients from Example 17.19, we may compute F(jw)
Once the Fourier coefficients are determined, the series can be written using (17.94) at w = nWI = n, n = 0, 1, ... , 9.
to describe the original function. The response of a network to a periodic function can
now be found by applying the DC command for the dc component do to the circuit Finally, certain dialects of SPICE permit more complete information on the Fourier
and the . AC command for each harmonic component (amplitude D n , frequency nfo, and transform to be gathered. Using the postprocessor PROBE available with PSpice, the
phase 1ftn). The solution, by superposition, is then the sum of the individual components magnitude and phase, or if preferred the real and imaginary parts, of F (j w) may be
in the time domain. displayed as high-resolution graphics plots. In this book we have chosen to concentrate
SPICE may also be used to find 19 samples from the Fourier transform F(jw) of on those core features available using "plain-vanilla" SPICE and thus leave the reader
certain time functions f(t), those that are zero outside a finite interval of time. Suppose with access to PSpice with PROBE to explore this useful feature on her or his own.
that f(t) is zero outside the time interval [to, to + T]. Let WI = 2rr/T. Evaluating
F(jnwI) = i:
(17.75), the integral defining the Fourier transform F(jw), at w = nWI, is
f(t)e-njwlt dt
EXERCISES
17.8.1. Using SPICE, find the Fourier coefficients for the waveform of
Fig. 17.2(b).
Since f(t) is zero outside the time interval [to, to + T], we may reduce the limits of
integration and then note that the resulting integral is identical to (17.25), the equation 17.8.2. Using SPICE, find the Fourier coefficients for the waveform shown
for the exponential Fourier series coefficient Cn when scaled by T: if T = 10 /-is and A = 1.
f(t)
(17.96) A
-T :
Thus SPICE may be used to compute a set of samples from the Fourier transform, ,,
F(jnwj), n = 0, 1, ... , 9, of a function f(t) zero outside a finite interval [to, to + T]. .. .. '" :
"
f(t) is treated as the generator fT(t) for a Fourier series analysis as described above
and the resulting F 0 UR Fourier series coefficients are converted to exponential form.
These are in tum scaled by T, the duration of fT(t), and we have 19 samples of the EXERCISE 17.8.2
Fourier transform of f(t), that is, F(jw) for w = nWI, n = 0, 1, ... , 9.
734 Chapter 17 Fourier Series and Transform Section 17.8 SPICE and Fourier Analysis 735
(c) 4 cos(2t + 17) 17.12. Obtain the trigonometric series of f(t) defined by
(d) 4 cos(2t + 17)(u(t - 1) - u(t + 1
Fourier analysis is the decomposition of a waveform into sinusoids of different frequen- 17.6. If the t, f (t) axes are translated to the new r, g ( r ) f(t) = 1, -1 < t < 1
cies, amplitudes, and phases. Suppose that the input to a linear circuit is analyzed this axes shown so that the origin for the new axes is the point
(to, fo), the relations between the variables are g = f - fo = 0, 1 < It I < 3
way. Since the gain and phase shift of each sinusoidal component may easily be deter-
and r = t - to. Show that
mined from the frequency response, and the output is simply their sum, Fourier analysis f(t + 4) = f(t)
is often a convenient path to determining a circuit's response. f(t) = fo + g(t - to)
17.13. Find the trigonometric Fourier series for the peri-
Periodic functions are represented by Fourier series, nonperiodic functions by Fourier odic waveform shown.
f(t) g(r) :
transforms. ,,
,
The trigonometric Fourier series is an infinite sum of sines and cosines at integer f(t)
multiple, or harmonic, frequencies. Its Fourier coefficients are real numbers. ,
fa
:, \
--------e----- 2
The exponential Fourier series is an infinite sum of complex exponentials of integer
: (to, fa)
multiple (harmonic) frequencies. Its Fourier coefficients are complex numbers. ,
,
The two forms of the Fourier series are related by c~ = (an + jbn )/2, where Cn is the (0,0) to:
,
exponential Fourier coefficient, an the cosine coefficient, and bn the sine coefficient. -2 -1 2 3
-1
ICn 12 equals the power in the nth frequency component nwo.
FIGURE P17.6
The Fourier transform is defined as the integral transform
FIGURE P17.13
F(jw) = :00 f(t)e- jwt dt 17.7. Determine fay, Ifl;v, and fnns for the periodic
extensions of each of the generators described in Prob-
lem 17.1. 17.14. Find the trigonometric Fourier series for the peri-
IF(jw) I corresponds to the strength of the sinusoidal component at frequency w, /F(jw)
(a) T = 4, tl = -2 odic extensions of each of the generators taken from !1 (t)
to the phase shift at frequency w.
(b) T = 10, to = -2 in Problem 17.1:
The Fourier transform of a sum is the sum of Fourier transforms; scaling a time function (c) T = 2, to = 0 (a) T = 4, to = -2
scales its transform equally. Fourier transformation is a linear operation. 17.8. Determine fay, Ifavl 2 , and fnns for the following (b) T = 10, to = -2
Table 17.1 of Fourier transform operations may be used to derive new transforms from f(t)'s: (c) T = 2, to = 0
known ones. (a) Fig. 17.2(a) 17.15. Find the trigonometric Fourier series for get)
(b) Fig. 17.2(b) dfldt, with f(t) shown in the figure of Problem 17.13.
The Fourier transform of the impulse response h(t) is the frequency response H(jw). (c) Fig. 17.2(c) 17.16. (a) Find the trigonometric Fourier series for f(t)
(d) Fig. 17.2(d) shown.
17.9. For f(t) complex, if Re f(t) is periodic with pe-
PROBLEMS riod Tl and 1m f(t) is periodic with period h is f(t)
17.1. Let fl (t) = u(t + 1) - u(t -1). Sketch the periodic 17.3. Show that if the even part of f (t) is identically zero, necessarily periodic? Justify.
extensions of each of the following generators taken from then f(t) has odd symmetry and if the odd part is zero it 17.10. Find the trigonometric Fourier series for the func-
!1 (t) and indicate the period. T is the duration and to the has even symmetry. tion
initial time for the generator.
(a) T = 4, to = -2
17.4. Suppose!1 (t) and h(t) are periodic with periods f(t) = 41 cos 2tl
nTl and mT2, repeatedly, where nand m are non-negative
(b) T = lO,to =-2
integers. 17.11. Find the trigonometric Fourier series for the func-
(c) T = 2, to = 0 FIGURE P17.16
(a) Is afl(t) + fih(t) periodic? If so, what is its period? tion
17.2. Let h(t) = ret + 1) - 2r(t) + 2r(t - 1) - 2r(t -
(b) Is !1(t)h(t) periodic? If so, what is its period?
2) +r(t - 3). Sketch the periodic extension of the following rr rr
(c) Is (1 + !1(th(t) periodic? If so, what is its period? f (t) = 4 sin(2t + rr /3), --<t<-
generators taken from h (t) and indicate their period. T is 6 3 (b) Find the trigonometric Fourier series for f(t - 1).
the duration and to the initial time for the generator. 17.5. Find the even and odd parts fe(t) and fo(t) if (c) Use linearity to find the trigonometric Fourier series for
(a) T = 2, to = -1 f(t) = : rr 5rr
=0, - <t<- f(t - 1) - f(t).
(b) T = 4, to = -1 (a) h(t) from Problem 17.2 3 6
17.17. Fnd the trigonometric Fourier series for the f(t)
(c) T = 8, to = -4 (b) sin2 2m f(t + rr) = f(t) shown (T = 4). For -1 ::::: t ::::: 1, f(t) = _t 2 + 1.
b2n - 1= - 81
T 0
T 4
/ f(t) sin(2n - l)wot dt
to = -1, T = 4.
17.29. Let get) = f(ct) so that get) is a time scaled
version of f(t) with time compression factor c. Relate Cn,
(b) Find the trigonometric Fourier series for the function the exponential Fourier series coefficients of f(t), to c~,
FIGURE 1'17.17 f(t), one period of which is shown. (Note that this the coefficients for get).
function is even and half-wave-symmetric.) 17.30. Let get) = f(t-r). Relate the exponential Fourier FiGURE 1'17.37
17.18. Find the trigonometric Fourier series for the f(t) series coefficients C n of f(t) and c~ of get).
17.38. Determine the forced output i(t) if the input is(t)
shown. The equation is 17.31. (a) What does it signify about f(t) if all its expo-
f(t)
is the periodic extension of the generator iST (t) = oCt) with
nential Fourier series coefficients are purely real? to = -1 and T = 2. Express as in (17.53).
f (t) = {Sin~, -rr::s t < +rr
4 (b) Repeat for purely imaginary.
T -smt, -2rr ::s t < -rr and rr ::s t < 2rr
17.32. Find the exponential Fourier series for f(t) = IH
with T = 4rr and to = -2rr. 21cost -!I-
17.33. Show that if f(t) has half-wave symmetry, that is,
f[t + T /2] = - f(t), then Cn = 0 for all n even. In
-6 -3 o 3 6
17.34. Find the amplitudes An as in (17.53) of the dc,
-2
fundamental, and second harmonic components of i (t) in
-4 Fig. 17.7 if VI(t) = f(t) in Fig. 17.3.
17.35. Find the complex Fourier series for v(t) if i s, (t) FIGURE 1'17.38
FIGURE 1'17.18
is the waveform illustrated in the following figures. Take
T = 1 in each case, and VSl (t) = 12 V dc. 17.39. Determine the value of the amplitudes An of the
FIGURE 1'17.21
(a) Fig. 17.2(a) dc component and the first three harmonics of the forced
17.19. Find the trigonometric Fourier series for the func- response vet) if Vs (t) is the square wave f(t) of Exam-
(b) Fig. l7.2(b)
tions ple 17.5.
(c) Fig. 17.2(c)
(a) f(t) = 2, 17.22. Let bn be the Fourier series coefficients for
an,
= 1, rr < t < 2rr f(t). Define get) = f(t - r) to be f(t) time shifted to the In
f(t + 2rr) = f(t) right by r. For each of the following r's, determine a~, b~,
(b) f(t) =e t , -rr<t<rr the coefficients for get), in terms of an and bn . T is the
f(t + 2rr) = f(t) period of f(t). +
+
(c) f (t) = t 2 , -1 < t < 1 (a) r = T /2 v
2 n v(t)
f(t + 2) = f(t) (b) r = T IF
17.20. Find the trigonometric Fourier series for the func- (c) r = +T/4
tions 17.23. Suppose a time function f(t) is time scaled get) =
(a) f(t) = 2t + 1, O<t<I f(ct). Relate the coefficients an, bn and the Wo of the FIGURE 1'17.35
FIGURE 1'17.39
= 2t - 1, -1 < t < 0 Fourier series for the two functions f(t) and get).
f(t + 2) = f(t) 17.24. Find the trigonometric Fourier series for df /dt if 17.36. Find the forced response v(t) if Vg is the function 17.40. Find the power delivered by the source in the cir-
(b) f(t) = It I, -I<t<1 f(t) is the sawtooth waveform shown in Fig. 17.3. of Problem 17.12. cuit of Fig. PI7.37 if Vg is the function of Problem 17.19(a).
f(t + 2) = f(t)
A 17.25. Let f
(t) be the periodic extension of the generator g
17.41. Find ~e discrete amplitude, phase, and power
(c) f(t) = 1;1, 0< t < T h(t) = te- t
with to = 0 and T = 2. Find the trigonomet- IilliI spectra for i (t) m Problem 17.39. Sketch.
f(t + T) = f(t) ric Fourier series. D 17.42. Find the discrete amplitude, phase, and power
(d) f(t) = 1 - Itl, -1 < t < 1 17.26. Find the exponential Fourier series for f(t) of IilliI spectra for i(t) in Problem 17.38. Sketch.
f(t + 2) = f(t) Problem 17.13. 17.43. What fraction of the total power in the full-wave
17.21. 17.27. Find the exponential Fourier series for the periodic rectified waveform of Fig. 17.2(b) is in its dc component?
(a) Show that for f(t) with half-wave symmetry, if f(t) is extension of each of these generators taken from II (t) in In the dc plus fundamental components?
even, all the Fourier coefficients are zero except Problem 17.1. FIGURE 1'17.36 17.44. What is the strongest frequency component in
(a) T = 4, to = -2
a2n-1 = - 81
T 0
T 4
/ f(t) cos(2n - l)wot dt
(b) T = 10, to = -2
(c) T = 2, to = 0
17.37. Find the first three terms of the forced response
i(t) if vg(t) is the function of Problem 17.12.
f(t)? If f(t) = vs(t) is high-pass filtered by the circuit
of Problem 17.39, creating vet), what is its strongest fre-
quency component then?
-4 -2 t
fiGURE 1'17.47
I
2 4
riod T = 2rr if f(t) = 8(t) in the period -rr ::::
t < rr. Convert to the form (17.88) used by SPICE.
Now use SPICE to determine how small L\ must be in
the approximation 8(t) ::::: d(t) = IjL\ [u(tl - u(t - L\)]
in order that the dc and first two harmonic frequency
components reported by SPICE agree in both ampli-
17.48. Find the Fourier transforms of e- at cos f3tu(t) and tude (do. Dn) and phase (l/In) to within two significant
e+at cos f3tu( -t) (O! > 0). figures. Any L\ satisfying this requirement is accept-
17.49. Find the transform D(jw) of able.
17.57. Use SPICE to help sketch IF(jw)1 for f(t) =
d(t) = IjL\ [u (t + ~) - u (t - %)] (I - e-t)(u(t) - u(t - 2. First. find a simple single-
loop circuit containing a dc source whose forced response
Noting that d(t) ~ 8(t) as L\ ~ 0, show that the limit of is vet) = I - e- I Then use vU). 0:::: t < 2 and SPICE to
D(jw) as L\ ~ 0 coincides with the Fourier transform of find IF(jw)l. w = nrr for n = O. l. 2 ..... 9. Sketch
8(t) computed directly from the integral definition 17.75. IF(jw)1 based on these sample values.
17.50. Solve the following differential equations by More Challenging Problems
Fourier transforms:
dy 17.58. Let fl(f) = AI COS(Wlt +81). h = A2 COS (W2 t +
(a) dt + 2y = 0 82). Find necessary and sufficient conditions of WI and 'W2
so that fl (t) + h(t) is periodic. repeat for fl (t) . h(t)
(b) dy + 2y = e-tu(t)
dt
17.59. Let fret) be the generator with initial time to= 0
(c) d2Y2 + 3 dy + 2y = e-Itl for its periodic extension f(t). and let gr(t) = /'tto fr(r) dr
dt 2 dt J
Benjamin Franklin
1706-1790
Anne Turgot
witty f'e':6i",1~:ihlltd"si,Al1~~ltla~:;f1iaJ11dln entere.dipto secrd ne;gotiations in Paris
lJl:!:cia:rattpn 'of .fu~It:(p'en(lence in F76, which led to France's
fateful, .perhaps pivqtaj;
," ", ,,, ,\' .eritry into. the Revolutionary War two years later on the
" , '" ' ,
Amerlcanside.
Benjanlin Franklin was born in Boston, Massachusetts in 1706. A child-
hood apprenticeship to Pis brother, it printer, gave him an early love for the
written word .an.da self-taught lllastery of language. In his early twenties he
produced the most respected colonial newspaper, composed the widely popular
Poor Richatd's Almanac filled with his beloved aphorisms, and after convincing
the colonies of the need, even printed their first paper currency. Fascinated upon
seeing sparks emanating from a Leyden jar, Franklin reasoned that lightning has
much the same character, and that the earth and sky must form a massive electric
cell. In 1752 he performed his famous experiment with kite and key, confirming
the common nature of lightning and electricity. It is well for the colonies that
luck was with him; the next two who tried this experiment both died of electrical
shock. Based on this and other experiments, Franklin developed the single-fluid
theory of electrical flow, inventing terms such as battery, positive and negative
charge. His ingenuity and scope as an inventor is demonstrated in the Franklin
stove, bifocal glasses and the lightning rod, each still in common use today.
743
1
Chapter Contents Although this chapter only samples the field of filter design, it clearly demonstrates
the power and versatility of the analysis-based approach to circuit design. At the end of
II 18.1 Passive Filters II Summary our journey, the careful reader is rewarded with a perspective from which development
II 18.2 Active Filters II Problems of real skill at circuit design may confidently be undertaken.
II 18.3 Classical Filters
II 18.4 Filter Transformation and
Scaling
Filters, introduced in Chapter 14, are circuits whose input-output behavior depends
strongly upon frequency. For instance, lowpass filters respond to low-frequency in-
puts, those below a selected cutoff frequency, by passing them to the output relatively
unchanged while attenuating higher-frequency inputs. Filters are perhaps the most ubiq-
uitous of electric circuits, embedded in practically every electronic system of any com-
plexity.
Since inductors and capacitors have impedances that vary with frequency, filter
action can be induced from passive RLC circuits. Passive filters should be c(Jilsidered
The approaches we have explored in the course of our study have been wide ranging, where the power supply requirement of the alternative, active filters, is a significant disad-
beginning with real algebraic and differential equations in the time domain, through vantage (e.g., in battery-operated devices). In addition, they often have a low component
phasors in the static complex domain, to the transform domains of Laplace and Fourier. count, are inexpensive and reliable, and generally tolerate higher power levels than do
Whatever the approach, our focus has remained the same: to advance our growing ability most active filters. Passive filters have their limitations and disadvantages as well, in-
to formulate and solve the most general circuit analysis problems. Excepting only those cluding the requirement that inductors be included in most designs. Inductors tend to be
few sections entitled "Design of ... ," we have not looked beyond strengthening our bulky, nonideal, and poorly suited to miniaturization and integrated circuits.
scope and ability at circuit analysis. Our first design example is a crossover filter for a two-loudspeaker system. Since
Why have we so adamantly preferred analysis to design? While being able to the range of frequencies easily reproduced by a loudspeaker depends strongly upon its size
analyze a given circuit certainly has some satisfactions, to most students of electric and materials, it is usually more cost-effective, for high-quality reproduction, to combine
circuits, analysis is a means to an end rather than an end in itself. The real goal is large "woofers" for low frequencies and much smaller "tweeters" for higher frequencies.
to be able to create interesting and useful circuits, that is, to do circuit design. With It would be inefficient to present the entire input signal to each speaker, since each will
perhaps a hint of regret, it was observed in the initial chapter of this book that it bum power for inputs falling outside its frequency range without producing much audible
is vital to give analysis primacy over design because meaningful design requires and output. Crossover filters match the signal components with the speakers, routing lower
presupposes effective analysis. In retrospect this was certainly reinforced by our ex- frequencies to the woofer and higher frequencies to the tweeter. This is a favorable
perience in the design examples and exercises throughout this book, in which analysis setting for the use of passive filters since relatively high power must be handled and the
always played a prominent role. The payoff for studying circuit analysis comes when size of the speakers themselves can easily accommodate the bulk of the inductors used
we confront nontrivial design challenges armed with the appropriate tools to address in the design.
them.
In this chapter we conclude by addressing one very important class of circuit de-
sign problems, the design of linear filters. First we consider the design of passive filters, We wish to design a crossover network for a two-loudspeaker system.
those consisting exclusively of passive RLC elements. Adding op amps and their required The woofer and tweeter are both modeled as 8-Q resistors, and the
power supplies results in active filters, which can exhibit frequency response character- power amplifier feeding the loudspeaker system has 8-Q resistive
istics difficult or impossible for passive designs, and in addition permit inductor-free output impedance. Assume that the woofer transduces electric to
circuits. A general method for designing active filters of any order based on transfer acoustic energy relatively efficiently for frequencies below 1 kHz
function realization is presented. Classical filters, those with optimal properties such as and the tweeter for those above 1 kHz. Our design goals are to pass
the Butterworth, Chebyshev, and Bessel families, are next defined and their design con- the appropriate frequency ranges to each speaker while maintaining
sidered. The chapter concludes with a look at filter transformation and scaling, operations a good impedance match at all frequencies to the power amplifier
that greatly expand the flexibility of a single basic design. driving the system.
744 Chapter 18 Design of Linear Filters Section 18.1 Passive Filters 745
"
------------------------------ Cr~~~~~-:
At very low frequencies the capacitor blocks current to the tweeter
------!.-~~-_:_...,..,....._:_~...,..,.....~I network ' and the inductor is a very low impedance, so Z (j w) goes to 8 Q.
-----,1
, Similarly, at high frequencies the impedance goes to 8 Q, as can
, be seen from (18.4a). Thus for very low and very high frequencies
_ _ _ _ _ _ _ _ .J
,,
~
Woofer: wrL = _1_ = (f
1_-------
'Ie
",,_, __ ;;' __ ... J
wrC
amp into (18.4a) yields
FIGURE 18.1 Two-way speaker system with crossover circuit.
\
Z('w) = (8 + i./LlC)(8 - i./LlC) =4 ~
J r 16 + 16C (lS.4b)
A simple but effective approach is to p.ut the woofer in series
with an inductor L, the tweeter with a capaCItor C, and plac~ these To have an exact impedance match at the crossover frequency we
. 'ts I'n parallel across the input from the power need
two ClfCUI . . amplIfier as
. p'
shown III Ig. 18 1
.. The inductor will present high . Impedance
h t to L
. h f
hlg requencles,. blocking undesired current flow Illto
.. t e woo her Z(jwr ) = 4 + -16C =8
in that frequency range, and the capacitor will do lIkewIse for t e
low-frequency range. . , d or L =64C (1S.S)
In order that the crossover frequency (at WhICh the magmtu e
of the impedance, and thus the power delivered, t~ ~e two. arms of Substituting the last into (18.3) we have C 2 = 2.53 X 10-8 /64 or
the circuit are equal) be set to a desired value w r , It IS reqUIred that
C = 19.9 p,P, L = 1.27 mH (1S.6)
I
Ii Wr LI = i Wr1 C I (18.1) For these values, the impedance of the loudspeaker system is exactly
matched to the 8-Q power amplifier output impedance for very low
frequencies, very high frequencies, and at the I-kHz crossover. In
or (18.2) between the impedance remains very close to the value desired, as
can be verified with the help of SPICE.
The cutoff frequency is just the resonant frequency of the LC com-
ponents. Since
Design Example 18.1: overall loudspeaker
impedance
Wr = 2n (1000) = 6283 rad/s *
IS 1 0 AC 1
this yields the condition L 1 2 1.27M
C 1 3 19.9U
LC = ~ 10-8 RWOOF
w; = 2.53 X (18.3)
RTWEET 3
2 0
0
8
8
While any pair of LC values satisfying (18.1) wil~ result in highe~ .AC DEC 10 1 1MEG
power dissipation in the woofer at input frequencIes bel.ow 1 kH .PRINT AC VM(l) VM(2) VM(3)
d in the tweeter above 1 kHz as we desire, the overall Impedance .END
:eds to be considered. The equivalent impedance of the system at
frequency W is, by the parallel combination law,
The output VM(I) from this run is the magnitude of the impedance
. (8 + iwL)[8 - i(1/wC)] (18.4a)
Z(]w) = 16 + i(wL - l/wC) of the loudspeaker, since we are using a lLQ A current source and
measuring the resulting voltage magnitude. The values printed all
Power Tweeter Midrange We wish to supply a 300-Q resistive load with 60-Hz power drawn
amp speaker
from the inverter power supply shown in Fig. 18.3(a). Its Thevenin
FIGURE 18.2 Three-way speaker system with crossover circuit. equivalent impedance at 60 Hz is 300/45 Q and the open-circuit
(unloaded) power supply voltage is a square wave with period s.to
748 Chapter 18 Design of Linear Filters Section 18.1 Passive Filters
749
Lr Cr Re{Yd must be nonnegative, since we are limited to passive fil-
~r-----.---c ters consisting of passive elements, so the magnitude of the real
a b
+ part is minimized by setting Re{Yd = o. The imaginary part can
be set to zero by selecting Im{Yd = 0.00236 S. Thus the best
3000 Y j = jO.00236 S, or Zl = - j423.7 Q. This is a capacitor of
value
a' b'
1
(a) (b) C 1 = (2n)(1000)(423.7) = 0.376 /tF
I
60 Hz. . 1
Let Z be the parallel combination of Zl and the 300-Q reSIS- Cr = (0.015)(120n)2 = 469 /tF
tive load. fhe power to the load at 60 Hz is given by IVL 12/300,
a' 0
0"
where, by voltage division, FIGURE 18.4 Power coupling filter for the This large value of capacitance is most conveniently attained by con-
example. necting several smaller capacitors in parallel. The power coupling
filter is shown in Fig. 18.4.
Inverting amplifier +
EXERCISE 18.1.2
Answer \
Noninverting amplifier +
VI
RA
RF
+
with L = 79.6 MH, C = 318 MF Inverting summer
VI +
V2
RF
+
The transfer function H(s) of a stable linear circuit, or equivalently its frequency response Noninverting summer
function H(jw), tells us what the circuit "does." It describes how the circuit will process VI
any frequency component or combination of frequency components present in its input.
For a given H(s) we may sketch a Bode gain plot and see if the transfer function yields
the desired circuit behavior for the filter task at hand. The size and location of the pass-
and stopbands, the gains in these bands, and the slopes between them are determined
by H(s). These filter properties can be modified by adjustment of the poles, zeros, and
constant factor in H(s).
Once we have selected a transfer function H(s) that meets the needs of our design Ch Th; voltages in Table 18.1 may be interpreted in terms of time functions v(t) as in
task, we are faced with the realization problem: find a circuit whose transfer function is apter ,or as phas?rs Vi. or s-domain voltages VieS) where convenient. In Ute latter
H(s). In this section we develop a general design procedure for solving the realization two c~ses, a mo~ent s reVIew of the derivations in Chapter 3 confirms that the same
problem using only resistors, capacitors, and op amps, that is, a procedure leading to equatIOns ~ppl~ if ~he resistances in these building blocks are replaced by im edances
an inductorless active filter realization. Inductors have disadvantages as circuit elements The resultmg CIrCUIts are filters rather than amplifiers, and summing filters r:the th .
summers. r an
within modem miniaturized electronic designs, and we use op amps because the ability
to multiply transfer functions of op amp two-ports connected in cascade makes active
filter design generally simpler than passive design.
We will have need for the building-block circuits introduced in Chapter 3. Their ~et us analyze the circuit of Fig. 18.5. It is an inverting filter with
circuit diagrams and voltage transfer equations are included in Table 18.1. The two Impedances ~F(S) = RF and ZA(S) = RA + 1/sC. The voltage
summers shown may, of course, have more than two inputs, in which case the voltage transfer equatIOn for this circuit is that for the inverting amp with
transfer equation has the corresponding additional terms.
V2(t) = __1_
RC
f vj(t)dt (18.10)
-l--------l--------:"" w (log scale)
The output is an inverted and scaled integral of the input.
We are now ready to develop a procedure for realizing any rational transfer function
(a) (b) R(s) whose denominator polynomial is at least the same order as its numerator. This
includes all proper rational functions (numerator order less than denominator) and those
FIGURE 18.5 (a) Inverting filter H(s) = -ZF(S)/ZA(S); (b) Bode gain
ol equal order. Factor R(s) into a product of transfer functions
plot.
R(s) = KR j (S)R2(S) ... Rq (s) (1S.11)
impedances replacing resistances,
where each Ri (s) is either of the form
-ZF(S)
Vo(s) = ZA(S) VI(S)
One inverting filter will be of particular interest to us, the inverting integrator R(s) = b2s 2 + bls + bo
shown in Fig. IS.6. Its transfer function is given by S2 +als + ao
(18.13)
Vb + Rb
-RCsYa(s) - Vj(s)
(b) Rp +
(a)
Vp + +
FIGURE 18.7 (a) Inverting integrator block; (b) H(s) = l/(s + ao) VI + RI
756 Chapter 18 Design of Linear Filters Section 18.2 Active Filters 757
30 kn 60kn
50 kn 20 {IF 10 kn 100 kn
10 kn
Vb + +
+ +
30 kn
+ 50 kn + 10 kn + +
Vi +
V2 + 50kn
60kn ,
20 kn
I _________________ J
----------------_1 _________________ ~
\
Input summer Inverting integrator Output summer
(-1, +1) (RC = 1) (-10, -10)
FIGURE 18.10 General summer realizing Vo = -2Va - 6Vb + 3Vl + ~V2'
FIGURE 18.11 Circuit for Design Example 18.5.
If all terms match the desired voltage transfer equation, the design is complete. The
a general summer, as in Fig. 18.9, with one inverting input, one
inverting terms will always be correct, but more frequently the noninve~ing. terms will
noninverting input, and perhaps a dummy input as well. Following
each be off by a common factor, necessitating one final step to correct this discrepancy.
the procedure just outlined for designing a general summer, first
That is the case in our example. When we compare (18.21) and (18.22), we find that
select Rp = Ra = 50 kQ to set the inverting coefficient at the
each noninverting coefficient in (18.22) is too high by a factor of 2.
desired value of -1. Then for any noninverting input resistor Rr,
Suppose that we add a dummy input to the noninverting terminal, that. is, an inpu~ V 3
in the absence of a dummy input we have a general summer with
whose value is zero, connected to the noninverting terminal through a resistor R 3 . Smce
RT = RI , RA = 50 kQ, and overall voltage transfer equation, by
we now have one additional resistor in the parallel equivalent defining R T , the value of
(18.20), Vo = -Va + 2V I. Since the noninverting coefficient is too
RT will be reduced. Examining (18.20), the only effect of the dummy input is to decrease
high by a factor of 2, a dummy input at the noninverting terminal
R T , which scales all the noninverting input coefficients downward proportionally. In the
with R2 = RI will have the desired effect of reducing RT, and thus
present case we wish to reduce all the noninverting coefficients by a factor of 2, so we
the noninverting coefficient in the voltage transfer equation, by a
require our corrected RT to equal half its original value of 20 kQ or 20 kQjj R3 = 10 kQ.
factor of 2. The final circuit is shown in Fig. 18.11.
In terms of conductance,
50 X 10-6 + G3 = 100 X 10-6 (18.23) The other type of transfer function in the cascade (18.11) is the biquad given in
(18.13). As with the bilinear circuit, we first consider the related transfer function with
which yields G 3 = 50 X 10-6 S or R3 = 20 kQ. This completes the design, and the unity numerator,
resulting circuit is shown in Fig. 18.10.
In our example the noninverting coefficients were initially too high, leading to a H(s) = Vo(s) = 1
VI(S) s2+als+aO
(18.24)
dummy input at the noninverting terminal. If the noninverting coefficients are, instead,
initially too low, a dummy input must be placed at the inverting input terminal inste~d. Since this is a second-order system we need two integrators, as shown in Fig. 18.l2(a).
RA in (18.20) will then be reduced, not RT, and the effect is to increase all noninve~l~g Cross-multiplying (18.24), we have
coefficients. The value for this dummy input resistor is calculated above as R 3 . ThiS IS
demonstrated in Design Example 18.6.
s2Vo(s) = -alsVo(s) - ao Vo(s) + VI (s) (18.25)
H(s) =
bls + bo = lO(s - 1)
---':'--'--
s+ao s+1 'l(s) ~(s)
summer with coefficients bo = -bl = -10. The input summer is FIGURE 18.12 (a) Double integrator; (b) H(5) = 1/(5 2 + al 5 + ao).
18.2.1. Design a circuit with bilinear transfer function H(s) = (2s + 1)1
(s + 5). Make the largest resistance 100 kQ.
Answer
+
+
100kO
V2(t)
100kO
+
Inverting integrator Inverting integrator Inverting integrator
.l =5 1 1
+ + RC =5 RC =4
RC
Vj(t)
v2(t) EXERCISE 18.2.3
50kO
EXERCISE 18.2.1
Ideal filters are characterized by a constant gain of unity in their passband and dis-
continuous jump to zero gain everywhere outside their passband. Although a useful
idealization, such pure binary frequency-domain behavior cannot be realized in practice.
18.2.2. Design a circuit with biquadratic transfer function H(s) = Filters designed with linear lumped elements such as those described in this book (RLC
s2/(s2 + s + 1).
Make 10 kQ the smallest resistance. What kind of circuit elements, op amps, and linear controlled sources), for instance, have rational transfer
is this? functions H(s), and any rational transfer function with p poles has a maximum roll-
Answer Unity gain high-pass
off of -20 p dB/decade. The gain for real filters with rational transfer functions varies
smoothly and continuously with frequency and is neither exactly constant in the passband
nor exactly equal to zero everywhere else.
Since real filters cannot behave ideally, a performance metric by which real fil-
ters may be judged and compared is needed. In practice, there are several useful
measures by which the deviation of a real filter H(s) from the ideal can be assessed.
lOk!1
Each measures a different feature of the filter's deviation from the ideal, and the rel-
o-----'w.r-----r-+ + ative importance of each depends upon the application at hand. Consider the gain
+
+ + plot of a typical real lowpass filter shown in Fig. 18.15. If this were an ideal filter,
V2(t)
the deviation of gain IH(jw)lmax - IH(jw)lmin in the passband, or passband ripple,
10 k!1 would be zero since the gain is everywhere constant at unity in the ideal filter pass-
band.
Inverting integrator
For an ideal filter the passband slope would be 0 dB/decade throughout the pass-
Input summer (-1.1,1) Inverting integrator
band. The rolloff, or maximum transition band slope, would be infinite and there would
EXERCISE 18.2.2 be no stopband ripple. In contrast to ideal filters, real filters H(s) take on finite nonzero
value for each of these measures. In addition, the behavior of the filter's phase response
(w) = !H(jw) is important in assessing filter performance. A phase shift (w) rad
18.2.3. Design a circuit with the third-order transfer function H(s) = corresponds to (w)/2n periods, or a time shift (time advance) of
-1001s3. If the input to this circuit was a unit step and all initial conditions (w) T = (w) 2n = (w)
were zero, what would the output be? 2n 2n w w
762 Chapter 18 Design of Linear Filters Section 18.3 Classical Filters 763
Passband slope
Gain ?nly o~e. measure of performance is relevant, these filters represent extreme performance
IH(jw) I m explICIt senses and are thus of great value to the filter designer.
(dB)
We describe three such families: the Butterworth, Chebyshev, and Bessel filters.
We focus attenti~n on the normalized lowpass filters of various order for each family.
A~ we shall see m the next section, filters of different bandwidth and of different type
(hIgh-pass, ba.ndpass, bandstop) can easily be derived from the normalized lowpass filter
transfer functIOn R(s) once that has been determined. All three of these "classical"
lowpass filters are of the all-pole type; that is, their transfer functions R(s) possess
constant numerators, so each is of the form
Stopband
ripple
\
(18.28)
--+-----------------~------_t--~--------------~w
Passband ~ I Transition band I ~ Stopband ~ (log scale)
!hus each may be specified by the value of its numerator bo, and either the locations of
ws
Its poles Pk, k = 1, ... , n or its denominator polynomial coefficients ak, k = 0, ... , n-1.
FIGURE 18.15 Filter parameters. Butterwort~ filters are optimal in the sense of being maximally flat in the passband.
No other normalIzed lowpass filter of a given order has a lower rate of change of gain
The time delay r(w) at frequency w is the negative of this time advance:
at w = than the Butterworth filter of that order. The poles of the Butterworth filter of
order n are the n left-half plane solutions of the equation
OdB (s)
3.0
-10 dB
2.0
\
1.0
~--------------------~---------------+--~ w
0.1 (log scale)
-wdB L -_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _- L_ _- L_ _ ~ _ _L __ _ ~ ______ ~ w 1.00 10.0
1.00 1.35 1.84 2.15 2.24 (log scale) FIGURE 18.17 Time delays of selected normalized lowpass filters.
To use this table most effectively, we need to generalize the cascade method for
realization developed in Section 18.2. We seek to avoid factoring third- or higher-order
polynomials for filter orders greater than or equal to 3. Such factorization is required to
reduce a higher-order filter realization to a cascade of biquadratic and bilinear blocks.
Following the logic of Section 18.2 to realize the general transfer function cited in
Table 18.2 and (18.28) in the form of a single block rather than a cascade, we begin with
a chain of n inverting integrators. Labeling the output of the rightmost V(s), the other
voltages are labeled as shown in Fig. 18.19. Designating the input to our filter as Vl(s)
2 and the output as V2(s) gives
S2 + 0.765s +1 S2+ 1.85s + 1
H(s) = V2(S) = bo (18.32)
FIGURE 18.18 Fourth-order Butterworth real ization.
Vl (s) sn + an_ls n- l + ... + als + ao
Table 18.2 lists the coefficients of the transfer functions of selected classical nor-
malized lowpass filters. It should be emphasized that the normalizations are somewhat
different for each filter shown, so comparisons must be made with care. The Bessel (-l)nsnv(s)
filters are normalized to a I-s time delay at dc, the Butterworth filters to -3-dB gain at
w = 1 rad/s, and the ~- and I-dB Chebyshev filters to s-passbands of w = 1 rad/s for
FIGURE 18.19 Chain of n inverting integrators with output Vo(s).
their corresponding s (0.350 and 0.509, respectively).
~--------------------1 +ao
~----------------I-a1 50 JLF
~-----------j+a2 20kO
+
V ou,
770 Chapter 18 Design of Linear Filters Section 18.3 Classical Filters 771
-------------------------~~----~
analysis.
FIGURE 18.23 SPICE gain plot for a Bessel lowpass filter.
FREQ VP (10) 18.3.1. For some radio direction-finding systems it is necessary to deter-
(*)---------- -8.0000E+01 -6.0000E+01 -4.0000E+01 -2.0000E+01 O.OOOOE+OO mine which of two input sinusoids, VI (t) at frequency WI or V2(t) at W2, is
stronger. If the frequencies are close, IWI - w21 is small. What filter family
1.000E-06 -3.607E-04 * would you use to filter out high-frequency noise?
4.083E-03 -1. 472E+00
8.164E-03 -2.994E+00 Answer The Butterworth filters have the least passband slope, so
*
1.225E-02 -4.416E+00 it is least likely that the stronger will be attenuated enough to have a lower
1.633E-02 -5.888E+OO * output amplitude.
2.041E-02 -7.360E+00
2.449E-02 -8.832E+OO 18.3.2. Determine the poles, transfer function, and realization for a third-
2.857E-02 -1.030E+01 * order normalized Butterworth lowpass filter with passband gain +6 dB. Use
3.265E-02 -1.178E+01 \ as many 50-kQ resistors as possible.
3.674E-02 -1.325E+01 Answer Poles -1, 1/120o;
4.082E-02 -1. 472E+01
4.490E-02 -1. 619E+01 * H(s) = (s2+s+~)(S+I) BiquadC2+~+I) plus bilinear L~I) realization:
4.898E-02 -1. 766E+01
5.306E-02 -1. 913E+01 *
5.714E-02 -2.061E+01
6.123E-02 -2.208E+01 SOk!! 20 JLF
6.531E-02 -2.355E+01 *
6.939E-02 -2.502E+01
7.347E-02 -2.649E+01 *
7.755E-02 -2.796E+01
8.163E-02 -2.943E+01 *
8.571E-02 -3.090E+01
8.980E-02 -3.237E+01 *
9.388E-02 -3.384E+01
9.796E-02 -3.531E+01
1.020E-01 -3.678E+01 *
1.061E-01 -3.825E+01
1.102E-01 -3.972E+01 *
1.143E-01 -4.119E+01 EXERCISE 18.3.2
1.184E-01 -4.266E+01 *
1. 224E-01 -4.413E+01
1. 265E-01 -4.560E+01 *
1. 306E-01 -4.706E+01 18.3.3. Find a realization of a second-order Chebyshev (! dB) normal-
1. 34 7E-01 -4.853E+01 * ized lowpass filter.
1.388E-01 -5.000E+01 Answer
1.429E-01 -5.146E+01
1.469E-01 -5.293E+01
1.510E-01 -5.439E+01 *
1.551E-01 -5.586E+01
1.592E-01 -5.732E+01 *
1.633E-01 -5.878E+01
1.673E-01 -6.025E+01 SOk!!
1.714E-01 -6.171E+01
1.755E-01 -6.317E+01 *
1.796E-01 -6.462E+01
1.837E-01 -6.608E+01
1.878E-01 -6.754E+01 *
1.918E-01 -6.899E+01
1.959E-01 -7.044E+01 *
2.000E-01 -7.189E+01
o ,
IH'(}w)I
o IW(jw) I
I
- - - - - - -:- - - - - - - - I - - - - - -
,
In this section we consider ways in which one filter design may be mapped into many.
We will first determine how to transform a filter from one type to another: lowpass
3 dB
to high-pass, high-pass to bandpass, and so on. This technique will permit us to use
lowpass filter transfer functions, such as those specified in Table 18.2, to design clas-
sical filters of many types, not just lowpass filters. We will also determine how to
impedance scale and frequency scale these circuits. Impedance scaling permits easy
modification of a circuit designed to drive a high-impedance load to work effectively
with a low-impedance load or to reduce the power dissipated by a circuit by uniform
reduction of the current levels of all elements. Frequency scaling allows us to dilate
or contract the frequency axis for a given frequency response function H(jw). This w= 1
permits us to modify a normalized filter, such as those normalized to 1 radls listed in (a) (b)
Table 18.2, to any bandwidth desired. The flexibility permitted by filter transformation
and scaling permits the filter designer to work with basic normalized transfer functions, FIGURE 18.25 Transformation H'(s) = H(l/s): (a) lowpass transformed
yet in the end create circuits to meet almost any application need without extensive to high-pass; (b) bandpass with passband below OJ = 1 transformed to one
with passband above OJ = 1 .
redesign.
Given the transfer function H(s), let us consider the effect of creating a new transfer
function H'(S) = H(l/s), that is, of replacing s by its inverse lis. The gain associated Design a normalized second-order Chebyshev high-pass filter with
with the new transfer function is I-dB passband ripple. The table entry for the normalized lowpass
H(s) = 0.983
sZ + 1.l0s + 1.10
But for any transfer function H(s) with real coefficients, H( - jw) = H*(jw), and not- Replacing s by lis, we have our desired high-pass filter transfer
ing that the magnitude of a complex number and its complex conjugate are equal, the function:
expression above may be simplified slightly:
H'(S) = 0.983 0.894s z
(l/s)z + 1.10 (lIs) + 1.10 sZ + s + 0.909
(18.35)
H' (s) is a biquad transfer function that may be realized based on the
block diagram of Fig. 18.13, as several previous biquad examples
After replacement of s by its inverse lis, the new transfer function H'(S) = H(1/s) has, were. Here we modify the method slightly to save one op amp.
for each w, the same gain that H(s) does at the frequency 1/w. As shown in Fig. 18.25, Since there is only a single numerator term, the usual biquad out-
the gain curve is reflected around w = 1. Lowpass filters are mapped to high-pass, put summer is just an output amplifier, in this case a noninverting
bandpass filters with passbands below w = 1 to bandpass filters with pass?ands above amplifier with gain 0.894. However, since this is a linear circuit, if
w = 1, and so on. A common use of this transformation is to map a normalIzed lowpass we scale the input to this circuit by a factor ex, all responses will
filter such as the entries of Table 18.2 into a normalized high-pass filter. The new filter scale by the same factor ex. So if we choose ex = 0.894, we need
will have the same gain at w = 1 radls as the old one, but its passband and stopband will not apply any amplification at all at the output. The resulting block
be reversed. diagram is shown in Fig. 18.26(a), and corresponds to the general
The poles of the transformed filter H(1/s) will clearly be the inverses of the poles block diagram for all-pole filters of even order in Fig. 18.20.
of the original. A pole of H(s) in the second quadrant, for instance, will become a From Fig. 18.26(a) the input summer will have three terms
pole in the third quadrant of H'(S) = H(l/s) (if is an angle in the second quadrant, with coefficients +0.894, + 1, and -0.909. Following the general
_ will be in the third quadrant), and vice versa. Thus left-half plane poles of H(s) summer design procedure and using the notation of (18.29), RF =
become left-half plane poles of the transformed filter, and stability is preserved. It 50 kQ, Ra = 55 kQ, Rl = 100 kQ, and Rz = 89.4 kQ. Then
would not be useful to define a transformation that did not preserve the stability of the RA = Ra = 55 kQ and RT = 100 kQ1I89.4 kQ = 47.2 kQ. The
circuit. noninverting coefficient, which is required to be 0.894, would be,
776 Chapter 18 Design of Linear Filters Section 18.4 Filter Transformation and Scaling 777
**** 04/06/96 15 : 27 : 34. ********* Evaluation PSpice (January 1991) *************
Desl'gn
Example 18.9 2nd order Chebyshev HPF
50kil
**** AC ANALYSIS TEMPERATURE = 27.000 DEG C
********************************************************************************
.-----l--O.909
FREQ VDB(4)
*
1.003E+02 -9.717E-Ol
This has the effect of transforming a lowpass filter H(s) to a bandpass H/(S) with center 1.263E+02 -9.717E-01
frequency at w = 1. To verify this, note that the frequency response of the transformed 1.590E+02 -9.717E-Ol
and by similar reasoning both the dc and high-frequency gains IH/(jO) I and \HI (j 00) I TOTAL JOB TIME .40
of H/(S) correspond to the high-frequency gain \H(joo) I of H(s). For instance, we may
transform the second-order normalized Chebyshev lowpass of Design Example 18.9 to a FIGURE 18.27
Note that by substituting s = jWe = j2n(100), the scaled filter behaves atf = 100 Hz The genera~ion of bandpass and bandstop filters from normalized lowpass filters
just as the original one did at W = 1 rad/s. should be done WIth care. Suppose that we require a bandpass filter with passband Band
Viewed in the circuit domain, frequency scaling by the factor f3 is the specification
of a new set of circuit parameters such that the scaled circuit responds at frequency f3w
center frequ~ncy We. Sin~e the lowpass-to-bandpass transformation H' (s) = H ( 1) s2:
maps the umt frequency Interval 0 ::; w ::; 1 into a unit interval of frequency for the
exactly as the original one does at w. There is only one way W enters into the calculation
of any circuit response, through the impedances jwL and 1/ jwC. If we arrange that the
b~ndpass, the passband of both filters will be the same size. Frequency scaling by We
wIll move the center frequency to the desired location, but also increase the bandwidth by
780 Chapter 18 Design of Linear Filters Section 18.4 Filter Transformation and Scaling
781
IH(jw) I IH'(jw) I
Gain
all voltages are unchanged but all currents are scaled by the constant l/a. To see how a
IH(jw) I
circuit may be impedance scaled, consider its mesh equations in vector matrix form,
ZI=Vs (18.40)
Here I is the vector of independent mesh currents, Vs the independent source vector,
and Z a square connection matrix. If we can find a new circuit with the same Vs whose
connection matrix is scaled to aZ but whose source vector Vs is unchanged, then since
(18.40) implies that
I
,, (aZ)- = Vs (18.41)
a
>!
1 the currents will each be scaled by Ija. Having studied mesh analysis, we know that the
(a) on-diagonal (i, i) element of Z contains the sum of all impedances around ith mesh, and
the off-diagonal elements the negative sum of impedances on the boundary between the
ith and jth meshes. If there are no controlled sources in the circuit, Z consists entirely
Gain of these impedances and will be scaled as in (18.41) if each impedance is scaled by a.
IH(jw) I That is, we must replace each RLC in the original circuit by
C
Rs = aR, Ls = aL, Cs = - (18.42)
a
where the subscript s indicates the values after scaling. To impedance scale by a, mul-
tiply each Rand L value by a, and divide each C value bya. The difference arises, of
course, because the impedance of resistors and inductors is proportional to their element
parameter R or L, while that of a capacitor is inversely proportional to C. To keep Vs
,, ,, the same, as required by (18.41), independent voltage sources are left unchanged, while
we
:< >: independent current sources must be scaled by Ija.
We
The gains of current-controlled current sources and voltage-controlled voltage
(b)
sources in the circuit will not change when impedance scaling, since the ratio of controlled
fiGURE 18.28 (a) Lowpass-to-bandpass transformation H'(s) = He 2 : 1 ); to controlling variable is the same after scaling as before. But current-controlled voltage
(b) subsequent frequency scaling by We sources will have their gain (transresistance r) scaled by a, just as the other elements
whose units are ohms, and voltage-controlled current sources, with gain (transconduc-
the same factor as in Fig. 18.28. We will not have the desired bandwidth B. T? remedy tance g) measured in siemens, must have these gains scaled by Ija. For instance, if
this we must anticipate the final scaling by Wc by producing a bandpass filter wIth. center a circuit with a current-controlled voltage source with source function v(t) = 4i c (t)
frequency w = 1 rad/s but bandwidth B /wc' This may be done by f~equency sc~hng the is to be impedance scaled by 20, the new controlled source will have source function
normalized filter by B /wc before applying the bandpass transforma~lOn. I~, for mstance, v(t) = 80i c (t). Then, with the currents in the new circuit reduced by a factor of 20, the
we frequency scale the lowpass by !,
the bandpass bandwidth wIll be '2 rad/s. After increase in the transresistance of the controlled source by a factor of 20 will compensate
and its voltage will remain unchanged, as required.
scaling by wc , its bandwidth will be w c ! 2 . . . _
Finally, we note that throughout thIS book numencal values such as R. - 1 Q, A common use of impedance scaling arises in the design of op amp circuits. While
C = 2 F, and w = 1 rad/s have been used in almost all ex~ples, exerCIses, and it may be convenient to use values close to unity for R and C in the design process,
problems. Although convenient for pedagogical purposes, and easmg the lab?r of .han~ this will inevitably result in a circuit whose impedance levels are too low for the use of
calculation, use of these values may concern the reader since they are not practIcal ClrcUlt popular low-power op amps such as the /LA 741. These devices operate best at milliampere
parameters for most realistic design applications. . current levels when their voltage levels are an appreciable fraction of the power supply
Fortunately, we can have the best of both worlds. We can frame a ~roblem m a voltage, typically 15 V. We may use convenient values such as R = 1 Q during the
simple numerical setting, solve it, then convert the solution to a. m?re pr~c~cal rang~ of design process, but if we are not to "fry" the op amp by drawing excessive current, we
circuit operation. We saw above how frequency scaling may aid m des~gnmg classIcal had best impedance scale such values to the order of kilohms before flipping the switch.
filters. We conclude with a second type of useful scaling, impedance scah~g. !mpedance This scaling also reduces capacitors to more practical values than 1 F or so. Note that
scaling by a is the specification of a new set of circuit parameters for the ClrcUlt such that since the scaled circuits have the same voltages as prescaled, there is no change to any
voltage transfer function H (s) = V out (s) jVin (s) when a circuit is impedance scaled.
+ o,--"""AII----"
+
EXERCISE 18.4.4
EXERCISE 18.4.1
18.4.2. Compare the op amp count for the cascade realization of the R(s)
of Exercise 1S.3.2 with that for the realization suggested in Exercise 1S.4.1. Filters, modules whose input-output behavior depends upon the input frequencies, are
Answer Cascade realization: six or seven op amps, depending on found in almost all practical electrical and electronic circuits. Filter design is the process
how the numerators of the biquad and bilinear circuits are defined. The real- of specifying a circuit whose input-output behavior complies with prescribed design
ization above has five op amps: two summers and three inverting integrators. specifications, including the desired frequency response. In this concluding chapter a
18.4.3. Design a third-order Butterworth high-pass filter with a -3-dB general method for schematic design of linear filters agreeing with a given frequency
frequency of 5 kHz. response H(j w) or transfer function H(s) is given. The result is a schematic or circuit
Answer First realize the normalized table entry, then scale down diagram of the desired filter. Physical design, that is, selection of appropriate physical
the capacitors by 104 n. Factor into biquad plus buffered RC highpass elements, their physical environment and packaging, is not considered here.
filter.
Passive filters, those containing only passive elements, are practical solutions to filter
design problems in which power supplies or other energy sources are undesirable and
0.637 nF
where discrete rather than integrated circuit implementation is planned.
Inductorless active filters consisting of cascades of bilinear and biquadratic circuits may
be used to realize any stable linear filter. The transfer functions of the several modules
are found by factoring the desired transfer function.
The general bilinear circuit consists of the interconnection of two summers and an
inverting integrator.
The general biquadratic circuit consists of the interconnection of two summers and two
inverting integrators.
Each summer and inverting integrator may be realized using a building block circuit
EXERCISE 18.4.3 with a single op amp.
784 of Summary
Chapter 18 Design Linear Filters 785
Butterworth filters have maximally flat response at the center of thepass~and, C~eby
shev filters have minimal passband ripple, and Bessel filters have maxImally hnear
Z
,----------------------------
I rnH 1 rnH
sponse to the harmonic component. Design a circuit which
will restore the original shape of the received voltage signal
phase shift in the passband. waveform by adding a phase shift of +20 0 at 90 kHz and
The coefficients of normalized dassicallowpass filter transfer functions of all orders are +30 0 at 270 kHz. The phase compensation circuit should
widely found in circuit design reference books. Designs of normalized filters may be not change the gain at either of these frequencies by more
easily converted to filters of other types and passbands using filter conversion, frequency v, + 300n 300n than 3 dB.
and impedance scaling. 018.7. A bridge circuit can be used to detect a small change
[;ill] in resistance. In the circuit below, light falling on the pho-
A normalized lowpass filter may be converted to a high-pass by replacing s .by l/s,
toresistor will cause its resistance to drop, signaling pres-
to a bandpass by replacing s by (s2 + l)/s, and to a bandstop filter by replacmg s by
ence of the light beam.
s /(s2 + 1). ZL
1 ______ - - - - - - - - - - - - - - - - - _ . _ - - -
(a) Show that if RIRp = R2R3 then I = O. This is
Frequency scaling consists in dividing all capacitances and inductances by the frequency the bridge's "null."
FIGURE P18.3 (b) The dark resistance for Rp = 1 kn, and this
scale factor. \
value drops to 800 n when the photoresistor is illuminated.
Impedance scaling consists of multiplying all resistances, indu~tan~e~ ~d transresi~ 018.4. (a) Show how an ideal transformer may be used to
tances in the circuit by the desired impedance scale factor, while .dIVIdmg all .capacI- Specify values for RI, R2, and R3 so that the bridge is
[;ill] make a small capacitor behave like one of greater capaci- nulled in dark conditions, and the microammeter reaches
tances and transconductances by this value. To impedance scale WIthout c~angmg any tance (except at dc).
voltages in the circuit, independent voltage sources are left unchanged but mdependent its full-scale 50 p,A reading when the photoresistor is il-
(b) Use (a) to design a passive two-terminal circuit luminated. Such a circuit can be used to detect an inter-
current sources are divided by the impedance scale factor. whose impedance attains its minimum value of 1 n at 1 Hz. rupted light beam, for instance with elevator door safety
Use only one each of an ideal transformer, resistor, inductor circuits.
with inductance less than or equal to 100 mH, and capacitor
with capacitance less than or equal to 10 mF.
PROBLEMS 018.5. A three-phase 60-Hz generator developing 4.4 kV
o 18.1. (a) Design a passive filter whose voltage gain has 018.2. Repeat 18.1(a) for the uncorrected Bode plot given.
[;ill] Use complex conjugate poles with zero dB correction, that
[;ill] RMS is to be connected via lengthy power lines to a load.
IlliiJ the uncorrected Bode plot shown in the figure. One phase of this balanced wye-wye system is shown in the
(b) Using the filter designed in (a) above, add a load is, damping factor of 1/2. figure. Design a passive delivery circuit by specifying its
resistance RL placed across the output port. For your filter, 1.5
two subcircuits 1 and 2 which will deliver maximum power Vdc-
what value for RL will produce -3 dB loading (that is, the to the load under the condition that the power line current
largest RL so that the loaded and unloaded gains at any IT cannot exceed 1 kA RMS and voltage V T cannot exceed Photoresistor
given frequency differ by 3 dB)? This RL may be s~eci~ed 44 kV RMS.
as the smallest permissible load resistance for the CIrcUIt. OdB
IT : - - - - - - - - - -:
.---------,--7 :o.lkn : :- -- - -- ---:
50kn: FIGURE P18.7
+
Gain (dB)
Subcircuit
1
VT Subcircuit
2
g 1~.8. Design a passive RLC circuit whose voltage gain is
[;ill] gIven by
-3 dB 1-------..
{J) 0.1 H
~~L----------------------T--~~
(log scale)
.. _ - - - ______ 1
, 1_ _ _ _ _ _ - - - - :
Load
OdB
Gain
(dB)
-----
FIGURE P18.2 Transmission
Gain (dB) line
FIGURE P18.5
018.3. (a) Design an RC coupling circuit Z which will
{J)
[;ill] maximize the power delivered to the load Z L at 1 kHz. g 18.6. After filtering a periodic voltage signal detected by
~~------------------------~~2n (b) Now make the design in (a) adjustable so it can ~e [;ill] an ultrasonic transducer, the two principal signal compo- -20 dB V V {J)
(log scale) set to maximize power delivered to ZL at any frequency m nents are a fundamental frequency component at 90 kHz 1 kHz 10 kHz 100 kHz
2n
30Hz (log scale)
the range 500 Hz ::: f ::: 2 kHz. Which circuit parameters and its harmonic at 270 kHz. The filter has introduced a
FIGURE P18.1 of Z need to be made adjustable, and over what ranges? phase shift of -20 0 in the fundamental and -30 0 in the re- FIGURE P18.8
Gain
D 18.11. Replace the resistors in the noninverting amplifier (dB)
lill!I of Table 18.1 by impedances, and use the resulting non-
R R inverting filter to design a circuit with uncorrected Bode -10 dB
+ ~------~-T-r--------------~ ~XlooO
+ -----.-----'VV\.-----~------'V'V\r--...._- plot shown. w
-+~----~r------r------~--~2n 2 4 8 (log scale)
(log scale) (kHz)
100Hz 400Hz 800Hz
C C FIGURE P18.20
FIGURE P18.15
+20dB/dec
R
2C
2 D 18.21. If the harmonics comprising a periodic signal are
D
.... 18.16. Design a circuit with voltage transfer function lill!I arbitrarily phase-shifted, the shape of the signal will be dis-
.... torted. This may be caused by frequency-dependent time
100s 2 +s +4 X 107 delays as the signal propagates through air, water or some
H(s) - - - - - - - . : - - -
FIGURE P18.9 OdB
- (s + 200)2 other medium. Design a circuit called an adjustable phase-
shifter which introduces a selectable phase shift at a se-
What kind of filter is this? Justify. lectable frequency while leaving all other phases, and all
D gains, relatively unchanged. The phase shift should be ad-
.... 18.17. Design a circuit with voltage transfer function
.... justable 90 in the frequency range 1 kHz-10kHz. The
-+------------1---------------~~ w gain should be 0 dB 1.5 dB at all frequencies, and the
2n H(s) = s2 - 30s + 200
200Hz (log scale) phase shift one octave or more away from the selected fre-
(b) Note that this circuit assumes exactly matched s2 + 30s + 200
quency should be less than 6. Using SPICE, show the
elements. Suppose the R /2 Q resistor in the figure is re- FIGURE P18.11 gain and phase shift curves when phase-shifting +30 at
What kind of filter is this? Justify.
placed by one with resistance (1 + p)(R/2) Q. Sketch a 2 kHz and -45 at 8 kHz. What circuit elements need to
graph showing the depth of the notch in dB as a function ~ 18.12. Repeat 18.11 for the voltage transfer function ~ 18.18. Design a circuit with voltage transfer function
be made adjustable? (Hint: consider complex poles and
of the mismatch parameter p. Note that for p = 0 there is zeros with the same break frequency but whose real parts
no mismatch and the notch is infinitely deep (in dB), and 3.77 x 106
H(s) = s2 + 6s + 4 H(s)= s+377 have opposite signs.)
the larger Ipi the greater the mismatch. Use SPICE to help
s2 +4s +4 18.22. A limb-lead electrocardiogram (ECG) has three
generate this curve.
Do not use any building block circuit with absolute value inputs: body surface voltages measured on the left arm
D 18.10. Design a passive circuit which will drive a display D 18.13. Replace the resistors of the noninverting summer of gain exceeding 10 on any of its inputs. (La), right arm (Ra), and left leg (Ll) all relative to the
im containing three light emitting diodes (LEDs) which show lill!I of Table 18.1 by impedances, and use the resulting build- D 18.19. Design an audio amplifier with 40 dB/decade roll-
right leg (Rl) voltage common ground as shown. De-
the dominant frequency range of the input signal. The input ing block circuit to design a circuit with voltage transfer sign an ECG amp so that its output Vo is proportional
to the circuit is an ac steady state 1 rnA RMS current. If equation Vo(s) = 1/3Vl (s) + 2/(s + I)V2(s). Note that a i!!! off in the uncorrected Bode plot in both the high and low
to VLa - 0.5(VRa + vu). Vo is referred to as aVL, the
more than half the power in the input is in frequencies in dummy input may be necessary. frequencies, lower break frequency 100 Hz and upper break
"augmented left arm lead voltage." The constant of pro-
the range 20 Hz-500 Hz the LED marked "bass" should 18.14. Replace the resistors of the circuits of Table 18.1 s
frequency 10 kHz, damping factor = 1/4 for all poles,
portionality should be such that when VLa = 1 mV and
by impedances and use the resulting building block circuits and + 100 dB gain in the center of the passband.
light up, if more than half is in the range 500 Hz-2 kHz the VRa = Vu = 0, we have Vo = 1 V within the pass-
LED marked "midrange" should light up, and if in the range to design to the desired triple-pole voltage transfer function D 18.20. Design a bandpass circuit whose passband is sur- band (0.1 Hz to 20 Hz). The output should roll off at
2 kHz-20 kHz the LED denoted "treble" should light up. mi rounded by guard bands in which the gain is required to be 40 dB/decade on either side of these frequency limits. Use
Assume that the LED's light up when the voltage across H s _ 10 very low, as shown in the accompanying figure. The gain a multi-op amp design so that no gain on any input exceeds
( ) - (s + 1)3
them reaches 5 V RMS. must be + 10 dB or higher at 4 kHz, the gains one octave 10 in magnitude for any single op amp.
Appendix
A
Matrix Methods
may be concisely written as a single vector-matrix equation, and matrix methods are
then used for its solution. Sets of equations of this type (A.I) arise naturally in applying
systematic methods of circuit analysis, such as nodal or mesh analysis, to linear circuits.
Here we summarize the elements of linear algebra leading to the vector-matrix form and
its solution by matrix inversion, Cramer's rule, and Gauss elimination.
793
~ ,
A of dimension mA x nA and B of dimension mB x nB (nA mB), their matrix the n unknowns in the set of scalar equations as the n elements of the vector x. Then
product following (AA), consider the first given equation as the first row of the vector-matrix
C=AB equation (A3). This requires that Yi with i = 1 in (AA) match YI in (AI) and that the
elements of the first row of the connection matrix A match the coefficients of the first
is defined to be the n A x m B matrix C whose elements are
equation in (Al). Continuing in this manner for each row, the vector-matrix form (A.3)
nA
Cij = L aikbkj
may be written by inspection of the set of scalar equations (AI).
For instance, consider the three equations in three unknowns
k=1
Matrix multiplication C = AB can be carried out by writing A below and to the left of B 3vI + V2 - 2V3 =0
and then extending the rows of A and columns of B into the space thus formed, as shown
VI - 2V2 - V3 = -6 (AS)
in Fig. AI. Each intersection corresponds to an element of the matrix C. To determine
its value, move from a given intersection to the farthest elements in A and B along the -VI + V2 - 2V3 = 2
intersecting "tracks," and multiply these scalars together. Move one element closer to the We will convert to the vector-matrix form Ax = y with
intersection along both tracks, multiply these elements together, and add to the previous.
Repeat until the intersection is reached and enter the accumulated value there. If the
number of elements along the tracks do not pair exactly, the matrices are not conformable.
An m x 1 matrix x with a single column is called a column vector, or simply a
vector. The elements Xi, i = 1, ... , n of a vector x need only one index, indicating the which will give
row location of the element. Following the rules of matrix multiplication, the product of
an m x n matrix A and a n-vector x is an m x 1 matrix, or m-vector y,
Ax=y (A.3)
[ :~: :~~ :~:] [~~] [~6]
a31 a32 a33 V3
=
2
whose elements are, for i = 1, ... , m, According to (A4) with i = 1, we must have
n
+ a12V2 + al3V3 = 0
=L
all VI
Yi aikxk (AA)
k=1 Comparing the first equation in (AS), it follows that all = 3, al2 = 1, and al3 = -2.
We shall have need of the operation of matrix multiplication in converting equations to Repeating for the second and third rows, the desired vector-matrix form of the equations
our desired form and in solving the resulting equation. (AS) is
Finally, the transpose AT of an m x n matrix A is defined as an n x m matrix
whose (i, j) element is the (j, i) element of A. That is, AT is the mirror image of A
flipped around the main diagonal. The transpose matrix will be helpful in computing the
inverse of a matrix, which we shall do shortly. In the remainder of this appendix we will summarize methods for solving the vector-
matrix equation Ax = y in the case that the connection matrix A is square, in other
B: [t ; ;] words, that there are as many unknowns as equations.
A:[~
-e---e--_ W] .---.-
AB AB
(a) (b)
Both principal methods for solving Ax = y, Cramer's rule and matrix inversion, depend
FIGURE A.l Matrix multiplication. on calculation of determinants. Let A be a square matrix of dimension n x n. The
determinant ~ of A, denoted IAI, is a scalar function of A computed as follows. For
n = 1, A is the 1 x 1 matrix all, and it is its own determinant.
~ = all (A6)
For n = 2,
A set of m equations in n unknowns, such as (AI), may be easily converted to a single
equation of the form (A3), said to be in vector-matrix form, with x referred to as the
unknown vector, y the known vector, and A the connection matrix. Begin by assigning (A7)
al1
a12 a131
D. = Ia21
a22 a23 (A.9) = al1 A l! - a12A12 + a13 A 13
a31 a32 a33
Writing out the minors explicitly, we have
may also be evaluated by a diagonal rule, given by
A = all
'-" I
a22 I
a231 - a12 a21
a32 a33 a31
D.= By the diagonal rule, this may be written as
(A.lO)
D. = al1 (a22a33 - a23a32) - a12(a21a33 - a23a3d + a13(a21a32 - a22a31)
D. = I;
-1
~11 2~ I = (2 - 1) - (1 + 1) + 2(-1 - 2) = -7
6xI +X2 = 4
The determinant of A is
Cramer's rule states that
t. = (1)(1) - (-2)(6) = 13
Each cofactor Cij is found by striking out the ith row and jth column of A, computing
t.1 5(1) - 4(-2)
XI=-= - -1 the determinant of what remains, and multiplying by (_l)i+ j After striking out the first
t. - 1(1) - 6(-2) -
'row and column, C II is just the determinant of the remaining 1 x 1 matrix +1, multiplied
by (-1) 1 + 1 = 1, or + 1. C 12 is found by striking out the first row and second column,
and computing the determinant of the remaining matrix, 6, and multiplying by (-1) I+2 = -1.
The result is C12 = -6. Proceeding in this fashion, the matrix of cofactors is
_1~(4-=--)_-_5-=--(6-=--) =-2
1(1) - 6(-2)
and after transposing, to get the adjoint, the inverse is found as
A_I=adjA=~[1-6
where the determinants are computed by (A7). Cramer's rule is particularly efficient
2
when only one or a small number of elements of the vector of unknowns are needed. 2J=[113 13]
t. 13 1 -6 1
- -
13 13
Then applying (AI6), the solution is
The second method we shall summarize for solving the vector-matrix equation
[~: 1J ~~ ~ ][~J [~2 J
Ax=y (AI3)
where A is a square n x n connection matrix and x and y are n-vectors, is called matrix
13 13l =
which agrees with our result using Cramer's rule. Matrix inversion is frequently more
mverSIOn. Suppose we can find an n x n matrix A -I, which has the property that efficient than repeated use of Cramer's rule when most or all of the elements of the
A-IA = I (A14) unknown vector are required.
where I is the n x n identity matrix [each (i, i) element is 1, each (i, j) element is 0 for
i not equal to j). Then, premultiplying both sides of (A 13) by A-I,
A-lAx = Ix = A-Iy (AlS)
or, since Ix = x for any conformable matrix x, A final method for solving Ax = y for square matrices A, based on successive elimination
of unknowns, is called Gauss elimination. Unlike Cramer's rule and matrix inversion, it
x =A-Iy (AI6)
can be developed without resort to the matrix formalism. Here we present it as a matrix
which is the desired solution. The matrix A -I is called the inverse of A. method, since it is useful in that context also.
To determine A -I, begin by computing the n x n matrix whose (i,j)th element is We shall illustrate the method for a 3 x 3 case, the generalization to other dimensions
the cofactor Cij defined in (A12). The transpose of this matrix of cofactors is denoted is evident. Consider the vector-matrix equation
adj A (the adjoint of A). Then
+]m=[+]
1 1
[ o -3
o 0
The matrix is in upper-triangular form. The final row equation may now be easily solved
for X3, because no other unknowns enter this equation. The solution is X3 = 3. Moving
up one row, after back-substitution of the value for X3 the second equation has only the
single unknown Xz. Upon division of this equation by -3, we have
X3
Xz = 3- 3" = 3 - 1=2
and back-substituting into the first row equation,
Xl = 6- Xz - X3 = 6- 2 - 3= 1
which completes the solution.
The method of Gauss elimination can be viewed as the formation of an augmented
matrix containing both A and y. In the case of the example just completed, this is the
augmented matrix
1
-1
1 6]
1 3
2 7
Subtracting twice the first row from the second and adding the first row to the third and
the augmented matrix becomes
[ 0~ 0~ 1~ 3~]
800 App. A Matrix Methods Section A.6 Gauss Elimination 801
Appendix
B
Complex Numbers and the
Complex Exponential
\
From our earliest training in arithmetic we have dealt with real numbers, such as 3, -5,
~, :rr, and so on, which may be used to measure distances along the real line from a fixed
point. A number such as x that satisfies
(B.1)
is not a real number and is customarily, and unfortunately, called an imaginary number.
To deal with imaginary numbers, an imaginary unit, denoted by j, is defined by
j=~ (B.2)
Thus we have j2 = -1, j3 = - j, j4 = 1, and so on. (We might note that mathematicians
use the symbol i for the imaginary unit, but in electrical engineering this might be
confused with current.) An imaginary number is defined as the product of j with a real
number, such as x = j2. In this case x 2 = (j2)2 = -4, and thus x is a solution of
(B.1).
A complex number is the sum of a real number and an imaginary number, such as
A= a + jb (B.3)
where a and b are real. The complex number A has a real part, a, and an imaginary
part, b, which are sometimes expressed as
a = ReA
b=ImA
It is important to note that both parts are real, in spite of their names.
The complex number a + j b may be represented on a rectangular coordinate plane,
or a complex plane, by interpreting it as a point (a, b). That is, the horizontal coordinate
is a and the vertical coordinate is b, as shown in Fig. B.l, for the case 4+ j3. Because of
803
~
'j
Imaginary axis 1m
4
3 .4+j3
~~-L------L--Re
4
...L_.....L_-+_---1_---1_ _L-_L-_ Real axis
-2 -\ 0 2 fiGURE B.3 Two forms of a complex number A.
fiGURE B.l Graphical representation of a complex number. We may easily convert from rectangular to polar form, or vice versa, by means of
{B.4) and (B.5). For example, the number A, shown in Fig. B.3, is given by
this analogy with points plotted on a rectangular coordinate system, (B.3) is sometimes
called the rectangular form of the complex number A. A = 4 + j3 = 5/36.9
The complex number A = a + j b may also be uniquely located in the complex since by (B.4)
plane by specifying its distance r along a straight line from the origin and the angle e
that this line makes with the real axis, as shown in Fig. B.2. From the right triangle thus r= J4 2 +3 2 = 5
formed, we see that 3
r = Ja 2 +b2 e= tan- 1 4" = 36.9
1 b
(BA) The conjugate of the complex number A = a + j b is defined to be
e= tan- -
a A*=a-jb (B.7)
and that That is, j is replaced by - j. Since we have
a = rcose
b=rsine
(B.5) IA*I = Ja 2 + (-b)2 = Ja 2 +b2 = IAI
and
We denote this representation of the complex number by
A=rM (B.6) argA* = tan- 1 (~b) = -tan- 1 ~ = -argA
which is called the polar form. The number r is called the magnitude and is sometimes we may write, in polar form,
denoted by
(rM)* = r/-e (B.8)
r = IAI
We may note from the definition that if A * is the conjugate of A then A is the
The number e is the angle or argument and is often denoted by
conjugate of A *. That is, (A *)* = A.
e = angA = argA The operations of addition, subtraction, multiplication, and division apply to com-
plex numbers exactly as they do to real numbers. In the case of addition and subtraction,
1m we may write, in general,
804 App. B Complex Numbers and the Complex Exponential Section B.l Complex Numbers 805
1m 1m
results in an irrational denominator, since j = R. We may rationalize the denominator
A A and display the real and imaginary parts of N by writing
A+B B _ AB* _ a + j b c - jd
N - -----._-
BB* c+jd c-jd
A+B which is
N = (ac + bd) + j (bc - ad)
c2 +d 2
(B.I5)
B We may also show by the method used to obtain (R13) that
FIGURE 8.4 Two methods of graphical addition. rIl81 r1
-=. = -/8 1 - 82 (B.16)
r2/ 82 r2
This may also be done graphically, as shown in Fig. B.4(a), where the number A and \
B are represented as vectors from the origin. The result is equivalent to completing the
As examples, let A = 4 + j3 = 5/36.9 and B = 5 + j12 = 13/67.4. Then we
have
parallelogram or to connecting the vectors A and B in head-to-tail manner, ~s shown
in Fig. B.4(b), as the reader may check by comparing the numbers. For thiS reason, AB = (5)(13)/36.9 + 67.4 = 65/104.3
complex number addition is sometimes called vector addition. and
In the case of multiplication of numbers A and B given by
A 5
A = a + jb = r1 cos81 + jr1 sin 81 Ii = 13/ 36 .9 - 67.4 = 0.3S5/-30S
(B.Il)
Evidently, it is easier to add and subtract complex numbers in rectangular form and
to multiply and divide them in polar form.
we have
AB = (a + jb)(c + jd) = ac + jad + jbc + j 2bd
(B.12)
= (ac - bd) + j(ad + bc)
Alternatively we have, from (B.lI), Complex numbers are widely used in electrical engineering because of the intimate link
AB = (r1 cos 81 + jr1 sin8])(r2 cos 82 + jr2 sin 82) between real sines and cosines and the complex exponential function e je . To expose this
relationship, define
= r1 r2 [(cos 81 cos 82 - sin 81 sin 82) + j (sin 81 cos 82 + cos 81 sin 82)]
g = cos 8 +j sin 8 (B.l7)
= r1r2[cos(81 + 82) + j sin(81 + 82)] Differentiating,
Therefore, in polar form we have
dg
d8 =]
.(cos 8 + ism
.. 8) =]g
.
(rIl81 )(r2/82) = r1 r2/ 81 + 82 (B.13)
This is a first-order unforced linear differential equation with general solution ke se , where
and hence we may multiply two numbers by mUltiplying their magnitudes and adding s satisfies the characteristic equation
their angles.
From this result we see that s-j=O
AA* = (rl.2.)(r/-8) = r 2LSl = IAI 2LSl and, evaluating (B.17) at 8 = 0, k = 1. Thus g = e je and, by (B.17),
The polar fonn of the Euler fonnula (B .18b) suggests that any complex number may
be represented as a complex exponential. Let x have magnitude r 2:: 0 and angle e. Then
x = r!Ji
= (rLQ)(I!Ji)
Thus, along with the rectangular and polar fonns to represent complex numbers, we have
the exponential form
x = a + j b (rectangular fonn) (B.21a)
= r!Ji (polar fonn) (B.21b)
= re jlJ (exponential fonn) (B.21c)
For instance, the complex number in rectangular fonn x = 1 + j has polar representation
wt
x = J214So and exponential representation x = J2e j45.
A principal use of the exponential fonn is in fonning phase-shifted complex expo- fiGURE 8.5 Graphical illustration of Euler's formula.
nentials. For instance, suppose
Euler's fonnula is illustrated graphically in Fig. B.S. A unit vector is rotating
i(t) = xe jwt (B.22) around a circle in the direction shown, with an angular velocity of w rad/s. Therefore, in
where x is a complex number. Then, using the exponential fonn(B.21), t seconds it has moved through an angle wt as shown, and thus the vector may be specified
i (t) = re jlJ e jwt = rej(wt+lJ) by 11.!EJ.. or e jwt Its real part is the projection on the horizontal axis, given by cos wt,
and its imaginary part is the projection on the vertical axis, given by sin wt. That is,
By the Euler fonnula (B.18a), the last is easily recognized to have real part r cos(wt +e)
and imaginary part r sin(wt + e). Extracting real and imaginary parts of expressions such
e jwt = cos wt +j sin wt
as (B.22) is an integral part of phasor analysis. which is Euler's fonnula in rectangular fonn (B.18a). The projections trace out the cosine
The exponential fonn can be used to concisely justify the multiplication and division and sine waves, as shown, as the vector rotates with period wT = 2n or T = 2n jw.
fonnulas presented earlier. Let
x = rJ/!J, y = r2fJJ:
Then
xy = (r1ejIJI)(r2ejIJ2 = (r1r2)ej(IJI+1J2)
as in (B.16).
808 App. B Complex Numbers and the Complex Exponential Section B.2 Complex Exponential Function 809
Appendix
........................ ., ................................................ " ....... "
c....
Circuit Topology
The general methods of nodal and mesh analysis emphasized in this text result in an
equal number of equations as unknowns. This in itself is not enough to guarantee that we
can use these equations to "break" the circuit, that is, evaluate their unknowns, the node
voltages or mesh currents, and from them all other voltages and currents. It is easy to write
k equations in k unknowns for which there are no solutions or for which an infinite number
of equally valid solutions exist. In neither case could we go on to complete the circuit
analysis from these equations, that is, find unique values for all currents and voltages.
In this appendix we shall present some of the key results from a study of the
circuit topology, or the manner in which the elements are interconnected. This proves an
effective way to determine how many equations are required to analyze a given circuit,
which ones are independent, and the most efficient set of equations for analysis. We seek
the smallest sets of equations with equal numbers of unknowns, while, also seeking to
avoid the dual pitfalls of an infinite number of solutions or no solutions at all.
In studying the topology of a circuit, we will not distinguish individual elements by what
they are, resistors, current sources, and so on, but by where they are. The configuration
of lines and nodes formed by replacing elements of a circuit by featureless lines is called
its circuit graph or network graph. To emphasize the great generality of this graph, its
capacity to represent networks of all kinds, not just electric circuits, we will refer to it
as a network graph hereafter. The lines in a network graph are called its branches.
A network graph is connected if there is a path between any two nodes. Our
discussion will assume the graph is connected, since, if it is not, each disjoint part may
be analyzed separately as a connected graph.
A tree is a sub graph with all nodes of the full network graph but no loops. An
example of a simple circuit, its network graph, and two distinct trees are shown in
Fig. c.l. While a given network graph with B branches and N nodes may have many
811
,'I
c 3 2 3 remaining voltages as well. From the resulting complete set of voltages we may recover
the currents by the element i-v laws. Thus, if we can find N - 1 independent equations
in the tree branch voltages, these will suffice for complete network analysis.
Let us write KCL at anyone node in the network graph. Moving to any other node,
the KCL equation at this second node must be independent of the first if N > 2. This
follows from the fact that the second equation contains at least one branch not referenced
in the first equation. For if the second equation contained all the same branches as the
o o first, every branch connected to one node would terminate at the other, in other words,
(a) (b) N = 2. Moving to a third node, its KCL contains a new branch not referenced in either
of the first two as long as N > 3, or, if it does not, then N = 3. We conclude that any
set of N - 1 KCL node equations is independent.
2 3
e-----..'2 --- ------- 3 \ For instance, referring to Fig. C.1 (c), let inm refer to the current flowing in branch bnm
from node n to node m. Then KCL at node 0 is
o o ilO + i l2 = 0
(c) (d) This second equation has a new branch current, i l2 , not in the first, so is independent.
KCL at node 2 is
FIGURE C.l (a) Circuit; (b) network graph; (c) one tree; (d) another tree.
i 2l + i 20 + i23 = 0
trees, each tree has exactly N -1 branches. This follows from the fact that it must connect
all N nodes, but cannot have any closed loops. The branches in a tree are called tree Once again the new KCL is independent of the others, since it contains a new branch
branches and the remaining branches are links. For instance, the branch b OI connecting current, i 23 The final KCL equation, at node 3, is
nodes 0 and 1 in Fig. C.I is a tree branch for both trees shown, while b02 is a tree branch
for Fig. C.1(c) and a link for the tree of Fig. C.1(d). Since there are N -1 tree branches, i32 + i30 = 0
there must be B - (N - 1) = B - N + 1 links in any tree. The set of these B - N + 1 Neither term is new to the set of four node equations, and this equation is not independent
links is called the co tree associated with a given tree. The union of the tree and cotree of the first three. In fact, it is simply the negative sum of the first three equations. In
is the entire network graph. this network graph there are four nodes, N = 4, there are N - 1 = 3 independent KCL
nodal equations, and the fourth is not independent of the first three.
Now suppose we choose to write N - 1 KCL node equations in terms of N - 1 tree
branch voltages. Each branch voltage, whether tree branch or link, may be expressed in
terms of the N - 1 tree branch voltages, and, with the help of the i-v laws, so can the
branch currents. Since the resulting N - 1 equations in N - 1 unknowns are independent,
Consider a network graph with N nodes, B branches, and a given tree. Adding a single it follows that we can find unique solutions for the tree branch voltages, and hence all
link to the tree results in a closed loop, since there is a tree path between the nodes at each voltages and currents. We conclude that any N - 1 KCL node equations in terms of the
end of the link, and the second path formed by the link itself must close a loop. Thus, by tree branch voltages for any tree are independent and will suffice for complete circuit
KVL, the link voltage is fixed by the tree branch voltages. On the other hand, the set of analysis.
tree branch voltages themselves are completely unconstrained by KVL, since they contain It is usually more convenient to work with node voltages than tree branch voltages,
no closed loops. The tree branch voltages may take on any set of values independently since then we are not required to construct a tree. N - 1 given node voltages fix the
of one another. Thus the N - 1 tree branch voltages are an independent set of voltages, full set of N - 1 tree branch voltages, since the tree branch voltages are differences of
and the remaining B - N + 1 link voltages are fixed by the tree branch voltages. All node voltages. Equally, the N - 1 tree branch voltages fix the N - 1 node voltages, as
branch voltages may be expressed in terms of the N - 1 tree branch voltages. can be seen by starting with a tree branch connected to the reference node, labeling the
It is well known that a set of k independent equations in k unknowns possesses voltage at the reference node zero, and at the other node voltage equal to the tree branch
a unique solution. Thus, if we can find N - 1 independent equations in the N - 1 voltage. Since the two sets of N - 1 voltages are equivalent to one another, and one is
tree branch voltages, we can find their values and, since they fix the link voltages, all independent, so must the other be.
---- -- - - - - - - - - - -
We conclude that KCL equations at any N - 1 nodes written in terms of the node there are as many KVL mesh equations as mesh currents. Moreover, each branch current
voltages are independent and will suffice for complete circuit analysis. This is what we may be expressed in terms of mesh currents, and, with the help of the i-v laws, each
refer to as nodal analysis in Chapter 4 and elsewhere, and we see that it is guaranteed to branch voltage. It follows that the KVL mesh equations written in terms of mesh currents
work for every circuit. I constitute an equal number of independent equations as unknowns and thus may be solved
unique~y for the mesh cun:ents. All branch currents, and hence branch voltages, may be
determilled from the resultlllg mesh currents. Mesh analysis is thus guaranteed to suffice
for complete analysis of any planar circuit. 2
The relative efficiency of mesh analysis depends on how many equations it requires
(compared to ~ - 1 nodal equations and B - N + 1 basic loop equations). Suppose we
Turning attention from voltages to currents, for any network graph with a given tree, the reconstruct a gIven planar network one mesh at a time. The first mesh has as many
set of tree branch currents is fixed by the values of the B - N + 1 link currents. For if nodes as branches, say kl of each. The next, and all subsequent meshes added, will
we set all link currents to zero, since there are no loops in the tree, there are no closed add one less node than branch to the growing reconstruction. Each branch in a new
loops anywhere and all tree branch currents are fixed at zero. Any individual link current 'mesh will also add a new node, except the last branch in the mesh, which returns
can be assigned independently of the others, as can be seen by setting all but one to zero to the previously reconstructed subcircuit. Thus, if k2 branches are added with the
and noting the last may take on any value, since it is part of a closed loop with no other second mesh, k3 with the third, and so on, upon full reconstruction we have M meshes
where '
links in the loop. We conclude that the set of B - N + 1 link currents is independent,
and the tree branch currents are fixed by the link currents. B = kl + k2 + ... + kM
If all but one link is removed from a network graph, there will remain one closed
loop. Call the set of these B - N + 1 loops, one per link, the basic loops associated with
N = kl + (k2 - 1) + (k3 - 1) + ... + (kM - 1)
the given tree. The set of KVL equations written around these loops is independent, since = B - (M -1)
each equation contains a branch not contained in any of the others, that is, its defining
so the number of meshes M is
link. For instance, referring to Fig. C.l(d), with B = 5 and N = 4, the B - N + 1 = 2
links are b20 and b23 . The two basic loops are closed by first inserting b20 into the tree, M=B-N+l (C.l)
and then replacing with b23 These basic loops are 2012 and 23012, enumerated by the
order that nodes are traversed. which is the same as the number of basic loops. We conclude that KVL equations around
Thus the B - N + 1 independent KVL equations around the basic loops, written the B - N + 1 basic loops associated with any tree, expressed in terms of basic loop
in terms of the B - N + 1 link currents, also referred to as basic loop currents, can be ~urrents, or KVL equations around the B - N + 1 meshes in a planar circuit, expressed
In terms of the mesh currents, are linearly independent and sufficient for complete circuit
solved for the link currents. The remaining tree branch currents are fixed by them, and,
given the full set of branch currents, the branch voltages follow from the i-v laws. We analysis.
conclude that for any given tree, KVL written around the B - N + 1 basic loops in terms To illustrate, consider the circuit of Fig. C.2(a). The network graph is shown in
of the B - N + 1 basic loop currents are independent and are sufficient for complete Fig. C.2(b), with a tree colored in green. Associated with this tree are four basic loops:
circuit analysis. This is referred to as basic loop analysis and, just like nodal analysis, is 014530, 0~520, 0120, an~ 3453. The current source lsi reduces the number of required
guaranteed to work for all circuits. loop equatIOns by one, Sillce the basic loop current marked i4 is in fact equal to I sl, a
Just as we found it convenient to convert in the end from tree branch voltages known value. The three basic loop equations, associated with the first three basic loops
to node voltages in our previous discussion of nodal analysis, it is preferable wherever cited above, are
possible to work with meshes rather than basic loops. Meshes are defined only for planar Basic loop 1: (RI+ R2 + R6 + R 7)i l - R7i2 + R l i3 = Vs2 - R6 I si
networks, those networks whose graphs can be drawn on a plane without any branches
crossing. A mesh is defined as an empty closed loop in a planar network graph, that is, a Basic loop 2: - R7il + (R3 + R4 + R7)i2 + R4i3 = - Vs2
loop containing no branches inside it. Mesh current paths can be identified by inspection, Basic loop 3: Rli l + R4i2 + (RI + Rs + R4)i3 = 0
while basic loops must be constructed from a tree.
The set of KVL mesh equations is independent, since each added mesh equation The four meshes and their mesh currents are shown for this planar circuit in Fig. C.3.
brings in at least one new branch (if the network graph contains meshes fully enclosed by There are three unknown mesh currents marked is, i6, and h. Temporary removal of
sets of other meshes, add them before the meshes that enclose them). By their definitions, the current source collapses the two meshes 01430 and 3453 into the single supermesh
1With the understanding that each voltage source defines a supemode that should be counted as a single 2With the u~derstandi.ng that each curr~nt source decreases the number of meshes by one (current sources
node (voltage sources are topologically equivalent to short circuits). Thus a circuit with five nodes and ar~ topologIcally eqUivalent to open cIrcuits). Thus a circuit with four meshes and one current source
one voltage source will have three nodal analysis equations. WIll have three KCL mesh analysis equations.
814 App. C Circuit Topology Section C.3 Basic Loop Analysis and Mesh Analysis 815
Appendix
\9 It ., 1111 III . . . . . . . It til ., "
D
., GIl *'
(a) (b)
This appendix lists the basic input statement formats and usage rules for the family
FIGURE C.2 (a) Circuit; (b) network graph showing tree and basic loop
of circuit simulators referred to throughout this book as SPICE. This family embraces
currents.
many variations, or "dialects," some public domain (SPICE 1, SPICE2, SPICE3) and
014530. The mesh equations are some proprietary (PSpice, HSPICE). The dialects differ somewhat both with respect
to restrictions and enhancements. Where dialects differ in their restrictions, the most
Supermesh: Rli5 + RZ(i5 - i6)+ R6(lsl - is - i6) + R7(i5 - h) = Vsz
restrictive form is listed here. For instance, some dialects permit the first character of
Mesh 6: R5i6 + R3(i6 - h) + R6(i6 - lsI + i5) + RZ(i6 - i5) = 0 the title statement to be white space and others do not, here you are warned not to use
white space. The dialects also differ widely in enhancements of various kinds; some for
Mesh 7: R4h + R7(h - i5) + R3(i7 - i6) = -Vsz
instance offer support for high resolution graphics and printing. For capabilities beyond
Which of these methods, nodal, mesh, or basic loop analysis, should be used to the universal common core discussed here, the user is referred to the documentation of
study a given circuit? If N - 1 B - N + 1, as is the case for circuits with many the specific software program. The syntax summarized here is a common core applicable
elements in parallel (many branches per node), nodal analysis will be more efficient. If to all dialects of SPICE.
B - N + 1 N + 1, as in circuits with many elements in series, and consequently few SPICE 1, which stands for Simulation Program with Integrated Circuit Emphasis
elements per node, either basic loop or mesh analysis would be a better choice. For planar (Version 1), was written by faculty and students at the University of California, Berkeley,
circuits, mesh analysis has the edge over basic loop analysis, since no tree is needed. largely with public grant support, and released into the public domain in 1972. The first
Since most circuits of low and medium complexity, such as we work with in this book, major rewrite, SPICE2, released in 1975, improved accuracy and robustness and expanded
are planar, nodal and mesh analysis are the methods we have emphasized in this book. the set of supported circuit devices. SPICE2 still remains the most widely used member of
the SPICE family in computing environments favoring multiuser batch processing, such
as large university mainframes. The mainframe model of digital computing prevalent
in 1975 gradually gave way to a networked server-workstation model where each user
had a high resolution screen and printer available. Recognizing this opportunity, SPICE3
appeared in 1985 complete with a graphical user interface, point-and-click controls, and
high resolution plots. The popUlarity of the C programing language in network-based
computing led SPICE's authors to abandon FORTRAN, the language of SPICEI and
SPICE2, in favor of recoding in C. While designed for interactive use and incorporating
menus and a graphical user interface, SPICE3 also supports ASCII batch mode operation,
making it backward compatible with earlier releases.
The first SPICE dialect designed for use on personal computers, PSpice, was re-
leased in 1984 by MicroSim, Inc. It remains the most widely used of the proprietary
FIGURE C.3 Meshes and mesh currents for the circuit of Fig. C.2(a).
K 10+3 M 10- 3
MEG 10+6 U 10- 6 The circuit to be investigated is specified by a netlist. A netlist is a list of all the elements
G 10+9 N 10-9 in a network, their types, values, parameters and locations. The circuit cannot contain a
T 10+ 12 P 10- 12 closed loop of voltage sources or a closed region into which only current sources flow.
F 10- 15 Each element in the circuit is completely specified by a single netlist statement. The
formats for netlist statements by element type follow.
For instance, the following are all equivalent numerical fields:
lThe restriction that only currents through voltage sources may be declared is not enforced in all SPICE
1.05E6 1.05MEG 1.05E3K 0.00105G dialects. Some permit forms such as I(RI). Consult your documentation.
818 App. 0 SPICE Reference Guide Section 0.3 Netlist Statements 819
Resistor c denotes a lO-mH inductor labeled 12 connected between nodes 100 and 0 (reference
node). If IC = -0.5 is included, then an initial current of -0.5 A from node 100 through
R<NAME> +) NODE> -) NODE> <VALUE> the inductor to node 0 exists at t = O. This is used only in the transient response.
(NAME) denotes any 1- to 7 -chapter alphanumeric string for labeling the element.
(+) and ( -) NODES define the polarity of the resistor connection. Positive current Linear Transformer
flows from (+) NODE through the resistor to (-) NODE.
(VALUE) is the nonzero resistance value (positive or negative) in ohms. K<NAME> L<INDUCTOR NAME A> L<INDUCTOR NAME B> <COUPLING VALUE>
Example D.l: RLOAD 12 3 10K K(NAME) couples two inductors A and B using a dot convention that is determined
by the node assignments of inductors A and B. The polarity is determined by the order of
the nodes in the L devices and not by the order of the inductors in the K statement. The
denotes a lO-kQ resistor labeled RLOAD connected between nodes 12 and 3 having
Aotted terminals of inductors A and B are those connected to the positive (first labeled)
current flowing from positive node 12 through the resistor to negative node 3.
nodes in their defining statements.
(NAME) denotes any 1- to 7-character alphanumeric string.
Capacitor (COUPLING VALUE) is the coefficient of mutual coupling in the range 0 :s k < 1.
C<NAME> +) NODE> -) NODE> <VALUE> [IC = <INITIAL VALUE>] Example D.4: LPRI 2 3 500M
LSEC 5 4 400M
(NAME) denotes any 1- to 7-character alphanumeric string. KXFRM LPRI LSEC 0.98
(+) and (-) NODES define the polarity of the capacitor connection. Positive
current flows from (+) NODE through the capacitor to (-) NODE. denotes a linear transformer having a mutual coupling of 0.98 between LPRI and LSEC.
(VALUE) is the nonzero value (positive or negative) in farads. Dotted terminals for the polarity of the coupling are the terminal of LPRI connected to
(INITIAL VALUE) is optional and denotes the initial capacitor voltage at time node 2 and the terminal of LSEC connected to node 5.
t = 0 of (+) NODE with respect to (-) NODE for a transient response analysis.
Independent Source
Example D.2: CEXT 2 3 lOU
CEXT 2 3 lOU IC=4
I<NAME> +) NODE> -) NODE> [TYPE <VALUE>] [TRANSIENT SPEC.]
denotes a 10-tiF capacitor labeled EXT connected between nodes 2 and 3. If IC = 4 is V<NAME> +) NODE> -) NODE> [TYPE <VALUE>] [TRANSIENT SPEC.]
included, then an initial voltage of 4 V for node 2 with respect to 3 exists at t = O. This
I denotes independent current source.
is used only in the transient response.
V denotes independent voltage source.
(NAME) denotes any 1- to 7 -character alphanumeric string.
Inductor (+) and (-) NODES define the polarity of the source. Positive current flows from
(+) NODE through the source to (-) NODE (the reference arrow points to the (-)
L<NAME> +) NODE> -) NODE> <VALUE> [IC = <INITIAL VALUE>] NODE).
TYPE is DC (default) for a dc source and ac for a sinusoidal ac source.
(NAME) denotes any 1- to 7 -character alphanumeric string. (VALUE) is a dc value for dc or a magnitude and phase (in degrees) for ac. Default
(+) and ( -
) NODES define the polarity of the inductor connection. Positive current values are zero.
flows from (+) NODE through the inductor to (-) NODE. [TRANSIENT SPEC.] is used in transient analysis only and can be one of the
(VALUE) is the nonzero value (positive or negative) in henrys. following:
(INITIAL VALUE) is optional and denotes the initial inductor current at time t = 0
form (+) NODE through the inductor to (:-) NODE for a transient response analysis. EXPxl> <x2> <tdl> <tel> <td2> <te2
Example D.3: L12 100 010M The EXP form causes the output current or voltage to be (xl) for the first (td1)
L12100 0 10M IC=-0.5 seconds. Then, the output decays exponentially from (xl) to (x2) with a time constant
820 App. D SPICE Reference Guide Section D.3 Netlist Statements 821
of (tel). The decay lasts (td2) seconds. Then, the output decays from (x2) back to (xl) (NAME) denotes any 1- to 7-character alphanumeric string.
with a time constant of (tc2). (+) and (-) NODES define the polarity of the source. Positive current flows from
(+) NODE through the source to (-) NODE.
PULSExl> <x2> <td> <tr> <tf> <pw> <per (+ CONTROLLING) and (- CONTROLLING) are in pairs and define a set of
controlling voltages that are multiplied by (GAIN). A particular node may appear more
The PULSE form causes the output to start at (xl) and remain for (td) seconds. than once, and the output and controlling nodes need not be different.
The output then goes linearly from (xl) to (x2) during the next (tr) seconds. The output The plus sign on the second line of this statement indicates continuation.
remains (x2) for (pw) seconds. It then returns linearly to (xl) during the next (tf)
seconds. It remains at (xl) for (per) - (tr) + (pw) + (tf) seconds and the cycle repeats,
excluding the initial delay of (td) seconds. Example D.6: EBUFFl 3 11 9 2.4
PWLtl> <xl> <t2> <x2> ... <tn> <xn fenotes a VCVS labeled BUFF connected between nodes 1 and 3, with the voltage of
node 1 with respect to 3 equal to 2.4 times the voltage of node 11 with respect to 9.
The PWL form describes a piecewise linear waveform. Each pair of time-output
values specifies a comer of the waveform. The output at times between comers is the Current-controlled Current Source (CCCS)
linear interpolation of the current at the comers.
F<NAME> +) NODE> -) NODE> CONTROLLING V DEVICE) NAME> <GAIN>
SINxoff> <xampl> <freq> <td> <itc
The SIN form causes the output to start at (xoff) and remain for (td) seconds. Then, (NAME) denotes any 1- to 7-character alphanumeric string.
the output becomes an exponentially damped sine wave described by the equation (+) and (-) NODES define the polarity of the source. Positive current flows from
(+) NODE through the source to (-) NODE. The current through (CONTROLLING
xoff + xampl . sin{2n . [freq .(TIME-td)]} . e-(TIME- td) * itc V DEVICE) NAME) multiplied by (GAIN) determines the output current. Positive
controlling current flows from the (+) NODE of (CONTROLLING V DEVICE) to the
(-) node.
(CONTROLLING V DEVICE) is an independent voltage source.
Example D.5: IG12 3 0.2A or IG12 3 DC 0.2A
denotes a dc current source labeled G 1 supplying 0.2 A from node 2 to 3 through the Example D.7: F23 3 7 VOUT 1.2
source.
denotes a CCCS labeled 23, with current flowing from node 3 through the source to node
ISOURCE 0 5 AC 10 64 7 having a value equal to 1.2 times that of the current flowing in independent voltage
source VOUT positive current (from + terminal to - terminal).
denotes a 10/64 A ac source labeled SOURCE connected between nodes 0 and 5 with
the positive terminal to node O.
Voltage-controlled Current Source (VCCS)
12 0 2 EXP(lOM OM 0 01 1)
G<NAME> < (+) NODE> < (-) NODE> < (+ CONTR.OLLING ) NO.DE>
denotes an exponential current of We-lOt rnA flowing from node 0 to 2 through the + - CONTROLLING) NODE> <TRANSC~NDUCTANCE>
source in the interval 0 < t < 0.1 s for transient analysis.
(NAME) denotes any 1- to 7-character alphanumeric string.
(+) and (-) NODES define the polarity of the source. Positive controlling current
Dependent Sources
flows from (+) NODE of the (CONTROLLING V DEVICE) to (-) NODE.
Voltage-controlled Voltage Source (VCVS) (+ CONTROLLING) and (- CONTROLLING) are in pairs and define a set of
controlling voltages that are multiplied by (TRANSCONDUCTANCE).
E<NAME> +) NODE> -) NODE> + CONTROLLING) NODE>
+ -CONTROLLING) NODE> <GAIN> Example D.8: GAMP 4 3 1 9 1.7
822 App. D SPICE Reference Guide Section D.3 Netlist Statements 823
denotes a VCCS labeled AMP connected between nodes 4 and 3, with the current flowing statements. The output control statements .PLOT and .PRINT will be described in the
from node 4 to 3 through the source equal to 1.7 times the voltage of node 1 with respect next section.
to 9.
(NAME) denotes any 1- to 7-character alphanumeric string. denotes an ac steady-state solution for a network having a frequency of 100 Hz.
[NODE]* denotes a list of nodes required by the subcircuit definition.
(SUB CIRCUIT) NAME} is the name of the subcircuit's definition (see .AC LIN 101 100KHZ 200KHZ
.SUBCKT statement below). There must be the same number of nodes in the call as in
the subcircuit's definition. denotes a linear frequency response having 101 points evenly distributed in the range
This statement causes the referenced subcircuit to be inserted into the circuit, with from 100 to 200 kHz.
the given nodes replacing the nodes in the .SUBCKT definition in order. It allows
defining a block of circuitry once and then the use of that block in several places.
.DC
Example D.lO: XBUFF 4 1 7 9 UNITAMP DC analysis: The .DC statement causes a DC sweep analysis to be performed for the
circuit. It has the form
denotes a call to a subcircuit that replaces the call statement by the content of the file
UNITAMP for SPICE analysis. .. DC SWEEP VARIABLE) NAME> START) VALUE> END) VALUE>
+ INCREMENT) VALUE>
(SWEEP VARIABLE) NAME} is a name of an independent current or voltage source.
It is swept linearly from (START) VALUE} to (END) VALUE}. The increment size
is (INCREMENT) VALUE}. (START) VALUE} may be greater or less than (END)
All statements beginning with a period (.) are control statements. The commands to be VALUE}; that is, the sweep may go in either direction. (INCREMENT) VALUE} must
discussed in this section are the .AC, .DC, .FOUR, .IC, .LIB, .SUBCKT, .TF, and .TRAN be greater than zero. The entire sweep may specify only one point if desired.
824 App. D SPICE Reference Guide Section DA Solution Control Statements 825
Example D.l2: . DC VIN 10V 10V lV .L1B
Library file: The .LIB statement is used to reference a subcircuit library in another file.
denotes a de solution for a circuit with independent voltage source VIN = 10 V. It has the form
denotes a dc sweep for independent current source IGEN being swept from 1 to 10 rnA ((FILE) NAME) is the name of the subcircuit library file.
in I-rnA steps.
Example D.15: . LIB OPAMP. LIB
.FOUR ,denotes that the subcircuit library file is named OPAMP.LIB.
Fourier analysis: Fourier analysis performs a decomposition into Fourier components
as the result of a transient analysis. A .FOUR statement requires a .TRAN statement
(described below). It has the form .OP
Operating (bias) point dc analysis: The .OP statement calculates all the dc node voltages
.FOUR FREQUENCY) VALUE> OUTPUT VARIABLE* and the currents in all voltage sources. It has the simple form
((OUTPUT VARIABLE)) is a list of one or more variables for which the Fourier com- .OP
ponents are desired. The Fourier analysis is done by starting with the results of the
transient analysis for the specified output variables. From these voltages or currents, the
dc component, the fundamental frequency, and the second through ninth harmonics are .SUBCKT
calculated. The fundamental frequency is ((FREQUENCY) VALUE), which specifies
the period for the analysis. The transient analysis must be at least 1/((FREQUENCY) Subcircuit definition: The .SUBCKT statement is used to define a subcircuit that is called
VALUE) seconds long. using the X statement described previously. It has the form
.SUBCKT<NAME> [NODE]*
Example D.13: .FOUR 10KHZ V(S) V(6, 7) I(VSENS3)
The .SUBCKT statement begins the definition of a subcircuit. The definition is ended
yields the Fourier components for variables V(5), V(6, 7), and I(VSENS3). The funda- with a .ENDS statement. All statements between .SUBCKT and .ENDS are included in
mental frequency for the decomposition is set equal to 10 kHz. the definition. Whenever the subcircuit is called by an X (subcircuit call) statement, all
the statements in the definition replace the calling statement.
.IC
.TF
Initial transient conditions: The .IC statement is used to set initial conditions for transient
analysis. It has the form Transfer function: The .TF statement produces the dc small-signal transfer function.
It has the form
.IC <VNODE = <VALUE*
.TF OUTPUT VARIABLEY> INPUT SOURCE) N~ME>
Each (VALUE) is a voltage that is assigned to (NODE) for the initial node voltage at
time t = 0 for the transient analysis. The gain from ((INPUT SOURCE) NAME) to ((OUTPUT VARIABLE)) is calculated
along with the input and output resistances. ((OUTPUT VARIABLE)) may be a current
or a voltage: however, in the case of a current it is restricted to be the current through a
Example D.14: .Ie V(2)=S V(S)=-4V V(101)=10 voltage source.
denotes setting the initial node voltages of nodes 2, 5, and 101 to 5, -4, and 10 V,
respectively, at t = O.
Example D.16: .TF V(3) lIN
826 App. D SPICE Reference Guide Section D.4 Solution Control Statements 827
produces the dc small-signal transfer function for V(3)/lIN, the input resistance seen
The range and increment of the x-axis are fixed by the analysis being plotted. The
from the terminals of independent current source lIN, and the output resistance seen at
node V(3). range of the y-axis can be set by adding ((LOWER LIMIT) VALUE), (UPPER LIMIT)
VALUE) to output variables with the same y-axis range. Each occurrence defines one
y-axis with the specified range. All output variables that come between it and the next
.TRAN range to the left are put on its corresponding y-axis. If no y-axis limits are specified, the
program automatically determines the plot limits.
Transient analysis: The .TRAN statement causes a transient analysis to be performed for
the circuit. It has the form
Example D.18: .PLOT DC V(2) VO, 5)
.TRAN PRINT STEP) VALUE> FINAL TIME) VALUE> [Ule]
plots the dc response for V(2) and V(3, 5),
The transient analysis calculates the behavior of the circuit over time, starting at t = 0
and going to (FINAL TIME) VALUE). (PRINT STEP) VALUE) is the time interval .PLOT AC VM(3) VP(3) IR(Vl) II(Vl)
used for plotting or printing the results of the analysis. The keyword mc (use initial
conditions) causes the initial conditions set for capacitors and inductors with the IC plots the magnitude and phase of V(3) and the real and imaginary parts of the current
specification to be used. through VI.
.PRINT
Print: The .PRINT statement allows results from dc, ac, and transient analysis to be
output in the form of tables. It has the form
The output control statements for plotting and printing are the .PLOT, .PRINT, and .PRINT [DC][AC][TRAN][(OUTPUT VARIABLE)]*
.WIDTH statements.
DC, AC, and TRAN are the analysis types that can be output. Exactly one analysis type
must be specified. [(OUTPUT VARIABLE)]* is a list of output variables desired (see
.PLOT
.PLOT statement above). There is no limit to the number of output variables. The output
The .PLOT statements allows results from dc, ac, and transient analysis to be output in format (characters per line) is determined be the specification of the .WIDTH command.
the form of line printer plots [for example, Fig. 14.44(a) or (b)]. It has the form
Example D.19: . PRINT DC V(1) !( V12)
PLOT [DC][AC][TRAN][OUTPUT VARIABLE]*
+ ([LOWER LIMIT) VALUE>, UPPER LIMIT) VALUE>J) prints the dc values for V(I) and I(V12).
DC, AC, and. TRAN are the analysis types that can be output. Exactly one analysis type .PRINT AC VM(1,5) VP(1,5) IR(V2) II(V2)
must be specIfied. [OUTPUT VARIABLE]* is a list of the output variables desired for
plotting. Currents are specified in the form I(VNAME), where VNAME is an independent prints the magnitude and phase of V(1, 5) and the real and imaginary parts of I(L2).
voltage source and I(VNAME) flows through the voltage source from its + terminal to its
- terminal. For ac analysis, the suffix M, P, R, or I can be added to the variable specifiers
.PRINT TRAN V(7) I(VCC) V(3,1)
V or I. These indicate magnitude, phase, real, and imaginary parts. A maximum of eight
output variables are allowed in one .PLOT statement.
prints the transient response for V(7), I(VCC), and V(3, 1).
End statements for subcircuits and circuit files are .ENDS and .END, respectively .
. ENDS
2.19. ,J2
End of subcircuit definition: The .ENDS statement marks the end of a subcircuit definition 2.21. V = -4i; i = -~v
(started by a .SUBCKT statement). It has the form 1.1. 456 MJ 2.23. (a) 800 n, 780 n; (b) 720 n, 702 n
1.3. 3529 kmJhr 2.25.40 V
.ENDS [(SUBCIRCUIT) NAME] 1.5. 1116 s 2.27. 5 n; 10 n
1.7. -4 C; 2 A; -3 A 2.29. 4 resistors; 240 resistors
1.9. 96 mW; 92 J.LC; 736 J.LJ 2 31 205 A 205 W
It is good practice to repeat the subcircuit name, although this is not required. 36 '9
1.11. -11.8 W 2.35. 10 kn; 60 rnA
1.13. -4(t 3 - 15t 2 + 50t) W; 0 J 2.37. 4 resistors; 200 resistors
.END 1.15. ~et A 2.39. VT = -6 V, RT H = 6 n
End of circuit: The .END statement marks the end of the circuit file. It has the form
1.17. -l J 2.41. Voltage across 1-n resistor, current through it.
1.19. V-source supplying; "3J V 2.43. VT = Vs; RTH = 0 n
1.21. (a) 14.4 kJ; (b) 1.2 kC 2.45. 96 kn
.END 1.23. 10 S; 6 J 2.47. in = 1; A; RN = ~ n
1.25. P odd 2.49. VT = ~ V, RTH = RN = ~ n, iN = ~ A
1.27. Never 2.51. RN = 00
1.29. 100; 43 2.53. All in series: 14, 4 copies of 10 in parallel, 7 copies
1.31. passive of 100 in parallel.
1.33. passive 2.55. VT = 28 V, RTH = RN = 8 n; iN = ~ A
!
2.57. (a) 6i2 + 24; (b) sint + ~ A, -~ sint + 8 V
2.59. VT = 11 V, RTH = RN = n, iN = A fI
2.1. il - i4 - i6 = 0; i2 + i4 + i5 = 0; -i3 - i5 + i6 = 0
2.5. (a) -V5 - V2 - VI + V3 + V4 = 0; (b) V5 + V2 + VI -
V3 - V4 = 0; (c) V3 + V4 = VI + V2 + V5 3.3. -1 S
2.11. 12 kW 3.5.3 A
2.13. 24 V; 12 V; -20 V; -32 V 3.7. -~ sin2t V
2.15. 10 V; -10 V; -10 V; -10 V
2.17. Vo = 5sin2t V; VI = ~ sin2t V;
3.9. VT = 8 V, RT = n 1
3.11. VCVS -1; CCVS -10
1O2V
sm t ; 10 = 11. = 612A
V2 = V3 = "3 sm t ; 3.13. -~ A
.12 = 912A
sm t ; 13 = 18 12A
sm t 3.15. 1 A
832 App. E Answers to Selected Odd-Numbered Problems App. E Answers to Selected Odd-Numbered Problems 833
9.7. VT = 24.7/-134 0 , ZT = ZN = 3 + j12 n, IN = 10.45. -~oo n 12.11. (a) ~(l- e- 2s ) + ~~~, Re(s) > -1; 13.7. VT = s2t:2+2, ZT = ~(3s + 1)
2/1'50 0 A
9.9. -.062 - j.012 n
10.47. (a) 0.943 lagging; (b) 93.7 fLF
1.62 Mn (b)
3e-4,
sz' 2n
10
Re(s) > 0, (c) s+l- s2+(2n)2' Re(s) > -1 (!
13.9. 1 n in series with H in parallel with 1 F)
10.1. 5~ kW 11.31. 2530/23 n, 4.7 kW 13.45. 0.21 eos(t - 143) A, 0.052eos(t + 129) V
(d) 3e- t + 2 sin t; (e) e- t - e- 3t - 2te- 3t ;
10.3. 500 W 11.33. 15v2 A rms (1) e- t (1 + 2t) - e- 3t (l + 4t) 13.49. -1,0
10.5. ~Rl;' 11.35. laA = 7.97 - j0.48 Arms, IbB = 10.8 - j1.83 A 12.35. (a) 2e- t - e- 2t ; (b) e t - 2e- t - 1; 13.51. Let (nl' d l ) be orders of numerator, denominator
!
10.7. W, independent of . rms, IcC = 2.86 + j2.31 Arms, 2280 W
11.37. 14 + 14,J3j n, -,J3 kVAR
(c) 6 [8(1)(t) + 8(t) + e- t ] in Z(s), (n2, d2) in Vs(s). Then dl - nl < d2-
n2 -1
10.9. -90 12.37. (a) - ~te-(3/2)t + ~e-(3/2)t; (b) [1 - 7t] e- 3t ;
10.11. 138 W 11.39. (a) 6 + 3j n, (b) 22~j21 n, 6 - j n, 6 - j n 13.53. 2(e- t - e- I )u(t - 1)
(c) [~t2 - 14t + 7] e- t 13.55. Transforms of both are sF(s)G(s).
10.13. (a) 2v2 A; (b) 2,J3 A; (c) ~ A; (d) Jzlm A 11.41. 111/14.80 A, 111/-178.5 A, 0 A
12.39. (a) _6e- 9t A
10.15. PR = 13.7 W, etc. 11.43. n 12.41. (a) cost; (b) sint; (c) (3 - t)e- t - 3e- 2t
13.57. 2s~2~il), 48' (t) - 68(t) + 12e-2t u(t) V,
10.19. 63.2 fLF 11.45. l11~oA, 111/-178.5 A, 0 A 12.43. 1 + 3e-4t (-4e- t + 12e- 2t )u(t) V
10.21. QLJ = 2.5 X 10-4 VA, QL2 = 2.5 X 10-4 VA, etc. t - ~ [1 - e-(5/6)t] , t + ~ [1 - e-(5/6)t] 13.59. e- 2t (1 - cos t)u(t)
11.47. Ilcci = 100 A (50v2 A rms) for R = 6.2 n 12.45.
10.23. PlkQ = ft; mW, P2kQ = -b
mW, PL = Pc = 0, 11.49. 350, 60 1247. 4s s3-3s s2_3 13.61. e- t (2 + t - !t
2) A
Pv = -fr, mW
(a) s4+6s2+25; (b) sC6s2+25; (c) (s2+9)(s2+1) 13.63. [.318e- 18st eos(1.56t - 57.9) +
11.51. Standard mesh equations.
10.25. 4 V 11.53. 188 n 12.49. 1 !
[e t - e- t ] + [cos t + eat eos(f3t + 45)+
.388e-44t eos(.093t + 48.6 0 ) ]
10.27. SI = -2 - j~; Sv = 2 - j120
10.29. 1.92 W, 15.4 W, 9.6 W
e- at eos(f3t - 45)] , a = f3 = l'
10.31. PIde = 8 kW, PIae = 479 W, PVae = -465 W
10.33. (a) Set ~iL = 0, ~;L = 0; (b) Set ~;L = 0
10.35. -1 s L L L
12.1. (a) S~9 - s~1 + S~I' (b) same as (a), (c) .114
10.37. ZL =4.19/24.8,RL circuit 12.3. No, not of exponential order. 4+16s 2+8 n 14.1. (a) _3_/30 - tan- 1 w; (b) ~/120 - tan- 1 !
10.39. "ES; = 0 + jO
13.1. 2s s3+6s -Jw2+1 -Jw2+25 5
12.5. t~1 not pw continuous 3+6s 2+30s+5 S
10.41. L!-..< 0 each case 12.7. tlOe-lOt . t* = 35.8 s 13.3. 12s 40s 2+20s (c) [(30-J1 + ( 2) / (-J4 + w211 - (21)]
2+3s+2)
10.43. ~ H 12
9 4.89s+6.66
s2+9 13.5. -(2s
2s 3+3s 2+6s
/30 - tan- 1 w - tan-I 'I -/ (1 - ( 2)
834 App. E Answers to Selected Odd-Numbered Problems App. E Answers to Selected Odd-Numbered Problems 835
-ju} 15.33. 4(1 - e-(6/19)t)u(t) V 16.45. Dependent variable pair is the superposition of a de- 17.21. (a) Break into two integrals, use symmetry;
14.3. (4-3w2)+ jZw
15.35. 2.67/-47.7 pendent variable pair plus new term due to sources. (b) Wo = ~, bn = 0, aZn-l =
14.5, w2+1 t -1 -1 (W-u} )
w6+Zw4-3w2+1' an W - tan l-Zw2 15.37. VI = 50cos377t V, il = 0.2cos(377t - 89.6) A, To find new term, set both independent variables to _ 4 _ [2 sin (Zn-l):n: _ 2 sin (Zn-l):n: +4(_l)n+l]
14
7 J w 2 +1 t
",2+4' an
-1
W -
t an -1 2"
w
Vz = 0.2 cos(377t - 89.3) V, iz = .033 cos(377t +
90.7) A
zero (open or short the ports), leave internal sources
in, compute dependent variables. They equal the
(Zn-l):n: 6
17.23. an, bn are all the same; Wo for get) is
3
C times Wo for
f(t).
14.9. s2~1' w~ll' -~cos2t V, no. 15.39. 761 H new terms. _ _ (l_n 211'2) -e-" (n 211'"+3)
14.13. (a) 9 dB; (b) ~ dB; (c) 50 dB; (d) 18 dB; 15.43. w::: !L1lf + !Lz1r - JLIL21112 = 17.25. Wo - n, an - (I + n"11'2)2
(e) -72 dB; (f) 612 dB; (g) 2 dB; (h) -18 dB !(JI\11 + ./"L2h)2 ::: 0 b - (-2n11') (I+e- 2[2-n 211'"])
n- (I + n2~)2
14.15. (a) .05, (b) 20; (c) 5.)2 (d) .01.J5; (e) .005.J5; 15.45. 1.21/-179.10 A 17.27. (a) Wo = ~, Cn = nf sin n;; (b) Wo ~, Cn -
(f) v'lO x 1O-z0
-.L sin n:n:. (c) C = - C = 0 (n even) C = _1_
14.17. (a) 121.62; (b) .254; (c) 7.08; (d) 3.16 x 108 , 17.1. (a) Tp = 4, (b) Tp = 10, (c) Tp = 2 n:n: Z ' 0 2' n ' n In.,,
(n odd)
(e) .0216, (f) 1.0001 !
17.3. fe(t) = [J(t) + f(-t)] = 0 implies f(t) = - f(t) 17.29. Same for all n.
14.19. zi:~!~;), triple real pole with break frequency 1 radls, (odd symmetry). Similar\ly fo(t) = 0 implies f(t) = 17.31. (a) f(t) even; (b) f(t) odd
zero with break frequency 112 radls. +f(-t). 17.33. Evaluate (16.25).
Is
14.21. s-:3/4' 3/4 radls
16.1. 2.31 A, .93~ A, 8.91 V, -5.62 V 17.5. (a) !
[(r(t + 1) + r( -t + 1 - 2(r(t) + r( -t +
4
17.35. (a) Co = + 12, Cn = (Zn:n:)/I!j nWo); (b) Co = : +
14.25. 18 dB, -38 dB, -38 dB, -20 dB, -20 dB 3 [ s+1
16 s+1 ] 2(r(t - 1) + r( -t - 1 - 2(r(t - 2) + r( -t - 2 +
3 Zs2+6s-1 !
(r(t - 3) + r( -t - 3))], [(r(t + 1) - r( -t + 1- 12, Cn = :n:(l-4n2~11+jnWo) (n =1= 0); (c) Co = -:;; +
14.27. WI = 25.1 rad/s, Wz = 631 rad/s, W3 = 15.8 krad/s, s(s+l)
W = 398krad/s (Sx 1O-7 )(s+1)2(s+w2f(s+w4)2 16.5. Solve z-parameter equations for 11 and Vz. 2(r(t) - r( -t + 2(r(t - 1) - r( -t - 1- 12 , Cl = 4(j-jA -A -
wo+l)' CZn = :n:(4n2-1)(jZnwo+l)' CZn-l -
4 , S(S+Wl)2(s+W3)2 16.7. Solve z-parameter equations for VI and Iz. 2(r(t - 2) - r( -t - 2 + (r(t - 3) - r( -t - 3]; o for n > 1
14.29. (s1~/3) 16 9
1 [ S+2 -s] Zs-1
4 2 - 4s s + 1 ' Zs+1O
(b) sinz 2n, 0;
17.37. i+~ :22:1~cos(tan-l~-tan-l~)-
. 1s2 (c) 3.83 cos 2t -1.17 sin 2t (d) 3.83 cos 2t [u(t - 1)-
14.31. (s+1.47)(si+. Z79s+.34)
14.39. Series: 1 H, 400 pF, 1250 D.; parallel: 1 H, 400 pF, 16.11. [~:: ;:~ ] u(t + 1)], -1.17 sin2t [u(t - 1) - u(t + 1)]
1 1 .fi'(b) 1 1 -./S.() 1 1.fi
2 9:n: 2+4 cos(3Jtt + tan- 1 3:n: -tan-1 3:n:)
3:n: 9:n: 2+16 Z Z 4
177 ()
2Mn 16.13. VI = t, V2 = ~, .. a 2' 2' 2' 5' 5' 5' c 2' 2' 2
17.9. No. f(t) possible if and only if Tz/Tl rational
17.39. n odd: An = (n:n:\~+4; n even: An = 0
14.41. Ao Wo AowoRA rad/s t -1 n:n: 64
I = 5,3 + 19,2 + 14, + 4 =-,( 5,a + 2) (.1 = 9s2 + 10 1741 . ~,- 8
J
'w+ AO"'ORA ' RA +RF
RA+RF I +4 number. v.. 4+ (!'Jf)2 an T' 4+(n:n:)2
I sa ' 2 S 00
8 16
14.61. 10.02, 10.02
s+3 _! ] [(ZS+3)(S+I) - (4s+3) ] 17.11. ~ + 2sin2(t +~) - ~ L 4nLl cos4n(t +~) 17.43. 11'" 911"
16.15. [ 6s 3 Zs2 s n=1 1 [jaw -jbw]. (b) a+j{J) . () b .
+-31 4s+3'
3S (s+3)
----y,- 3 00 ()
1745 a jw e - e , (a+ jw)2+b2' c (a+ jw)2+b2 ,
17 13 ! _ .2. " sin(Zn-l)Jtt
(d) l-e-2(~+j",)
Z :n: L." (Zn-l)
1 [2(SZ + s + 1) 2 ] n=l a+Jw
16.17. s(Zs2+Zs+3) 2 2(2sz + 2s + 1) , 00
17.47. ~ [sin4w - sin2w]
17.15. -6 L cos(2n - 1)m
15.1. ('it
Z
+ 1) x 10-4 W ' 'iZ X 10-4 W '
Z(s2+s+1) Z(Z2+Zs+1)
s(Zs2+Zs+3)' s(Zs2+Zs+3) n=1 17.49. lim
L\.-O
[1
sin ~w] = 1
(3t + ~) x 10-4 W, (~t + 4) x 10-4 W 16.19. Z21 -h21 1717 :n: b 0 [(nwo)3_(nwo)2-Z] .
Zit' hllh22-h12h21 Wo = 2"' n = , an = (nWo)3 sm nwo - 17.51. (a) -6e+ 3t u(-t); (b) _6e- 3Itl ; (c) (e t - e 3t )u(-t)
15.3. ~ H 1753 1 3+jw
16.23. Fig. 16.13 with Zll - ZIZ = sz--zz
+ s+ z' ZZZ - ZIZ = '.fi S-w 2+jZw
[ (n!)2 ] cos nwo
15.5. 0.9t + 1.05 W, 0.45t Z + 1.05t J 1 s 17.57. IF(jnn)1 = 0.565(n=0), 0.262(11=::t: 1),
s2+Zs+Z' ZZZ - ZZI = s2Zs+Z' ZIZ = s 00
15.7. 2 H, 0 H 16.25. Symmetric n-circuit with 6 n across each port, con- 1719. () 'i + 1 (b) w = 1
"sin(Zn-l)t 0.136(n = 2), etc.
15.9. 4(e- t - e- 3t ) A nected by 1.85 n.
a Z :n:
n=1
L." (Zn-1)'
n
0 ,
!
17.61. (a) 2X(jw) + [X(j [w - woD + X (j [w + wo])];
15.11. 0.174 cos(7t - 56.3) A _ (-I)n(e"-e-") b _ (-1) +l(e"-e") (b) 2X(jw)
16.27. (a) Apply KCL at input, output ports, (b) apply KCL an - :n:(I+n2) , n - :n:(I+n2) ,
15.13. Sf sin(2t + 135) V at input, output ports. (c) Wo -- n, an -- 4(_I)n b - 0
(n:n:)2' n-
15.15. 1250 W, 0.6 lagging 1629 [ .706 .765]
. . 1.00 .588
15.17. (a) 2e- Zt V; (b) z f cos(8t - 45) V
15.19. -2e- t V 16 33 [ -5.70 -9.00]
1.00 1.33
15.21. 35.9 cos(t + 152) A
15.23.
15.25.
2/-67.3~ 1/-157~ .343/-59.0
35.9cos(t + 152) A
16.35. 10-4 [:ilo ~ ]
s+IQ -17
YII = -1.5 YI2=4.79xl0 L-.044 ]
15.27. -2 [
16.37. Yn = 0.75 L 179.9
Y2I = 1.l1xlO- 16 L -0.22
15.29. .)2e-t cos(t + 45) - e- t A 16.43. Nodal analysis with negative terminal of VI as ref-
15.31. (a) Vz = - 8Vl, /z = kh, Vz = -8Vl, iz = kil; erence yields Yll = YIZ and YIZ = YZI. Inverting
(b) Vz = -~ VI, /z = ~h, Vz = -~Vl' iz = ~il yields the z-parameters.
836 App. E Answers to Selected Odd-Numbered Problems App. E Answers to Selected Odd-Numbered Problems 837
A British System of Units, 3
building block circuits, 753
ac. See Current, alternating Bunsen, Robert, 117
admittance, 340
matrices, 673 c
y-parameter, 658
ammeter(s), 61-62
candela, 3
d' Arsonval meter, 61
capacitance, 175
ideal,61
capacitor(s), 2, 174
Ampere, Andre Marie, 4,21, 185
discrete, 196
ampere, 4, 6, 21
energy storage in, 179
amplifier(s), 81. See also Operational amplifier(s)
linear, 175
ideal voltage, 84-85, 93, 99, 100
parallel-connected, 181-184
inverting, 93, 585, 753
parallel-plate 174
noninverting, 95, 96, 753
practical, 196
series and parallel, 181, 183
B
series-connected, 181-184
terminal law for, 175
bandwidth, 580 voltage-current relations for, 336
unity gain, 585 cascade interconnection, 98
basic loop analysis, 814-816 parallel, 672
battery, electric, 1 series, 674
bel,564 of two-ports, 670
Bell, Alexander Graham, 647 cathode ray tubes (CRTs), 8
block diagram, 760, 770 Cavendish, Henry, 30
Bode, Hendrick, 566 chain(s)
Bode gain plot, 567-570, 582, 752 of series current sources, 41
corrected, 569 of series elements, 37
for inverting amplifier, 586 of series resistors, 38
uncorrected, 569 of series voltage sources, 40
Bode phase plot, 567, 576 characteristic equation, 213
break frequency, 569 characteristic exponents, 278
839
~-----~~-- ---- --
9
charge, 46 second-order, 273, 303 direct, 8 first-order, 209
conservation of, 24, 177 series RLC, 273, 578, 581, 582 divider, 44 forcing term in, 222
electric, 4, 5 short, 32, 146 division, 44, 361 Dirac, Paul, 490
circuit(s), 23. See also Subcircuits simple, 37 electric, 1, 6 Dirichlet conditions, 699
analog computer, 304 single-node-pair, 42 full-scale, 61 discrete phase plots, 714
analysis of, 15-16 singular, 197-199 gain, 81 discrete spectra, 714
ac steady-state, 361 stable, 349 leakages, 196 dot convention, 616
computer-aided, 155-165 three-phase, 450-456 mesh,139 double-subscript notation, 447
building block, 753 and SPICE, 470-472 phase, 454 dual equations, 146
closed-loop, 84, 91 topology of, 29, 811 reference direction, 7, 11 duality, 145, 147-148
with current sources, 142 two-port, 98, 649 sensitivity, 63 of capacitor and inductor, 187
with dc sources, 222 with two storage elements, 273 short-circuit, 51 property, 728
with dependent sources, 82 unstable, 349 sinusoidal, 8
design of, 15-16, 255-261, 303-306 voltage follower, 90 terminal, 33 E
physical, 16 coefficient of coupling, 615 current-divider principle, 44, 45
schematic, 16 coil(s), 610 current sources, 13, 142 Edison, Thomas Alva, 359
electric, 2 coupled, 611, 616 controlled, 80 electric field(s), 26, 178
elements of, 3, 12 comment statements, 819 current-controlled (CCCS), 81 electric circuit, 2
equivalent, 35-36 complementary solution, 275 dependent, 80 electrical isolation, 628
first-order, 208, 218, 219, 221, 229, 255 complex number(s), 323, 803. See also Phasors element statements of, in SPICE, 159 electricity
graph,811 application of, to electric circuits, 325 ideal,56 animal, 1
integrated, 60 complex exponential function, 325, 807 independent, 13-14, 34, 53, 128 metallic, 1
integrator, 241 conjugate of, 805 Norton equivalent, 49 electrocardiogram (ECG), 764
ladder network, 121 exponential form, 808 in parallel, 46 electromagnetic (EM) field theory, 2
linear, 119 polar form, 323, 804 practical, 55-56 electromotive force, 9
lumped-parameter, 23 exponential, 324 in series, 41
mesh, 138 electrons, 6
rectangular form, 323, 804 voltage-controlled (VCCS), 81
with mutual inductance, 616 electrophorus, 1
complex plane, 323
natural response of, 210 element(s)
component. See Reactance D
negative feedback, 90, 153 active, 12-14
conductance, 32, 340
nonlinear, 121 ideal, 15
conductor( s)
nonplanar, 138 d' Arsonval meter, 61 linear, 119
metallic, 6 damping, 280
n-port, 98, 650 lossless, 408
perfect, 23, 32 critical, 279
one-port, 648 in parallel, 42
continuity principle, 177, 187 factor, 567
open, 32, 146 passive, 12, 32, 179, 189
conventional current, 6 ratio, 278 two-terminal, 13, 34
parallel RLC, 278, 578, 582 converter(s)
phasor, 325, 338, 347 dc. See Current, direct element law(s), 13, 23, 34
analog-to-digital (AD), 103 dc steady state, 193 and the s-domain, 521-522
planar, 138 digital-to-analog (DA), 103, 106
primal, 147 dc sweep, 161 energy, 10, 11
convolution, 545 decibel scale, 563 equation(s)
port termination, 651 graphical method for evaluating, 546
RC,208-21O delta connection, 461, 463-464, 466 characteristic, 213
Laplace transform of, 545 phasor diagram, 463 dual, 146
resistive, 37, 64 Coulomb, Charles Augustin de, 4
resonant, 577 describing equation, 272 of phasor, 347
Cramer's rule, 130, 797 determinant(s), 795 equivalence, 35
RL,212-213 current, 3, 4, 6 dielectric. See Insulator(s) equivalent circuits, 35-36
RLC, 194,273,278,321-322,578,581,582 alternating, 8
s-domain, 525 differential equation(s), 209 Euler identity (Euler's formula), 324, 705, 807, 809
conventional, 6 characteristic equation of, 213 excitations. See Sources, independent
840 Index
Index 841
F Fourier transform, 690, 720, 725
impedance, 338 K
inverse, 722
farad (F), 173, 175 equivalents, 342
operations, 727-730
Faraday, Michael, 173, 175 input, 591 KCL. See Kirchhoff's laws, of current
differentiation, 727-728
filter(s), 371, 588 matrices, 652, 674 kelvin, 3
duality property, 728
active, 593, 752 output, 591 kilogram, 3
linearity, 727
all-pass, 758 phase, 453 Kirchhoff, Gustav Robert, 22, 117
time shift, 728
bandpass, 589 reflected, 626 Kirchhoff's laws, 23
pair, 722
bandstop, 589 scaling, 595, 632, 782 of current (KCL), 23-25, 29
Franklin, Benjamin, 6, 743
Besse1,766 z-parameter, 658 and impedance equivalents, 342
frequency,
lowpass, 771 improved op amp model, 86-87, 93, 100 and phasors, 342
crossover, 746
Butterworth, 765 impulse product rule, 490 and the s-domain, 521-525
cutoff,588
Chebyshev, 765 impulse response, 541 of voltage (KVL), 23, 25-27
response, 560, 729, 752
classical, 763 inductance, sum of drops form, 26
and forced response, 562
lowpass, 771 circuits with, 616 sum of rises form, 26
function, 561, 752
crossover, 745 mutual, 612, 624 KVL. See Kirchhoff's laws, of voltage
and SPICE, 598
lowpass, 588 and steady-state response, 562 self-, 612
L
ideal,588 scaling, 596, 780 and transformers, 624
linear, 744 shift, 497 induction, electromagnetic, 186 ladder network, 121
normalized, 765 function, generalized, 489 inductor(s), 2, 185 Laplace, Pierre Simon, 479
classical, 769 energy storage in, 188-189 Laplace transform(s), 481-485
parameters, 764 G practical, 196 differentiation, 493
passive, 593, 745 series and parallel, 190, 191 integration, 494
power coupling, 750 gain, 81, 561 voltage-current relations for, 335
constancy of, 91 inversion integral, 485
transformation, 776 initial condition generator, 526 i-v laws of, 617
final value theorem, 540 current, 81
initial value theorem, 540 pairs, 501, 541
first-order circuit(s), 208, 218, 219, 221, 229, 255 0l?en-Ioop, 84, 89, 584 input, 15
voltage, 81 poles of, 483
design of, 255-261 dummy, 758
gain-bandwidth product, 585, 587 properties, 500
and superposition, 229 inverting, 84
Galvani, Luigi, 1 frequency shift, 496
flux. See Flux linkage; Magnetic flux noninverting, 84
Gauss elimination, 130, 799 time-frequency sealing, 498
flux linkage, 185, 610 input-output,
generalized function, 489 time shift, 496
force analysis, 530
ground, 84, 128 t-multiplication, 499
electrodynamic, 185 relationship, 16
loop, 628 zeros of, 483
electrostatic, 185 insulator, 174
node, 128 Laplace transformation, 481
forced response, 285 perfect, 32 conditions for, 481
and forcing functions, 287 integrated circuits, 60
H leakage flux, 611
and frequency response, 562
integration, 494 Leibniz's rule, 492
to periodic inputs, 708 harmonic(s), 698, 700 integrodifferential equations, 509 linear circuit, 119
and transfer functions, 534 Heaviside, Oliver, 234,490, 519 interconnection laws. See under Kirchhoff's laws linear combination, 120
forced solution, 223 henry (H), 185, 207 International System of Units, 3, 211 linearity, 119, 121, 484, 727
trial, 223, 288 Henry, Joseph, 185, 207 inverse-inverse law, 44 loading, 100, 591
Fourier, Jean Baptiste Joseph, 689 Hertz, Heinrich Rudolph, 318, 557 integrator, inverting, 754 loop, 138
Fourier coefficients, 697
ground,628
Fourier series, 690, 697, 725
loop analysis, generalized, 138
conversion between trigonometric and exponential,
ideal switch, 179 loudspeaker system, 745
707
joule (J), 4 three-way, 748
exponential, 705-706, 713 imaginary number, 803
Joule, James Prescott, 4, 401 lumped-parameter circuit, 23
842 Index
Index 843
nonplanar circuits, 138 passband slope, 763 pulse responses, 242
Norton equivalent(s), 48-54, 57, 363 passive sign convention, 11, 12, 17 leading edge, 242
magnetic field, 185 Norton form, 49 periodic function(s), 691, 701 and SPICE, 247
magnetic flux, 185 n-port circuits, 98, 650 symmetry properties of, 692 and trailing edge, 242
matrix fundamentals, 793 phase plot, discrete, 714 pulse source, 247
conformity, 793 o phase shift, 561
methods, 793 linear, 764 Q
multiplication, 794 Oersted, Hans Christian, 21 phasor(s), 328-332
matrix inversion, 130, 798 ohm (Q), 30 analysis, 350, 559, 708 quadrature representation, of sinusoid, 320
Maxwell, James Clerk, 609 Ohm, Georg Simon, 30, 79 circuit, 338, 347 quality factor, 581
mean square value. See Power, average ohmmeter(s), 61, 63 current-voltage laws for, 333
mesh analysis, 138, 139, 372, 814-815 Ohm's law, 30-32, 87 diagram, 381 \ R
short-cut method, 139-140 one-port circuit, 648 equation of, 347
meter, 3 op amps. See Operational amplifiers and Kirchhoff's laws, 342
RC circuit, 208-210
metric system, 3 open circuit, 32, 146 locus of, 383
rational function, 503
scaled, 211 voltage, 51 rms,411
proper, 503
mho, 32 operational amplifier(s), 3, 80, 83-89, 584 and sinusoids, 330
reactance, 339
microminiaturization, 104 building blocks of, 92, 97, 98, 100 planar circuits, 138
Morse, Samuel F. B., 271 capacitive, 340
inverting amplifier, 93 pole(s), 534-535
inductive, 340
inverting summer, 94 repeated, 506
N reactive component, See Reactance
noninverting amplifier, 95-96 port(s), 98, 624, 648
realization problem, 752
noninverting summer, 96 condition, 649
natural frequency, 567 resistance, 339
voltage follower, 92 termination circuit, 651
undamped, 278 critical parallel, 279
design of, 104-109 potential difference. See Voltage input, 86
natural response, 210, 276
and impedance scaling, 783 power, 4, 10, 11 output, 86
and transfer functions, 534
improved model, 86-87, 93, 100 apparent, 432
natural solution(s), 212, 224, 289 resistive circuits, 37, 64
p,A709, 102 average, 402, 421, 694 resistor(s), 2, 22, 59
trial, 212
p,A741, 88-89, 103, 783 complex, 413, 421 carbon, 59, 60
negative feedback, 90-91, 93, 153, 585
circuit, 90 practical, 102-104 conservation of, 428, 429 discrete 59
netlist statement(s), 819 frequency dependence of, 584 factor, 432 integrated-circuit, 60
network, microminiaturization, 104 instantaneous, 11, 32,402, 459 linear, 31
electric, 2 physical packaging of, 102 maximum transfer, 425 metal film, 60
single-loop, 37 virtual open principle, 151 peak, 402 nonlinear, 31
network graph, 811 virtual short principle, 151 reactive, 417, 421 in parallel, 45
branches, 811 output, 15, 84 single-phase, 460 practical 31, 59
tree, 811 and superposition, 420 tolerance, 59
p three-phase, 460
neutrons, 6 voltage-current relations for, 333
Newton, Sir Isaac, 4 power delivery resonance, 577
nodal analysis, 127,366-367,812,813 parallel equivalents, 42, 363 instantaneous, 459 and parallel RLC circuits, 578
node(s), 23, 24. See also Supernode parallel RLC circuit, 278, 578, 582 single-phase, 457 and series RLC circuits, 581
chassis-earth, 127 Parseval relation, 718 three-phase, 457 resonant circuits, 577
equation, 128 partial fraction expansion, 503 PROBE, 251, 301 resonant frequency, 577, 579
nonreference, 127 complex conjugate poles, 505 product-by-sum rule, 44 RL circuit, 212-213
reference, 127 repeated poles, 506 proportionality, 119, 121 RLC circuit, 194, 321-322
voltages, 127 simple real poles, 505 protons, 6 parallel, 278, 578, 582
nonlinear circuit, 121 passband ripple, 763 PSpice, 17, 156, 159,249,817 series, 273, 578, 581, 582
v w
848 Index
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . @(I o tit ~
f(t) F(s)
Singularity functions
l. Unit impulse 8(t) 1
2. Unit step u(t) lis
3. Unit ramp ret) = tu(t) 1/s2
4. Unit parabola pet) = l/2t 2u(t) 1/s3
5. nth integral of impulse 8(-n) (t) l/sn
6. Unit doublet 8' (t) s
7. nth deriv. of impulse 8(n) (t) sn
Ordinary functions
8. Constant 1 lis
9. t t 1/s2
10. Power of t tn-l/(n -I)! l/sn
11. Exponential e- at l/(s + a)
12. t-mult. exponential te-at l/(s + a)2
13. repeated t-mult. expo t n- l e- at I(n - I)! l/(s + a)n
14. sine sinwt wl(s2 + ( 2)
15. cosine coswt s I(S2 + (2)
16. damped sine e- at sinwt wl([s + a]2 + ( 2)
17. damped cosine e- at coswt (s + a)/([s + a]2 + ( 2)
18. t-mult. sme t sinwt 2ws l(s2 + ( 2)2
19. t-mult. cosine t cos wt (s2 _ (2)/(s2 + ( 2)2 621.3192
w TrS
ELE
20. rectified sine Isinwtl
s2 + w 2
coth-
2w
1997
1
2l. quartic I - 3 [sinhwt- sinwt] 1/(s4 - a 4)
2w
1
22. quartic II -2 [cosh wt - cos wt] S/(S4 - a 4)
2w