Eed 3016 Control Systems: Assoc - Dr. Güleser K Demir Assist. Dr. Hatice Do Ğan

EED 3016 CONTROL SYSTEMS Assoc.Dr.
Güleser K Demir
Assist. Dr. Hatice Doğan
CONTROL SYSTEM ENGINEER
Regulates a is a combination of understands and
mechanical or components that act controls the materials
together and and forces of nature
scientific process for the benefit of
perform a certain
objective. humankind
A control system is an interconnection of components

forming a system configuration that will provide a desired
system response.
Control system engineers are concerned with understanding and controlling segments of
their environment, often called systems, to provide useful economic products for society.
1. INTRODUCTION
Control applications
are all around us:
´ A self-guided vehicle
´ The rockets fire
´ Robotic arms
´ Air-conditioning systems
´…
´…
Automatically controlled
systems that we can create
1. INTRODUCTION
Control applications
also exist in nature:
´ Regulating blood sugar
by the pancreas
´ Tracking moving objects
by our eyes
´ Adjusting our heart rate
by increasing adrenaline
in time of fight
´…
Automatically controlled
systems in our own bodies
1. INTRODUCTION
Even the nonphysical
world appears to be
automatically
regulated.
´ In quality evaluation
services
´ In environmental systems
such as global carbon
cycle or microbial
ecosystems
´ In financial systems such
as markets, supply and
service chains
´…
CONTROL SYSTEM DEFINITION
A control system consists of subsystems and processes (or plants) assembled for the purpose of controlling the
outputs of the processes.
PLANT: Furnace
Aim: Regulating the temperature of a room by
Flow of fuel Heat controlling the heat output from the furnace
Subsystems: Fuel valves, Fuel-valve actuators,
Thermostats which act as sensors, measure the room
temperature
Input; Desired response : Output; Response :

valves controller
Desired room temperature Actual room temperature
thermostats furnace
A control system provides

an output or response for a
given input or stimulus
ADVANTAGES OF CONTROL SYSTEMS
Power amplification
For example, a radar antenna, positioned by the low-power rotation of a knob at the input, requires
a large amount of power for its output rotation.
Remote control
For example, a remote controlled robot arm can be used to pick up material in a radioactive
environments.
Convenience of input form
For example, in a temperature control system, the input is a position on a thermostat. the output is
heat. Thus, a convenient position input yields a desired thermal output.
Compensation for disturbances.
The system must be able to yield the correct output even with a disturbance. For example, if wind
forces the antenna from its commanded position, the system must be able to detect the disturbance
and correct the antenna’s position.
1.2 A HISTORY OF CONTROL SYSTEMS
Liquid-level control
Around 300 B.C. … Water clocks, oil lamp,
etc.
Steam pressure and temperature
controls
Regulation of steam pressure began around
1681 with Papin’s invention of the safety
valve.
Also in the 17 th century Drebbel invented a
purely mechanical temperature control
system for hatching eggs.
CORNELIS DREBBEL’S OVEN:
In this artist’s rendering, Drebbel’s circulating oven incubates eggs. The oven consists of three nested metal boxes atop an enclosed fire. Fumes
from the fire rise up through vents into the outer box, and out of an opening at its top. The alchemist’s early thermostat sits in the water just
touching the center box. It is an L-shaped glass tube filled with alcohol and topped off by mercury. A metal rod floats in the quicksilver.
When the heated alcohol expands, it pushes up the quicksilver, and the rod rises in the tube. If the heat of the fire rises beyond a set point,
the floating rod elevates high enough to lower the damper, cutting off the air that feeds the fire. The fire would then diminish, as would the
temperature in the center box. If the heat falls too far, so does the rod, with its lever opening the vent to admit air.
Water Level Float Regulator Steam Engine
1765 Polzunov…
Flyball governor used on a steam engine in a cotton factory near Manchester.

It regulates the speed of a water wheel driven by the flow of the river.
Invented! by J. Watt 1788 to reject the load disturbances also

STEAM ENGINE
J. WATT 1736-1819
● 1769 James Watt’s Steam Engine and Governor
● 1868 J. Clerk Maxwell formulates a mathematical model for governor control

of a steam engine
● 1927 Harold Black discovers and patents the feedback amplifier
● 1927 Hendrik Bode analyzes feedback amplifiers
● 1932 Nyquist develops methods for analyzing feedback amplifier stability

• 1940s Norbert Wiener leads gun positioning effort;
becomes an engineering discipline
• 1950s Increased use of Laplace transform,

s-plane, root locus
• 1960s Sputnik, highly accurate control
systems for space vehicles,
missiles and robotics
• 1980s Routine use of digital

computers as control elements
• 1990s Feedback control in automobiles, automation,

planetary exploration
Modern Control Application Areas
1.3 RESPONSE CHARACTERISTICS
Represents what we would like the
Fourth button output to be after the elevator has
of an elevator stopped. (desired output)
is pushed
The difference
between input
and actual
output is
steady-state
error.
Physical entities cannot change After the transient response, a physical system
their state (such as position or approaches its steady-state response, which is
velocity) instantaneously. Thus, the its approximation to the desired response.
elevator location changes gradually.
OPEN LOOP SYSTEMS Disturbance means the interruption of
a settled and peaceful condition.
An example of Open-loop system: Toaster

The device is designed with the assumption
Open-loop that the toast will be darker the longer it is
systems do not subjected to the heat. The toaster does not
correct for measure the color of the toast; it works
disturbances only for a pre-specified duration. It does
and are simply not correct/change the duration of heat for
commanded by the fact that the toast is rye, white or
the input. sourdough, nor does it correct for the fact
that toast comes in different thicknesses.
CLOSED LOOP SYSTEMS
Objective: Make
the system output
Closed-loop toaster system
and It is more complex and expensive than
reference(/input) open loop toaster systems. The device
as close as has to measure both color (through
possible, i.e. light reflectivity) and humidity inside
the toaster oven. The heat duration is
make the error as
adjusted automatically for different Morphy Richards Toaster
small as possible type of breads. (The browning control)
Systems that perform measurement and correction are called closed-loop, feedback
control systems.
Systems that do not have this property of measurement and correction are called open-
loop systems.
CLOSED LOOP SYSTEMS VS OPEN LOOP SYSTEMS
qHigher accuracy ´ Simple
qLess sensitive to noise, ´ Inexpensive
disturbances, and changes in the
environment
qTransient response and steady
state error can be controlled
more conveniently
´ Disturbances cause errors, and
the output may be different
from what is desired.
´ Recalibration is necessary from
qMore complex time to time.
qExpensive
ANALYSIS AND DESIGN OBJECTIVES
What are our focal points in analysis and design of control system?
1. Producing the desired transient response

2. Reducing steady-state error
3. Achieving stability
4. Satisfying other design considerations such as cost, sensitivity, etc.
1. Producing the desired transient response
Transient response is also important

for structural reasons: Too fast
response can cause permanent
physical damage.
In an elevator example;
• Slow transient response à passengers impatient
• Rapid transient response à passengers uncomfortable
• Oscillation at the 4th floor more than a second à disconcerting feeling
2. Reducing steady-state error
In an elevator example;
• The steady state response à An elevator stopped near 4th floor.
• Our concern à The accuracy of the steady-state response
• An adequate accuracy (low error) à easy exit for the passengers
3. Achieving Stability
Total response= Natural response + Forced response
Homogeneous solution. Particular solution.

Depends only on the system, Depends on the input.
not the input.
For a control system to be useful à the natural response must

(i) eventually approach zero, or
(ii) oscillate.
If the natural response grows without bound à the system

cannot be controlled,
is called instable and
can self-destruct itself (physical damage).
In an elevator example; the elevator would crash through the floor or exit through
the ceiling!
4. Other considerations
´Factors affecting hardware selection (power

requirements, sensors for accuracies, etc.)
´Finance
´Robustness to system parameter changes

EXAMPLE: ANTENNA AZIMUTH
An Introduction to Position Control Systems
The purpose of this system:

To have the azimuth angle output of the antenna,q0(t) , follow the input angle of the
potentiometer, qI (t).
EXAMPLE: AUTOMOBILE STEERING CONTROL SYSTEM
The driver uses the difference between the actual and the desired direction of travel to
generate a controlled adjustment of the steering wheel.
EXAMPLE: REGULATING THE LEVEL OF FLUID
§ A basic, manually controlled closed-loop system for regulating the level of fluid in a tank is
shown in figure below.
§ The input is a reference level of fluid that the operator is instructed to maintain. (This
reference is memorized by the operator.)
§ The power amplifier is the operator, and the sensor is visual.
§ The operator compares the actual level with the desired level and opens or closes the valve
(actuator), adjusting the fluid flow out.
EXAMPLE: NATIONAL INCOME FEEDBACK CONTROL
A simple model of the national income feedback control system is shown in figure below.
EXAMPLE: ROTATING SPEED CONTROL
Our goal is to design a system for rotating disk speed control that will ensure that the actual
speed of rotation is within a specified percentage of the desired speed
Open-loop (without feedback) control of the Closed-loop control of the speed of a rotating disk.
speed of a rotating disk
DRILL QUESTIONS
In a chemical process control system, it is valuable to control the chemical composition of the product. To do so, a measurement
of the composition can be obtained by using an infrared stream analyzer, as shown in the figure below. The valve on the
additive stream may be controlled. Complete the control feedback loop, and sketch a block diagram describing the operation
of the control loop.
Consider the inverted pendulum shown in the figure. Sketch the block diagram
of a feedback control system using the figure below as the model. Identify the
process, sensor, actuator, and controller. The objective is keep the pendulum in
the upright position, that is to keep q = 0, in the presence of disturbances.
1.6 THE DESIGN PROCESS STEPS
MODELING IN THE FREQUENCY
DOMAIN
To understand and control complex systems, one must obtain quantitative mathematical
models.
Develop mathematical models applying the fundamental physical laws of science and
engineering.
Transfer functions in the frequency domain
State equations in the time domain
2.2 LAPLACE TRANSFORM REVIEW
´ Differential equations for modeling ´ Laplace Transform for modeling
´ Difficult to represent as a block diagram ´ Easy to represent as a block diagram
´ Computationally complex ´ Algebraic relationship
The Laplace transform is defined as The inverse Laplace transform is

# '*()
1
ℒ 𝑓(𝑡) = 𝐹 𝑠 = ) 𝑓(𝑡)𝑒 "$% 𝑑𝑡 ℒ "& 𝐹(𝑠) = 𝑓 𝑡 𝑢(𝑡) = ) 𝐹(𝑠)𝑒 $% 𝑑𝑠
2𝜋𝑗
!" '"()
where 𝑠 = 𝜎 + 𝑗𝑤, a complex variable. where 𝑢 𝑡 is the unit step function.
PARTIAL FRACTION EXPANSION
´ To find inverse Laplace transform a complicated function, we can convert it into
sum of simpler terms for which we know the Laplace transform.
𝑁(𝑠)
𝐹 𝑠 =
𝐷(𝑠)
Roots of the denominator of F(s) can be

´ real and distinct (case1 )
´ real and repeated (case 2)
´ Complex or imaginary (case 3)
CASE 1. ROOTS OF DENOMINATOR OF F(S) ARE REAL AND DISTINCT
𝑁(𝑠) 𝑁(𝑠)
𝐹 𝑠 = =
𝐷(𝑠) 𝑠 + 𝑝! 𝑠 + 𝑝" … 𝑠 + 𝑝# … (𝑠 + 𝑝$ )
𝐾! 𝐾" 𝐾# 𝐾$
= + + ⋯+ + ⋯+
𝑠 + 𝑝! 𝑠 + 𝑝" 𝑠 + 𝑝# 𝑠 + 𝑝$
𝑁 𝑠 𝑠 + 𝑝#
𝐾# = -
𝑠 + 𝑝! 𝑠 + 𝑝" … 𝑠 + 𝑝# … (𝑠 + 𝑝$ ) %&'(
!
Problem: Given the following differential equation, solve for y(t) if all initial
conditions are zero. Use the Laplace transform.
𝑑"𝑦 𝑑𝑦
" + 12 + 32𝑦 = 32 𝑢(𝑡)
𝑑𝑡 𝑑𝑡
32
𝑠 "𝑌 𝑠 + 12𝑠𝑌 𝑠 + 32𝑌 𝑠 =
𝑠
32 32 𝐾! 𝐾" 𝐾)
𝑌 𝑠 = " = = + +
𝑠(𝑠 + 12𝑠 + 32) 𝑠(𝑠 + 4)(𝑠 + 8) 𝑠 𝑠+4 𝑠+8 )"
𝐾! = : =1
(%.*)(%.,) %&0
1 2 1
𝑌 𝑠 = − + )"
𝑠 𝑠+4 𝑠+8 𝐾" = %(%.,) : =-2
%&'*
𝑦 𝑡 = 1 − 2𝑒 '*+ + 𝑒 ',+ 𝑢(𝑡) 𝐾) =

)"
: =1
%(%.*) %&',
CASE 2. ROOTS OF DENOMINATOR OF F(S) ARE REAL AND REPEATED
𝐹 𝑠 = =
𝐷(𝑠) (𝑠 + 𝑝!)1 𝑠 + 𝑝" … (𝑠 + 𝑝$ )
𝐾! 𝐾" 𝐾1 𝐾" 𝐾$
= 1 + 1'! + ⋯+ + …+
(𝑠 + 𝑝!) (𝑠 + 𝑝!) 𝑠 + 𝑝! 𝑠 + 𝑝" 𝑠 + 𝑝$
1 𝑑2'!𝐹!(𝑠)
𝐾2 = 2'!
=
𝑖 − 1 ! 𝑑𝑠
%&'("
where 𝐹! 𝑠 = 𝑠 + 𝑝! " 𝐹 𝑠
Problem: Find the inverse Laplace transform of
2
𝐹 𝑠 =
(𝑠 + 2)"(𝑠 + 1)
2 𝐾! 𝐾" 𝐾)
𝐹 𝑠 = " = + " +
(𝑠 + 2) (𝑠 + 1) 𝑠 + 1 (𝑠 + 2) 𝑠+2
"(%.!)
𝐾! = :
(%.")# (%.!) %&'!
=2
"(%.")#
𝐾" = :
(%.")# (%.!) %&'"
=-2
2 2 2
𝐹 𝑠 = − "−𝑠+2
𝑠+1 𝑠+2
'"
𝐾) = : =-2
(%.!)# %&'"
𝑓 𝑡 = 2𝑒 '+ − 2𝑡𝑒 '"+ − 2𝑒 '"+ 𝑢(𝑡)
CASE 3. ROOTS OF DENOMINATOR OF F(S) ARE COMPLEX OR IMAGINARY
𝐹 𝑠 = =
𝐷(𝑠) 𝑠 + 𝑝! 𝑠 " + 𝑎𝑠 + 𝑏 …
𝐾! 𝐾"𝑠 + 𝐾)
= + " +⋯
(𝑠 + 𝑝!) 𝑠 + 𝑎𝑠 + 𝑏
𝐾! 𝐾"(𝑠 + 𝛼) 𝐵!𝛽
= + " " +
(𝑠 + 𝑝!) (𝑠 + 𝛼) +𝛽 (𝑠 + 𝛼)"+𝛽"
𝑓 𝑡 = 𝐾!𝑒 '(" + + 𝐾" 𝑒 '3+ cos 𝛽𝑡 + 𝐵!𝑒 '3+ sin 𝛽𝑡 𝑢(𝑡)

2.3 THE TRANSFER FUNCTION 1. Consider nth order --- linear --
output time-invariant differential equation input
𝑑$ 𝑐(𝑡) 𝑑$'!𝑐(𝑡) 𝑑# 𝑟(𝑡) 𝑑#'!𝑟(𝑡)
𝑎$ $ + 𝑎$'! $'! + ⋯ + 𝑎0𝑐 𝑡 = 𝑏# # + 𝑏#'! #'! + ⋯ + 𝑏0𝑟 𝑡
𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡
2. Take Laplace Transform
𝑎$ 𝑠 $ 𝐶 𝑠 + 𝑎$'!𝑠 $'!𝐶 𝑠 + ⋯ + 𝑎0𝐶 𝑠 + 𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛 𝑡𝑒𝑟𝑚𝑠 𝑖𝑛𝑣𝑜𝑙𝑣𝑖𝑛𝑔 𝑐 𝑡

= 𝑏# 𝑠 # 𝑅 𝑠 + 𝑏#'!𝑠 #'!𝑅 𝑠 + ⋯ + 𝑏0𝑅 𝑠 + 𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛 𝑡𝑒𝑟𝑚𝑠 𝑖𝑛𝑣𝑜𝑙𝑣𝑖𝑛𝑔 𝑟(𝑡)
3. Assume all initial conditions are
zero
𝑎$ 𝑠 $ 𝐶 𝑠 + 𝑎$'!𝑠 $'!𝐶 𝑠 + ⋯ + 𝑎0𝐶 𝑠 = 𝑏# 𝑠 # 𝑅 𝑠 + 𝑏#'!𝑠 #'!𝑅 𝑠 + ⋯ + 𝑏0R s
4. Obtain transfer function G(s)
𝐶(𝑠) 𝑏# 𝑠 # + 𝑏#'!𝑠 #'! + ⋯ + 𝑏0

=𝐺 𝑠 =
𝑅(𝑠) 𝑎$ 𝑠 $ + 𝑎$'!𝑠 $'! + ⋯ + 𝑎0
Problem: Find the transfer function, G(s)=C(s)/R(s), corresponding to the differential equation
4$ 5 4# 5 45 4# 1 41
+3 # +7 + 5𝑐 = +4 + 3𝑟
4+ $ 4+ 4+ 4+ # 4+
6(%) %# .*%.)
Answer: 𝐺 𝑠 = =
7(%) %$ .)%# .8%.9
"%.!
Problem: Find the differential equation corresponding to the transfer function 𝐺 𝑠 =
%# .:%."
4# 5 45 41
Answer: # +6 + 2𝑐 = 2 4+ +𝑟
4+ 4+
%
Problem: Find the ramp response for a system whose transfer function is 𝐺 𝑠 =
(%.*)(%.,)
! ! '*+ ! ',+
Answer: 𝑐 𝑡 = − !: 𝑒 + )" 𝑒
)"
2.4 ELECTRIC NETWORK TRANSFER FUNCTIONS
TRANSFER FUNCTION – +
VIA THE DIFFERENTIAL EQUATION 𝑑𝑖(𝑡) 1
𝐿 + 𝑅𝑖 𝑡 + Y 𝑖 𝜏 𝑑𝜏 = 𝑣(𝑡)
𝑑𝑡 𝐶
0
Using 𝑖 𝑡 = 𝑑𝑞(𝑡)/𝑑𝑡
𝑑"𝑞(𝑡) 𝑑𝑞(𝑡) 1
𝐿 " +𝑅 + 𝑞 𝑡 = 𝑣(𝑡)
𝑑𝑡 𝑑𝑡 𝐶
𝒒 𝒕 = 𝑪𝒗𝒄 (𝒕)
𝑑"𝑣5 (𝑡) 𝑑𝑣5 (𝑡)
𝐿𝐶 " + 𝑅𝐶 + 𝑣5 𝑡 = 𝑣(𝑡)
𝑑𝑡 𝑑𝑡
LC𝑠 " + 𝑅𝐶𝑠 + 1 𝑉5 𝑠 = 𝑉(𝑠)
Assume zero initial
conditions to obtain
𝑉5 (𝑠) 1/𝐿𝐶 transfer function !
=
𝑉(𝑠) 𝑠 " + 𝑅 𝑠 + 1
𝐿 𝐿𝐶
TRANSFER FUNCTION –
VIA THE TRANSFORM METHODS
1 𝐼 (𝑠) 1
𝐿𝑠 + 𝑅 + 𝐼 𝑠 = 𝑉(𝑠) ⇒ =
𝐶𝑠 𝑉(𝑠) 𝐿𝑠 + 𝑅 + 1
𝐶𝑠 𝑉5 (𝑠) 1/𝐿𝐶
=
𝑉(𝑠) 𝑠 " + 𝑅 𝑠 + 1
! 𝐿 𝐿𝐶
𝑉5 𝑠 = 𝐼 𝑠 ⇒ 𝐼 𝑠 = 𝐶𝑠𝑉5 (s)
6%
VIA THE TRANSFORM METHODS
𝐺" = 1/𝑅"
𝐺! = 1/𝑅!
𝑉< 𝑠 − 𝑉 𝑠 𝑉< 𝑠 𝑉< 𝑠 − 𝑉6 𝑠
+ + =0 𝐺!𝐺"
𝑅! 𝐿𝑠 𝑅" 𝑉6 𝑠 𝑠
= 𝐶
𝑉(𝑠) 𝐺 𝐺 𝐿+𝐶 𝐺
𝑉6 𝑠 − 𝑉< 𝑠 𝐺! + 𝐺" 𝑠 " + ! "𝐿𝐶 𝑠 + 𝐿𝐶"
𝐶𝑠𝑉6 𝑠 + =0
𝑅"
VIA THE TRANSFORM METHODS 1 10 8
𝑍" 𝑠 = 𝑅" 𝑠 + = 220 x 10) +
𝐶"𝑠 𝑠
𝑉0 𝑠 𝑍" 𝑠
=
𝑉2 (𝑠) 𝑍!(𝑠)
𝑉0 𝑠 𝑠 " + 45.95𝑠 + 22.55

= −1.232
𝑉2 (𝑠) 𝑠
This circuit can be used

as a PID controller. It
1 1 360 x 10) improves the
𝑍! 𝑠 = = = performance of a control
1 1 2.016𝑠 + 1
𝐶!𝑠 + 5.6 x10':𝑠 + system. Details later…
𝑅! 360 x 10)
2.4 TRANSLATIONAL MECHANICAL SYSTEM TRANSFER FUNCTIONS
Analogies between electrical

and mechanical systems:
Force ⟺ Voltage
Velocity ⟺ Current
Displacement ⟺ Charge
Spring ⟺Capacitor
Viscous damper ⟺ Resistor
Mass ⟺ Inductor
TRANSFER FUNCTION
𝑑"𝑥(𝑡)
𝑓$=+ =𝑀
𝑑𝑡 "
𝑑"𝑥(𝑡) 𝑑𝑥(𝑡)
𝑀 " + 𝑓> + 𝐾𝑥 𝑡 = 𝑓(𝑡)
𝑑𝑡 𝑑𝑡
𝑀𝑠 " + 𝑓> 𝑠 + 𝐾 𝑋(𝑠) = 𝐹(𝑠)
𝑋 (𝑠) 1
𝐺 𝑠 = =
𝐹(𝑠) 𝑀𝑠 " + 𝑓> 𝑠 + 𝐾
Problem: Find the transfer function, G(s)=X2(s)/F(s), for the translational
mechanical system shown in below.
Answer:
3𝑠 + 1
𝐺 𝑠 =
𝑠(𝑠 ) + 7𝑠 " + 5𝑠 + 1)
2.5 ROTATIONAL MECHANICAL SYSTEM TRANSFER FUNCTIONS
2.7 TRANSFER FUNCTIONS FOR SYSTEMS WITH GEARS
Gears allow you to match the drive system and the load
---a trade-off between speed and torque.
Rotational mechanical impedances can
be reflected through gear trains by
multiplying the mechanical impedance
by the ratio
"
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑡𝑒𝑒𝑡ℎ 𝑜𝑓 𝑔𝑒𝑎𝑟
𝑜𝑛 𝑑𝑒𝑠𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛 𝑠ℎ𝑎𝑓𝑡
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑡𝑒𝑒𝑡ℎ 𝑜𝑓 𝑔𝑒𝑎𝑟
𝑜𝑛 𝑠𝑜𝑢𝑟𝑐𝑒 𝑠ℎ𝑎𝑓𝑡
2.8 ELECTROMECHANICAL SYSTEM TRANSFER FUNCTIONS
Motor’s Schematic. 𝑑𝜃#
𝑣? 𝑡 = 𝐾? 𝑉? 𝑠 = 𝐾? 𝑠𝜃# (𝑠)
𝑑𝑡
𝑅@ 𝐼@ 𝑠 + 𝐿@ 𝑠𝐼@ 𝑠 + 𝑉? (𝑠) = 𝐸@ (𝑠)
𝑇# 𝑠 = 𝐾+ 𝐼@ (𝑠)
(𝑅@ +𝐿@ 𝑠)𝑇# (𝑠)

+ 𝐾? 𝑠𝜃# 𝑠 = 𝐸@ (𝑠)
𝐾+
𝑇# 𝑠 = (𝐽# 𝑠 " + 𝐷# 𝑠)𝜃# (𝑠) (𝑅@ +𝐿@ 𝑠)(𝐽# 𝑠 " + 𝐷# 𝑠)𝜃# (𝑠)
+ 𝐾? 𝑠𝜃# 𝑠 = 𝐸@ (𝑠)
Neglect 𝐾+
𝜃# 𝑠 𝐾+ /(𝑅@ 𝐽# )
=
𝐸@ (𝑠) 1 𝐾+ 𝐾?
𝑠 𝑠+ 𝐷 +
𝐽# # 𝑅@
𝜃# 𝑠 𝐾
=
𝐸@ (𝑠) 𝑠(𝑠 + 𝛼)
2.10 NONLINEARITIES
A linear system possesses two properties: Superposition and homogeneity
The output response of a system to the Multiplication of an input by a scalar

sum of inputs is the sum of the yields a response that is multiplied by
responses to the individual inputs. the same scalar.
"#$%& "#$%&
𝑟! 𝑡 𝑐! 𝑡 𝑟! 𝑡 𝑐! 𝑡
"#$%& "#$%&
𝑟' 𝑡 𝑐' 𝑡 𝐴𝑟! 𝑡 𝐴𝑐! 𝑡
"#$%&
𝑟! 𝑡 + 𝑟' 𝑡 𝑐! 𝑡 + 𝑐' 𝑡
Linear models are useful because

´ They are easy to compute, understand ´ The analysis of linear theory is complete,
and visualize. developed and efficient
´ They give predictable outputs, in time ´ Linear differential equations are easy to
and over iterations solve!
Motor does not Gears do not fit
respond at very low tightly exhibit
input due to backlash. The input
frictional forces moves over a small
range without the
output responding.
LINEAR NONLINEAR NONLINEAR NONLINEAR

Make a linear approximation to a nonlinear system.
• Simplifies the analysis and design
• Used as long as the results yield a good approximation to reality
• Analysis of a system represented by nonlinear partial differential equations is
extremely difficult and requires heavy computations
• There is no general analytic method available for solving nonlinear systems
2.11 LINEARIZATION
• Write the
Recognize the nonlinear
nonlinear differential
component equation
• Linearize it for small-

signal inputs about the
Linearize a steady-state solution
nonlinear when the small signal
differential eq. input is equal to zero.
• Steady-state solution is
called equilibrium 𝑓 𝑥 − 𝑓(𝑥( ) ≈ 𝑚) (𝑥 − 𝑥( )
Take the Laplace
transform of the • Separate input 𝛿𝑓(𝑥) ≈ 𝑚) 𝛿𝑥
linearized diff. eq., and output
assuming zero variables and 𝑓 𝑥 ≈ 𝑓 𝑥( + 𝑚) 𝑥 − 𝑥( ≈ 𝑓 𝑥( + 𝑚) 𝛿𝑥
form transfer
initial conditions. function.
LINEARIZING A FUNCTION
Example: Linearize 𝑓 𝑥 = 5 cos 𝑥 about 𝑥 = 𝜋/2
𝑓 𝑥 ≈ 𝑓 𝑥( + 𝑚) 𝑥 − 𝑥( ≈ 𝑓 𝑥( + 𝑚) 𝛿𝑥 ⟹ 𝒇 𝒙 = −𝟓 𝜹𝒙
𝑑𝑓 𝑑𝑓
= −5 sin 𝑥 ⇒ < = −5
𝜋 𝜋 𝑑𝑥 𝑑𝑥 *+,/'
𝑓 𝑥( =𝑓 = 5 cos = 0
2 2
Taylor series expansion: Ignore higher-order terms
y = f ( x) = f ( x0 ) + ( x - x0 )
df
+
(x - x )0
2
d2 f
+ ...
dx x = x0 2! dx 2 x = x0
df
y = f ( xo ) + ( x - x0 ) = y0 + m( x - x0 )
dx x = x0
y - y0 = m( x - x0 )
Dy = mDx
LINEARIZING A DIFFERENTIAL EQUATION
&! * &*
Example: Linearize +2 + cos 𝑥 = 0 about 𝑥 = 𝜋/4
&. ! &.
𝑥 = 𝛿𝑥 + 𝜋/4
𝑑 ' (𝛿𝑥 + 𝜋/4 ) 𝑑(𝛿𝑥 + 𝜋/4 ) 𝒅𝟐 𝜹𝒙 𝒅𝜹𝒙 𝟐 𝟐
+2 + cos(𝛿𝑥 + 𝜋/4 ) = 0 ⇒ +𝟐 − 𝜹𝒙 = −
𝑑𝑡 ' 𝑑𝑡 𝒅𝒕𝟐 𝒅𝒕 𝟐 𝟐
𝑑𝛿𝑥 Linearized Diff. Eq.

𝑑 ' 𝛿𝑥 Solve it for 𝛿𝑥 then obtain 𝑥 = 𝛿𝑥 +
𝑑𝑡
𝑑𝑡 ' π/4
Notes:
• Linearized diff. eq. is non-homogeneous though
𝑑𝑓(𝑥) the non-linear diff. eq. is homogeneous!
𝑓 𝑥 − 𝑓(𝑥( ) ≈ < (𝑥 − 𝑥( ) • The additional term can be thought of as an
𝑑𝑥 *
𝜋 𝜋
"
𝑑 cos 𝑥 𝜋 input.
cos 𝛿𝑥 + − cos = < 𝛿𝑥 = − sin( )𝛿𝑥 • Since the roots of the equation are positive,
4 4 𝑑𝑥 , 4
the homogeneous solution of linearized eq.
𝜋 𝜋 𝜋 0
cos 𝛿𝑥 + = cos − sin( ) 𝛿𝑥 grows without bound. Thus this system
4 4 4 𝝅
linearized around is not stable.
𝟒
TRANSFER FUNCTION--NONLINEAR ELECTRICAL NETWORK
Example: Find the transfer function VL(s)/V(s) for the electrical network which contains a
nonlinear resistor whose voltage-current relationship is defined by 𝑖3 = 2𝑒 (.!5# . Also, v(t) is a
small-signal source.
1
𝑖3 = 2𝑒 (.!5# ⇒ 𝑣3 = 10 ln 𝑖3 Linearize this diff. equation.
2
(Set small-signal source, v(t), equal to zero and find the steady-
𝑑𝑖 1 state current (/equilibrium point), 𝑖6 . Then linearize non-linear
𝐿 + 10 ln 𝑖 − 20 = v(t)
𝑑𝑡 2 function around 𝑖6 by using 1st order taylor expansion)
𝑑𝛿𝑖 1 1 𝑑𝑖( 1
𝐿 + 10 (ln 𝑖( + 𝛿𝑖) − 20 = v(t) 𝐿 + 10 ln 𝑖( − 20 = 0
𝑑𝑡 2 𝑖( 𝑑𝑡 2
'
⇒ 𝑖( = 2𝑒 ⇒ 𝑖6 = 14.78amps
𝑑𝛿𝑖
+ 0.677δ𝑖 = v(t) 𝑑(𝑖( + 𝛿𝑖) 1
𝑑𝑡 𝐿 + 10 ln (𝑖( + 𝛿𝑖) − 20 = v(t)
𝑑𝑡 2
𝑉(𝑠)
𝛿𝑖 𝑠 = 1 1 𝑑(ln 0.5𝑖)
𝑠 + 0.677 ln (𝑖! + 𝛿𝑖) − ln 𝑖! = J 𝛿𝑖
2 2 𝑑𝑖 "!
𝑑(𝑖( + 𝛿𝑖) 𝑑𝛿𝑖 1 1 1
𝑣7 𝑡 = 𝐿 =𝐿 ln (𝑖! + 𝛿𝑖) − ln 𝑖! = 𝛿𝑖
𝑑𝑡 𝑑𝑡 2 2 𝑖!
𝑉 (𝑠) 𝑠 1 1 1
𝑉7 𝑠 = 𝐿𝑠𝛿𝑖(𝑠) ⇒ 7 = ln 𝑖! + 𝛿𝑖 = ln 𝑖! + 𝛿𝑖
𝑉(𝑠) 𝑠 + 0.677 2 2 𝑖!
MODELING IN THE TIME DOMAIN Assoc.Dr. Güleser K Demir
Assist.Dr. Hatice Doğan
Mathematical Models
3.1 INTRODUCTION
Classical or Frequency Domain Technique Modern or State-Space or Time Domain
Technique
´ Advantages
´ Advantages
´ Converts system’s differential equation into a
transfer function, thus gives a model that ´ Provides a unified method for modeling,
algebraically relates output to input. analyzing, and designing a wide range of
systems using matrix algebra.
´ Rapidly provides stability and transient
response information. ´ Useful to represent nonlinear systems that
have backlash, saturation, and dead zone.
´ Disadvantages
´ Nonlinear, Time-Varying, Multiple-input,
´ Applicable only to Linear, Time-Invariant (LTI) multiple-output systems
systems or their close approximations.
´ Disadvantages
For example, LTI modeling in space ´ Not as intuitive as classical method.
applications is inadequate. Models
´ Calculations required before physical
for time-varying systems (for interpretation is apparent
example, missiles with varying fuel
levels) are necessary.
3.3 STATE SPACE REPRESENTATION
System matrix Input matrix
̇
𝒙(𝑡) = 𝑨𝒙(𝑡) + 𝑩𝒖(𝑡)
𝒚(𝑡) = 𝑪𝒙(𝑡) + 𝑫𝒖(𝑡)
Input vector State vector Output vector
𝑢! (𝑡) 𝑥! (𝑡) 𝑦! (𝑡)
𝑢 (𝑡) 𝑥 (𝑡) 𝑦 (𝑡)
𝒖(𝑡) = " 𝒙(𝑡) = " 𝒚= "
⋮ ⋮ ⋮ Output
𝑢$ (𝑡) 𝑥# (𝑡) 𝑦% (𝑡) Feedforward
matrix
matrix
DEFINITIONS
System variables: Any variable that responds to an input or initial conditions.
1
State variables: The smallest set of linearly independent system variables such that knowledge of
these variables at t=t0 , together with knowledge of the input, completely determines the
behavior of the system for any time t≥t0 .
Note that the concept of state is by no means limited to physical systems. It is applicable to
biological systems, economic systems, social systems, and others.
State Vector: A vector whose elements are the state variables.
1
Linear combination: A linear combination of n variables, xi, for i=1 to n, is 𝑆 = 𝐾! 𝑥! +
𝐾!"# 𝑥!"# + ⋯ 𝐾# 𝑥# where eack Ki is a constant.
Linear Independence: A set of variables is linearly independent if none of the variables
can be written as a linear combination of the others.
State Space: The n-dimensional space whose coordinate axes consist of the x1 axis, x2 axis,
….., xn axis, where x1, x2,…… , xn are state variables, is called a state space.
"State space" refers to the space whose axes are the state variables. The state of the
system can be represented as a vector within that space.
(state variable-1)
(state variable-2)
State Equations: A set of n simultaneous, first order differential equations with n state
variables.
Output Equation: The algebraic equation that expresses the output variables of a system as
a linear combinations of the state variables and the inputs.
𝑥#
𝑥$
State vector 𝒙 = ⋮
𝑥!
State Equation 𝒙̇ = 𝑨𝒙 + 𝑩𝒖
Output Equation 𝒚 = 𝑪𝒙 + 𝑫𝒖
Input or control vector
𝑢#
𝑢$
𝒖= ⋮
Output vector
𝑦# 𝑢%
𝑦$
𝒚= ⋮
𝑦&
3.4 APPLYING THE STATE-SPACE REPRESENTATION
The state vector must be chosen according to the following considerations:

´ A minimum number of state variables should be selected.
´ The minimum number of state variables must be linearly independent.
The minimum number of state variables is equal to:

ü The order of the differential equation describing the system.
ü The order of the denominator of the transfer function
ü The number of independent energy-storage elements
Example: Find a state-space representation of the circuit given below if the output is the current
through the resistor.
3) Substitute the equations in step 2 into the

ones in step-1.(i.e. Obtain state equations)
𝑑𝑣3 1 𝑑𝑣3 1 1
𝐶 = − 𝑣3 + 𝑖1 ⇒ =− 𝑣3 + 𝑖1
𝑑𝑡 𝑅 𝑑𝑡 𝑅𝐶 𝐶
𝑑𝑖1 𝑑𝑖1 1 1
𝐿 = −𝑣/ + 𝑣 ⇒ = − 𝑣/ + 𝑣
𝑑𝑡 𝑑𝑡 𝐿 𝐿
1) Select the state variables by writing the 4) Find the output equation:
derivative equation for all energy-storage 1
elements: Choose as the 𝑖2 = 𝑣3
,-! ,0" state variables 𝑅
𝐶 ,. = 𝑖/ 𝐿 ,. = 𝑣1 5) Write state-space representation in vector-
matrix form:
2) Obtain 𝑖/ and 𝑣1 in terms of state variables: 𝑣/̇ 0
−1/(𝑅𝐶) 1/𝐶 𝑣/
1 ̇ = 𝑖 + 1 𝑣
𝑖/ = −𝑖2 + 𝑖1 = − 𝑣3 + 𝑖1 𝑖1 −1/𝐿 0 1
𝑅 𝐿
𝑣/
𝑣1 = −𝑣/ + 𝑣 𝑖2 = 1/𝑅 0 𝑖
1
3.5 CONVERTING A TRANSFER FUNCTION TO STATE SPACE
Ø Consider the differential equation Note: Corresponding transfer
function is
dny d n -1 y d y Y (s)
= n
b0
n
+ an -1 n -1 + ! + + a0 y = b0u U ( s ) s + an -1s n -1 + ! + a1s1 + a0
dt dt dt
where y is the measure variable and u is the input.
Ø The minimum number of state variables is n since the differential equation is nth order.
Ø Convenient way: Choose the output, y(t), and its (n-1) derivatives as state variables.
x1 = y
x2 = y" Þ x"1 = x2 First row of state equations
#
d n -1 y
xn = n -1 Þ x"n -1 = xn
dt
dny
x"n = n = -a0 x1 - a1 x2 ! - an -1 xn + b0u Last row of state equations
dt
Ø Arrange in vector-matrix format.
é x1 ù é 0 1 0 # 0 ù é x1 ù é 0 ù
êx ú ê 0 0 1 # 0 ú ê x2 ú ê 0 ú
d ê 2
ú ê úê ú ê ú
ê ! ú=ê ! ! ! " ! ú ê ! ú + ê ! úu
dt ê ú ê úê ú ê ú
ê xn -1 ú ê 0 0 0 " 1 ú ê xn -1 ú ê 0 ú
êë xn úû êë- a0 - a1 - a2 " - an -1 úû êë xn úû êëb0 úû
é x1 ù
êx ú
ê 2ú
y = [1 0 0 " 0]ê ! ú
ê ú
ê xn -1 ú Note the transfer function
êë xn úû format
Y ( s) b0
= n
U ( s) s + a n -1 s n -1 +!+a1 s1 + a 0
Example: Find the state-space representation for the transfer function given below.
𝐶(𝑠) 24 ' + 9𝑠 $ + 26𝑠 + 24 C s = 24R(s)

= ' ⇒ 𝑠
𝑅(𝑠) (𝑠 + 9𝑠 $ + 26𝑠 + 24)
𝑐⃛ + 9𝑐̈ + 26𝑐̇ + 24𝑐 = 24𝑟

𝑥̇ # = 𝑥$
𝑥# = 𝑐
𝑥̇ $ = 𝑥'
𝑥$ = 𝑐̇
𝑥̇ ' = −24𝑥# − 26𝑥$ − 9𝑥' + 24𝑟
𝑥' = 𝑐̈
𝑦 = 𝑐 = 𝑥#
𝑥#̇ 0 1 0 𝑥# 0
𝑥$̇ = 0 0 1 𝑥$ + 0 r
𝑥̇ ' −24 −26 −9 𝑥' 24
𝑥#
𝑦= 1 0 0 𝑥$
𝑥'
TRANSFER FUNCTION WITH NUMERATOR POLYNOMIAL
𝑌 𝑠 = 𝐶 𝑠 = 𝑏$ 𝑠 $ + 𝑏# 𝑠 + 𝑏( 𝑋# 𝑠
X 1 (s) 1 / a3
=
R( s ) s 3 + a2 s 2 + a1 s + a0 𝑑 $ 𝑥# 𝑑𝑥#
𝑦 𝑡 = 𝑏$ $ + 𝑏# + 𝑏( 𝑥#
a3 a3 a3 𝑑𝑡 𝑑𝑡
é ù é ù
é x1 ù ê 0 1 0 ú é x1 ù ê 0 ú 𝑦 𝑡 = 𝑏( 𝑥# + 𝑏# 𝑥$ + 𝑏$ 𝑥'
d ê ú ê
ê x2 ú = 0 0 1 ú êê x2 úú + ê 0 úu 𝑥#
dt ê a a a2 ú ê1ú
êë x3 úû ê- 0 - 1 - ú êë x3 úû ê ú 𝑦 = 𝑏( 𝑏# 𝑏$ 𝑥$
êë a3 a3 a3 úû êë a3 úû 𝑥'
Example: Find the state-space representation for the transfer function given below.
1) Seperate the transfer

function into two cascaded
transfer functions.
𝑥̇ # = 𝑥$ 𝑥̇ # 0 1 0 𝑥# 0
𝑥̇ $ = 𝑥' 𝑥̇ $ = 0 0 1 𝑥$ + 0 r
𝑥̇ ' = −24𝑥# − 26𝑥$ − 9𝑥' + 𝑟 𝑥̇ ' −24 −26 −9 𝑥' 1
C s = 𝑠 $ + 7𝑠 + 2 𝑋# (s)
𝑥̈ # = 𝑥'
𝑐 = 𝑥̈ # + 7𝑥̇ # + 2𝑥#
𝑥̇ # = 𝑥$ 𝑥!
𝑦 𝑡 = 2𝑥! + 7𝑥" + 𝑥& ⇒ 𝑦= 2 7 1 𝑥"
𝑥&
𝑥̇ # 0 1 0 𝑥# 0
𝑥̇ $ = 0 0 1 𝑥$ + 0 r
𝑥̇ ' −24 −26 −9 𝑥' 1
𝑥!
𝑦= 2 7 1 𝑥"
𝑥&
equivalent
block diagram
Note: y(t) = c(t)
3.6 CONVERTING FROM STATE SPACE TO A TRANSFER FUNCTION
𝒙̇ = 𝑨𝒙 + 𝑩𝒖
𝒚 = 𝑪𝒙 + 𝑫𝒖
Ø Take the Laplace transform assuming zero initial conditions

𝑠𝑿(𝑠) = 𝑨𝑿(𝑠) + 𝑩𝑼(𝑠)
Identity matrix 𝒀 𝑠 = 𝑪𝑿 𝑠 + 𝑫𝑼 𝑠
Ø Solve for 𝑿 𝑠
!"
𝑠𝑰 − 𝑨 𝑿 𝑠 = 𝑩𝑼 𝑠 ⇒ 𝑿 𝑠 = 𝑠𝑰 − 𝑨 𝑩𝑼(𝑠)
Ø Use 𝑿 𝑠 in output equation
𝒀 𝑠 = 𝑪 𝑠𝑰 − 𝑨 !" 𝑩𝑼 𝑠 + 𝑫𝑼 𝑠
𝒀 𝑠 = 𝑪 𝑠𝑰 − 𝑨 !" 𝑩 + 𝑫 𝑼 𝑠 Transfer function matrix
Ø For Single-Input, Single-output (SISO) systems, the transfer function is:

𝑌(𝑠)
𝑇 𝑠 = = 𝑪 𝑠𝑰 − 𝑨 !" 𝑩 + 𝑫
𝑈(𝑠)
Example: Find the transfer function for the system defined by
0 1 0 10
𝒙̇ = 0 0 1 𝒙+ 0 𝑢
−1 −2 −3 0
𝑦= 1 0 0𝒙
𝑠 0 0 0 1 0 𝑠 −1 0
𝑠𝑰 − 𝑨 = 0 𝑠 0 − 0 0 1 = 0 𝑠 −1
0 0 𝑠 −1 −2 −3 1 2 𝑠+3
𝑠 $ + 3𝑠 + 2 𝑠+3 1
−1 𝑠(𝑠 + 3) 𝑠
"#
𝑎𝑑𝑗 𝑠𝑰 − 𝑨 −𝑠 −(2𝑠 + 1) 𝑠 $
𝑠𝑰 − 𝑨 = = 𝑌(𝑠)
𝑑𝑒𝑡 𝑠𝑰 − 𝑨 𝑠 ' + 3𝑠 $ + 2𝑠 + 1 𝑇 𝑠 = = 𝑪 𝑠𝑰 − 𝑨 "# 𝑩 + 𝑫
𝑈(𝑠)
𝑌(𝑠) 10(𝑠 $ + 3𝑠 + 2)
10 𝑇 𝑠 = =
𝑩= 0 𝑪= 1 0 0 𝑫=0 𝑈(𝑠) 𝑠 ' + 3𝑠 $ + 2𝑠 + 1
0
Example: Find the transfer function for the multiple-input multiple-output (MIMO) system
defined by é x!1 ù é 0 1 0 ù é x1 ù é0 0ù
ê x! ú = ê 0 - 4 3 ú ê x ú + ê1 0 ú é u1 ù
ê 2ú ê úê 2 ú ê ú êëu2 úû
êë x!3 úû êë- 1 - 1 - 2úû êë x3 úû êë 0 1úû
é x1 ù
é y1 ù é1 0 0ù ê ú
ê y ú = ê0 0 1ú ê x2 ú
ë 2û ë ûê ú
adj ( sI - A) ë x3 û
( sI - A) -1 =
sI - A 𝑌# 𝑠+2
=
𝑈# 𝑠 ' + 6𝑠 $ + 11𝑠 + 3
é s 2 + 6 s + 11 s + 2 3 ù
1 ê ú
= - 3 s 2
+ 2 3s 𝑌# 3
s ( s + 4)( s + 2) + 3 + 3s ê ú =
𝑈$ 𝑠 ' + 6𝑠 $ + 11𝑠 + 3
êë s + 4 - s - 1 s + 4 s úû
2
G ( s ) = [C ( sI - A) -1 B + D] 𝑌$ −(𝑠 + 1)
𝑌! 𝑌! =
1 é s+2 3 ù 𝑈# 𝑠 ' + 6𝑠 $ + 11𝑠 + 3
𝑈 𝑈"
= = !
s 3 + 6 s 2 + 11s + 3 êë- ( s + 1) s ( s + 4)úû 𝑌" 𝑌" Transfer 𝑌$
=
𝑠(𝑠 + 4)
𝑈! 𝑈" function matrix 𝑈$ 𝑠 ' + 6𝑠 $ + 11𝑠 + 3
3.7 LINEARIZATION
Ø A prime advantage of the state-space representation over the transfer function representation is the
ability to represent systems with nonlinearities.
Ø We can linearize the state eq. about the equilibrium point and can use only for a limited range of
operation.
Ø Let us represent a nonlinear system by the following vector-matrix formed state equations:
Note that it cannot 𝑓" (𝑥" , 𝑥# , … 𝑥$ , 𝑢" , 𝑢# , … 𝑢% )

be written in the 𝑓 (𝑥 , 𝑥 , … 𝑥$ , 𝑢" , 𝑢# , … 𝑢% )
form 𝒙̇ = 𝑨𝒙 + 𝑩𝒖
𝒙̇ = 𝒇 𝒙, 𝒖 = # " #
⋮
𝑓$ (𝑥" , 𝑥# , … 𝑥$ , 𝑢" , 𝑢# , … 𝑢% )
ØFor small-signal linearization, first determine equilibrium points 𝒙& , 𝒖& from
𝒙̇ & = 𝟎 = 𝒇 𝒙& , 𝒖&
ØLet 𝒙 and 𝒖 be perturbed about the equilibrium point and use taylor series to expand
𝒙 = 𝒙& + 𝛿𝒙
𝒖 = 𝒖& + 𝛿𝒖
𝜕𝒇 𝜕𝒇
𝒙̇ & + 𝛿 𝒙̇ ≈ 𝒇(𝒙& , 𝒖& ) + @ 𝛿𝒙 + @ 𝛿𝒖
𝜕𝒙 (𝒙 𝜕𝒖 (𝒙
' ,𝒖' ) ' ,𝒖' )
0 0
𝜕𝒇 𝜕𝒇
𝒙̇ A + 𝛿 𝒙̇ ≈̇ 𝑓(𝒙A, 𝒖A) + = 𝛿𝒙 + = 𝛿𝒖
𝜕𝒙 (𝒙 𝜕𝒖 (𝒙
# ,𝒖# ) # ,𝒖# )
𝜕𝑓C 𝜕𝑓C 𝜕𝑓C 𝜕𝑓C

= ⋯ = = ⋯ =
𝛿 𝑥̇ C 𝜕𝑥C (𝒙 𝜕𝑥D (𝒙 𝛿𝑥C 𝜕𝑢C (𝒙 𝜕𝑢E (𝒙 𝛿𝑢C
# ,𝒖# ) # ,𝒖# ) # ,𝒖# ) # ,𝒖# )
⋮ = ⋮ ⋱ ⋮ ⋮ + ⋮ ⋱ ⋮ ⋮
𝛿 𝑥̇ D 𝜕𝑓D 𝜕𝑓D 𝛿𝑥D 𝜕𝑓D 𝜕𝑓D 𝛿𝑢E
= ⋯ = = ⋯ =
𝜕𝑥C (𝒙 ,𝒖 ) 𝜕𝑥D (𝒙 𝜕𝑢C (𝒙 ,𝒖 ) 𝜕𝑢E (𝒙
# # # ,𝒖# ) # # # ,𝒖# )
𝑨 𝑩
Ø Then drop ′𝛿′ and abusing notation, write the linearized differential equation
Example: Find the linearized state equation for the following differential equation of nonlinear
spring.
𝑦̈ = 𝑘Z 𝑦 + 𝑘[ 𝑦 \
𝐿𝑒𝑡 𝑥! = 𝑦 𝑎𝑛𝑑 𝑥" = 𝑦̇

) 𝑥# 𝑥$ 𝑓# (𝑥# , 𝑥$ )
=
𝑥$ 𝑘# 𝑥# + 𝑘$ 𝑥# ' =
)* 𝑓$ (𝑥# , 𝑥$ )
To find the equilibrium points, we must solve
𝑓# (𝑥# , 𝑥$ ) 𝑥$
𝐟 𝐱 = = =0
𝑓$ (𝑥# , 𝑥$ ) 𝑘# 𝑥# + 𝑘$ 𝑥# '
Then
𝑓# 𝑥# , 𝑥$ = 0 Þ𝑥($ = 0
"+!
𝑓$ 𝑥# , 𝑥$ = 0 Þ𝑘# 𝑥(# + 𝑘$ 𝑥(# ' = 0 Þ 𝑥(# = 0 or 𝑥(# = ± Y+"
𝜕𝒇 𝜕𝒇
𝛿 𝒙̇ = _ 𝛿𝒙 + _ 𝛿𝒖
𝜕𝒙 (𝒙 𝜕𝒖 (𝒙
# ,𝒖# ) # ,𝒖# )
For the state space model
,-! ,-!
,.! ,." 0 1 0 1
𝐴 = = =
,-" ,-" 𝑘# + 3𝑘$ 𝑥# $ 0 .# 𝑘# + 3𝑘$ 𝑥(# $ 0
,.! ,." .
#
For the equilibruim point 𝑥(# = 0 and 𝑥($ = 0

0 1
𝐴=
𝑘# 0
which are the standard dynamics of a system with just a linear spring of stiffness 𝑘!
The linearized system is

𝛿 𝒙̇ = 𝑨𝛿𝒙 + 𝑩𝒖
𝛿 𝑥̇ # 0 1 𝛿𝑥#
=
𝛿 𝑥̇ $ 𝑘# 0 𝛿𝑥$
EXAMPLE
Given nonlinear system below, find linearized model
𝑥̇ C(𝑡)=𝑥L(𝑡)
𝑥̇ L(𝑡)=−𝑥C(𝑡)𝑥L(𝑡) − 𝑥L(𝑡) + 𝑢 𝑡 , 𝑥A = 𝑥AC 𝑥AL = 1 0 , 𝑢A= 0
Solution
𝑥̇ C(𝑡)=𝑥L 𝑡 = 𝑓C 𝑥C 𝑡 , 𝑥L 𝑡 , 𝑢(𝑡)
𝑥̇ L(𝑡)=−𝑥C 𝑡 𝑥L 𝑡 − 𝑥L 𝑡 + 𝑢 𝑡 = 𝑓L 𝑥C 𝑡 , 𝑥L 𝑡 , 𝑢(𝑡)
𝑥̇ C(𝑡)=𝑥L 𝑡 = 𝑓C 𝑥C 𝑡 , 𝑥L 𝑡 , 𝑢(𝑡)
𝑥̇ L(𝑡)=−𝑥C 𝑡 𝑥L 𝑡 − 𝑥L 𝑡 + 𝑢 𝑡 = 𝑓L 𝑥C 𝑡 , 𝑥L 𝑡 , 𝑢(𝑡)
𝜕𝒇 𝜕𝒇
𝛿 𝒙̇ = _ 𝛿𝒙 + _ 𝛿𝒖
𝜕𝒙 (𝒙 𝜕𝒖 (𝒙
# ,𝒖# ) # ,𝒖# )
MN$ MN$
MO$ MO% 0 1 0 1
𝐴 = F = =
MN% MN% −𝑥AL −𝑥AC − 1 0 −2
MO$ MO% O# ,P#
MN$
MP 0
𝐵= MN%
F =
1
linearized system: MP O# ,P#
𝛿 𝑥̇ # 0 1 𝛿𝑥# 0 𝛿 𝑥̇ # (𝑡)=𝑥$ (𝑡)

= + 𝑢
𝛿 𝑥̇ $ 0 −2 𝛿𝑥$ 1 𝛿 𝑥̇ $ (𝑡)=−2𝑥$ (𝑡) + 𝑢(𝑡)
EXAMPLE
Given the nonlinear system below, find the linearized model
𝑥̇ C(𝑡)=𝑥L(𝑡)
𝑥̇ L(𝑡)=-sin 𝑥C 𝑡 −𝑥L 𝑡 + 𝑢(𝑡)
Nominal (Equilibrium) point;

C Q
𝑢A = , 𝑥AC = , 𝑥AL = 0
L R
Solution 𝑥̇ " (𝑡)=𝑥# (𝑡) = 𝑓" 𝑥" 𝑡 , 𝑥# 𝑡 , 𝑢(𝑡)
𝑥̇ # (𝑡)=-sin 𝑥" 𝑡 −0.1𝑥# 𝑡 + 𝑢 𝑡 = 𝑓# 𝑥" 𝑡 , 𝑥# 𝑡 , 𝑢(𝑡)
𝜕𝒇 𝜕𝒇
𝛿 𝒙̇ = = 𝛿𝒙 + = 𝛿𝒖
𝜕𝒙 (𝒙 𝜕𝒖 (𝒙
# ,𝒖# ) # ,𝒖# )
,-( ,-(
,.( ,.) 0 1 0 1
𝐴 = J = =
,-) ,-) −𝑐𝑜𝑠𝑥&" −0.1 −0.707 −0.1
,.( ,.) .' ,/'
,-(
,/ 0
𝐵= J = ;
,-) 1
,/ .' ,/'
𝛿 𝑥̇ # 0 1 𝛿𝑥# 0 𝛿 𝑥̇ # (𝑡)=𝑥$ (𝑡)

= + 𝑢 𝛿 𝑥̇ # (𝑡)=−0.707𝑥# (𝑡) − 0.1𝑥$ (𝑡) + 𝑢(𝑡)
𝛿 𝑥̇ $ −0.707 −0.1 𝛿𝑥$ 1
4.10 LAPLACE TRANSFORM SOLUTION OF STATE EQUATIONS
Ø Consider the state and output equation: 𝒙̇ = 𝑨𝒙 + 𝑩𝒖
Ø Take the Laplace transform and solve for 𝑿 𝑠 :

𝑠𝑿 𝑠 − 𝒙 0 = 𝑨𝑿 𝑠 + 𝑩𝑼 𝑠
𝒀 𝑠 = 𝑪𝑿 𝑠 + 𝑫𝑼 𝑠
!" !"
𝑠𝑰 − 𝑨 𝑿 𝑠 = 𝒙 𝟎 + 𝑩𝑼 𝑠 ⇒ 𝑿 𝑠 = 𝑠𝑰 − 𝑨 𝒙 0 + 𝑠𝑰 − 𝑨 𝑩𝑼 𝑠
012(3𝑰!𝑨)
𝑿 𝑠 = 𝒙 0 + 𝑩𝑼(𝑠)
678(3𝑰!𝑨)
Ø By the definition of a transfer, let 𝒙 0 =0. Then use 𝑿 𝑠 in output equation

012(3𝑰!𝑨)
𝒀 𝑠 =𝑪 𝑩𝑼 𝑠 + 𝑫𝑼 𝑠
678(3𝑰!𝑨)
Ø The transfer function is:
𝑌(𝑠) 𝑪 𝑎𝑑𝑗 𝑠𝑰 − 𝑨 𝑩 + 𝐃det(𝑠𝑰 − 𝑨)
𝑇 𝑠 = =
𝑈(𝑠) det(𝑠𝑰 − 𝑨)
4.10 LAPLACE TRANSFORM SOLUTION OF STATE EQUATIONS
Ø The transfer function is:
𝑌(𝑠) 𝑪 𝑎𝑑𝑗 𝑠𝑰 − 𝑨 𝑩 + 𝐃det(𝑠𝑰 − 𝑨)
𝑇 𝑠 = =
𝑈(𝑠) det(𝑠𝑰 − 𝑨)
The roots of the denominator are the poles of the system. Find
poles from
det 𝑠𝑰 − 𝑨 = 0
The system poles equal the eigenvalues of the matrix A!!!
Note that the poles of the transfer function determine the nature
of the transient response of the system.
Example: a)Solve the state equation adj ( sI - A)
( sI - A) =
-1
and the output. sI - A

b)Find the eigenvalues and the system
é s 2 + 9 s + 26 s+9 1ù
poles 1 ê ú
= 3 - 24 s 2
+ 9 s s
é x!1 ù é 0 1 0 ù é x1 ù é0ù s + 9 s 2 + 26 s + 24 ê ú
êë - 24 s - (26 s + 24) s úû
2
ê x! ú = ê 0 0 ú ê ú
1 x2 + 0 e ê ú -t
ê 2ú ê úê ú ê ú
êë x!3 úû êë- 24 - 26 - 9úû êë x3 úû êë1úû
𝑠 ' + 10𝑠 $ + 37𝑠 + 29
é x1 ù é x1 (0) ù é1 ù (𝑠 + 1)(𝑠 + 2)(𝑠 + 3)(𝑠 + 4)
y (t ) = [1 1 0]ê x2 ú ê x (0)ú = ê0ú
ê ú ê 2 ú ê ú 𝑋# 𝑠 2𝑠 $ − 21𝑠 − 24
êë x3 úû êë x3 (0) úû êë2úû 𝑿 𝑠 = 𝑋$ 𝑠 = (𝑠 + 1)(𝑠 + 2)(𝑠 + 3)(𝑠 + 4)
𝑋' 𝑠
𝑠(2𝑠 $ − 21𝑠 − 24)
𝑠𝑿 𝑠 − 𝒙 0 = 𝑨𝑿 𝑠 + 𝑩𝑼 𝑠
(𝑠 + 1)(𝑠 + 2)(𝑠 + 3)(𝑠 + 4)
𝒀 𝑠 = 𝑪𝑿 𝑠 + 𝑫𝑼 𝑠
𝑿 𝑠 = 𝑠𝑰 − 𝑨 SC 𝒙 0 + 𝑠𝑰 − 𝑨 SC 𝑩𝑼 𝑠
𝑋C(𝑠)
d ! eZ[d " eZfdeg
𝒀 𝑠 = 𝑪𝑿 𝑠 = 1 1 0 𝑋L(𝑠) = 𝑋C(𝑠)+ 𝑋[ 𝑠 =
(deZ)(de[)(de\)(deh)
𝑋T(𝑠)
−6.5 19 11.5 0 The pole at -1

𝒀 𝑠 = + − +
𝑠+2 𝑠+3 𝑠+4 𝑠+1 canceled a zero at -1
𝑦 𝑡 = −6.5𝑒 SL. + 19𝑒 ST. − 11.5𝑒 SR.
b) The system poles equal the eigenvalues of the matrix A.
Therefore, both the poles of the system and eigenvalues are -2, -3, and -4.
4.11 TIME DOMAIN SOLUTION OF STATE EQUATIONS
The solution in the time domain is t
x(t ) = e x(0) + ò e
At A ( t -t )
B u (t ) dt
0
.
= 𝜱 𝑡 𝒙 0 + ^ 𝜱 𝑡 − 𝜏 𝑩𝒖 𝜏 𝑑𝜏
A
Zero-input response Zero-state response
Where 𝜱 𝑡 = 𝑒 𝑨. by definition, and which is called the state-transition matrix.

4.11 TIME DOMAIN SOLUTION OF STATE EQUATIONS
.
𝒙(t) = 𝜱 𝑡 𝒙 0 + ^ 𝜱 𝑡 − 𝜏 𝑩𝒖 𝜏 𝑑𝜏
A
For the unforced system (i.e. Zero input)
𝒙(𝑡) = 𝜱 𝑡 𝒙 0
SC 𝒙 U,V W𝑰S𝑨
ℒ𝑥 𝑡 = ℒ[𝜱 𝑡 𝒙 0 ] = 𝑠𝑰 − 𝑨 0 ⟹ ℒ SC 𝑠𝑰 − 𝑨 SC = ℒ SC
YZ[ W𝑰S𝑨
=𝜱 𝑡
Exercise: If u(t) is a unit step find the state-transition matrix and then solve for x(t)
0 1 0
𝒙̇ 𝑡 = 𝒙 𝑡 + 𝑢 𝑡
−8 −6 1
1
𝒙 0 =
0
You can find the solution in your textbook

(see example 4.12 and example 4.13).
CASE STUDY: PHARMACEUTICAL DRUG ABSORPTION
PHARMACEUTICAL DRUG ABSORPTION PROBLEM
We want to describe the distribution of a drug in the body
by dividing the process into compartments: dosage, absorption site,blood,
peripheral compartment, and urine.
Ø Each xi is the amount of

drug in that particular
compartment.
Ø The rate of change of

amount of a drug is equal to
the input flow rate
diminished by the output
flow rate.
Represent the system in state space, where the outputs are the amounts of drug in each compartment.
1. Assume dosage is released at a rate proportional to concentration.
d
x1 = - K1 x1
dt
2. Assume input(/output) flow rate into(/out) any given compartment is proportional to the
concentration of the drug in the previous(/its own) compartment. Then,
d
x2 = K1 x1 - K 2 x2
dt
d
x3 = K 2 x2 - K 3 x3 + K 4 x4 - K 5 x3
dt
d
x4 = K 5 x3 - K 4 x4
dt
d
x5 = K 3 x3
dt
3. Define the state vector as the dosage amount in each compartment. Then,
é x1 ù é- K1 0 0 0 0ù é x1 ù
êx ú ê K - K2 0 0 0ú ê x2 ú
d ê 2
ú ê 1
úê ú
ê x3 ú = ê 0 K2 - ( K3 + K5 ) K 4 0ú ê x3 ú
dt ê ú ê úê ú
ê x4 ú ê 0 0 K5 - K4 0ú ê x4 ú
êë x5 úû êë 0 0 K3 0 0úû êë x5 úû
1 0 0 0 0
0 1 0 0 0
𝒚= 0 0 1 0 0 𝒙
0 0 0 1 0
0 0 0 0 1
CHALLANGE: STORAGE OF WATER IN AQUIFERS (UNDERGROUND WATER SUPPLIES)
An aquifer system consists of a number of interconnected natural storage tanks. Natural water
flows through the sand and sandstone, changing the water levels in the tanks on its way to the
sea. A water conservation policy can be established whereby water is pumped between tanks
to prevent its loss to the sea.
Represent the aquifer system
in state space, where the state
variables and the outputs are
the heads of each tank.
Ø Each qn is the natural water flow to the sea and 𝑞! = 𝐺! (ℎ! − ℎ!"# )
Ø q02 , q03 are flows from the tanks for irrigation, industry, and homes
Ø q21 is created by the water conservation policy to prevent loss to the sea
If h1<H1 then water will be pumped from Tank2 to Tank1
If h1>H1 then water will be pumped back to Tank2 to prevent loss to the sea.
Flow is proportional to the difference between 𝐻# 𝑎𝑛𝑑 ℎ# ∶ 𝑞$# = 𝐺$# 𝐻# − ℎ#
Ø The net flow into a tank is proportional to the rate of change of head in each tank. Thus,
𝑑ℎ!
𝐶! = 𝑞4! − 𝑞(! + 𝑞!5# − 𝑞! + 𝑞 !5# ! − 𝑞!(!"#)
𝑑𝑡
TIME RESPONSE
4.1 INTRODUCTION
(1) We would like to
control/evaluate (2) Can be represented
by transfer function
Linear, Time-invariant System
(3)Should be analyzed for its

• Transient response and
• Steady-state responses
To see if these characteristics yield the
desired behaviour at the output.
4.2 POLES, ZEROS, AND SYSTEM RESPONSE
Output response= Natural response + Forced response
Homogeneous solution.
(If the system is stable, it is also called Particular solution, steady-state response
transient response.) Depends on the input.
Depends only on the system, not the input.
Transfer Numerator The zeros of a transfer function are the values of s

Function Polynomial that cause the transfer function to become zero.
N ( s ) Function
G (s) 
D( s)
Denominator The poles of a transfer function are the values of s
Polynomial that cause the transfer function to become infinite.
Function
Is there a relationship between the pole/zero locations and time response of a system?
Poles and Zeros of a first-order system: An Example
𝑠+2 𝐴 𝐵 2/5 3/5

𝐶 𝑠 = = + = +
𝑠(𝑠 + 5) 𝑠 𝑠 + 5 𝑠 𝑠+5
2 3 −5𝑡
𝑐 𝑡 = + 𝑒
5 5
1. A pole of the input generates the form

of the forced response
2. A pole of the transfer function
generates the form of the natural
response
3. The zeros and poles generate the
amplitudes for both the forced and
natural responses.
• A pole at −𝛼 generates an exponential
response of the form 𝑒 −𝛼𝑡 .
• The farther to the left a pole is on the

negative axis, the faster the
exponential transent response will
decay to zero.
System 1 System 2
System 2 reaches to its steady-state value earlier than the System 1.

(i.e., Its transient response decays faster) . System 2 is faster.
4.3 FIRST-ORDER
1
SYSTEMS
𝑠
𝑎
𝐶 𝑠 =𝑅 𝑠 𝐺 𝑠 =
𝑠(𝑠 + 𝑎)
𝑐 𝑡 = 𝑐𝑓𝑜𝑟𝑐𝑒𝑑 𝑡 + 𝑐𝑛𝑎𝑡𝑢𝑟𝑎𝑙 𝑡 = 1 − 𝑒 −𝑎𝑡
1
When 𝑡 = , → 𝑒 −𝑎𝑡 ȁ𝑡=1/𝑎 = 𝑒 −1 = 0.37
𝑎
𝑐(𝑡)ቚ = 1 − 𝑒 −𝑎𝑡 ቚ 1 = 1 − 0.37 = 0.63

𝑡=1/𝑎 𝑡=𝑎
Time Constant: Time it takes for the step

response to rise to 63% of its final value.
1
=
𝑎
Rise time: Time for the waveform to go Settling Time: Time for the response to reach, and
from 0.1 to 0.9 of its final value. stay within, 2% of its final value.
2.2 4
𝑇𝑟 = = 2.2 𝑇𝑠 = = 4
𝑎 𝑎
Laboratory results of an output
FIRST-ORDER TRANSFER FUNCTIONS VIA TESTING for a step input
What about if it is not possible/practical to obtain a
system’s transfer function analytically?
A possible approach:
• Apply a step input
• Measure the time constant and steady-state
value from the response
• Write the transfer function by using
measured values.
Consider a simple first-order

𝐾
system, 𝐺(𝑠) =  Final Value: 0.72
𝑠+𝑎
whose step response is  0.63 x 0.72 = 0.45
 Time constant is the time where response
𝐾 reaches 0.45→ 𝜏 = 0.13 𝑠𝑒𝑐.
𝐶 𝑠 =𝑅 𝑠 𝐺 𝑠 = 1
𝑠(𝑠 + 𝑎)  Then 𝑎 = 0.13 = 7.7
=
𝐾/𝑎 𝐾/𝑎
−  Forced resonse reaches a steady-state
𝑠 𝑠+𝑎 𝐾
value of 𝑎 =0.72. Then K=5.54
5.54
Identify K and a from  The transfer function, 𝐺(𝑠) = 𝑠+7.7
laboratory testing
4.4 SECOND-ORDER SYSTEMS:INTRODUCTION
The pole locations determine the form of the response!

Poles: Two real at −𝜎1 , −𝜎2

Natural Response: Two
exponentials with time constants
equal to the reciprocal of the pole
locations
𝑐 𝑡 = 𝐾1 𝑒 −𝜎1𝑡 + 𝐾2 𝑒 −𝜎2𝑡
Poles: Two complex at −𝜎𝑑 ± 𝑗𝑤𝑑

Natural Response: Damped
sinusoid with an exponential 𝑐 𝑡 =
𝐴𝑒 −𝜎𝑑𝑡 cos(𝑤𝑑 𝑡 − 𝜙)
Poles: Two complex at ±𝑗𝑤1
Natural Response: Undamped sinuoid
with radian frequency equal to the
imaginary part of the poles
𝑐 𝑡 = 𝐴 cos(𝑤1 𝑡 − 𝜙)
Poles: Two real at −𝜎1

Natural Response: an exponential + the
product of time, t, and an exponential.
𝑐 𝑡 = 𝐾1 𝑒 −𝜎1𝑡 + 𝐾2 𝑡𝑒 −𝜎2𝑡
4.5 THE GENERAL SECOND-ORDER SYSTEM
Natural Frequency, 𝑤𝑛 : The frequency of oscillation of the system without damping.
A quantity that compares the exponential decay frequency of the

Damping Ratio,  :
envelope to the natural frequency. It is constant regardless of the time
scale of the response.
𝐸𝑥𝑝𝑜𝑛𝑒𝑛𝑡𝑖𝑎𝑙 𝑑𝑒𝑐𝑎𝑦 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 1 𝑁𝑎𝑡𝑢𝑟𝑎𝑙 𝑝𝑒𝑟𝑖𝑜𝑑 (𝑠𝑒𝑐𝑜𝑛𝑑𝑠)

𝜁= =
𝑟𝑎𝑑 2𝜋 𝐸𝑥𝑝𝑜𝑛𝑒𝑛𝑡𝑖𝑎𝑙 𝑡𝑖𝑚𝑒 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡
𝑁𝑎𝑡𝑢𝑟𝑎𝑙 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 ( )
𝑠𝑒𝑐𝑜𝑛𝑑
𝑏
𝐺 𝑠 = 2
𝑠 + 𝑎𝑠 + 𝑏
𝑤𝑛 and 
can be used to 𝑤𝑛2
describe the 𝐺 𝑠 = 2
𝑠 + 2𝜁𝑤𝑛 𝑠 + 𝑤𝑛2
characteristics
of the response. The poles of a transfer
function are
𝑠1,2 = −𝜁𝑤𝑛 ± 𝑤𝑛 𝜁 2 − 1
SECOND-ORDER RESPONSE AS A FUNCTION OF DAMPING RATIO
𝑏
𝐺 𝑠 = 2
𝑤𝑛2
𝐺 𝑠 = 2
The poles of a transfer

function are
𝑠1,2 = −𝜁𝑤𝑛 ± 𝑤𝑛 𝜁 2 − 1
𝜎 𝑤𝑑
Attenuation Damped frequency
4.6 UNDERDAMPED SECOND-ORDER SYSTEMS
𝑤𝑛2
𝐺 𝑠 = 2
𝜁
𝑤𝑛2 𝐾1 𝐾2 𝑠 + 𝐾3 1 𝑠 + 𝜁𝑤𝑛 + 2 𝑤𝑛 1 − 𝜁 2
1−𝜁
𝐶 𝑠 = = + 2 = −
𝑠(𝑠 2 + 2𝜁𝑤𝑛 𝑠 + 𝑤𝑛2 ) 𝑠 𝑠 + 2𝜁𝑤𝑛 𝑠 + 𝑤𝑛2 𝑠 (𝑠 + 𝜁𝑤𝑛 )2 +𝑤𝑛2 (1 − 𝜁 2 )
1
𝑐 𝑡 =1− 𝑒 −𝜁𝑤𝑛𝑡 cos(𝑤𝑛 1 − 𝜁 2 t − ϕ)
1 − 𝜁2
𝜁
ϕ= tan−1 ( )
1− 𝜁2
The lower the value of 𝜁 ,

the more oscillatory the response.
Rise time: The time for the waveform to go
from 0.1 to 0.9 of its final value.
𝜋−𝜃
𝑇𝑟 =
𝑤𝑑
Peak time: The time required to reach the
first, or maximum, peak.
𝜋
𝑇𝑝 =
𝑤𝑛 1 − 𝜁 2
Percent overshoot: The amount that the

waveform overshoots the final or steady-state
value at the peak time, expressed as a
percentage of the steady-state value.
−𝜁𝜋
൘ 𝑐𝑚𝑎𝑥 −𝑐𝑓𝑖𝑛𝑎𝑙
1−𝜁2
%𝑂𝑆 = 𝑒 x 100, %𝑂𝑆 = x100
𝑐𝑚𝑎𝑥
𝑤𝑛2
𝐺 𝑠 = 2 C s = G s R(s)
𝑠 + 2𝜁𝑤𝑛 𝑠 + 𝑤𝑛2 Settling time: The time required for the
transient’s damped oscillations to reach and
The poles of a transfer function are stay within ±2% of the steady-state value.
𝑠1,2 = −𝜁𝑤𝑛 ± 𝑤𝑛 𝜁 2 − 1 4
𝑇𝑠 =
𝜁𝑤𝑛
Representation of a pole in s-domain. Lines of constant peak time Tp, settling time, Ts, and
percent overshoot %OS.
𝑤𝑛2 𝜋
𝐺 𝑠 = 2 C s = G s R(s) Peak time: 𝑇𝑝 =
𝑠 + 2𝜁𝑤𝑛 𝑠 + 𝑤𝑛2 𝑤𝑛 1−𝜁 2 −𝜁𝜋
൘
1−𝜁2
The poles of a transfer function are Percent overshoot: %𝑂𝑆 = 𝑒 x 100
𝑠1,2 = −𝜁𝑤𝑛 ± 𝑤𝑛 𝜁 2 − 1 4
Settling time: 𝑇𝑠 =
𝜁𝑤𝑛
𝑤𝑛2
𝐺 𝑠 = 2 C s = G s R(s)
𝑠1,2 = −𝜁𝑤𝑛 ± 𝑤𝑛 𝜁 2 − 1
Step responses of different 2nd-order systems having same

(a) real parts; (b) imaginary parts; (c) damping ratios.
Except for certain applications where oscillations cannot be tolerated, it is desirable that the
transient response be sufficiently fast & sufficiently damped.
Thus, for a desirable transient response of a second-order system, the damping ratio must
be between 0.4 and 0.8.
( Small values of  (<0.4) yield excessive overshoot.

Large values of  (>0.8) responds sluggishly. )
Note that, fastest response without oscillation is the critically damped response.
Problem: Given the pole plot shown in figure below, find 𝜁, 𝑤𝑛 , 𝑇𝑝 , %𝑂𝑆, 𝑎𝑛𝑑𝑇𝑠 .
𝑠1,2 = −𝜁𝑤𝑛 ± 𝑤𝑛 𝜁 2 − 1 = −3 ± 𝑗7
7
𝜁 = cos 𝜃 = cos[tan−1 ( )] = 0.394
3
𝑤𝑛 = 72 + 32 = 7.616
𝜋 𝜋
𝑇𝑝 = = = 0.449
𝑤𝑛 1−𝜁 2 7
−𝜁𝜋
൘
%𝑂𝑆 = 𝑒 1−𝜁 2 x 100 = 26%
4 4
𝑇𝑠 = = = 1.33 𝑠𝑒𝑐𝑜𝑛𝑑𝑠
𝜁𝑤𝑛 3
Problem: Given the system shown in figure below, find J and D to yield 20%
overshoot and a settling time of 2 seconds for a step input torque T(t).
1ൗ
𝐽
𝐺 𝑠 =
𝐷 𝐾
𝑠2 + 𝑠 +
𝐽 𝐽
𝐷
𝐾 2𝜁𝑤𝑛 =4=
From the transfer function G(s): 𝑤𝑛 = 𝑎𝑛𝑑 𝐽
𝐽
𝐷 𝐽
2𝜁𝑤𝑛 = 𝜁=
4
=2
𝐽 𝜁=2 = 0.456
𝐽 𝐾
2𝑤𝑛 𝐾
𝐽
4 = 0.052 ⟹
From the problem statement : 𝑇𝑠 = =2 𝐾
𝜁𝑤𝑛
−𝜁𝜋
൘
1−𝜁 2 𝜁=0.456
𝐷 = 1.04
%𝑂𝑆 = 𝑒 x 100 = 20%
𝐽 = 0.26
4.7 SYSTEM RESPONSE WITH ADDITIONAL POLES
 Up to now, we analyzed systems with one or two poles.
 Note that the formulae describing OS%, Ts and Tp are derived only for a system with two
complex poles and no zeros!
 What about if a system has more than two poles or has zeros ?
One of the Approaches:

 Approximate it as a second-order system that
has just two complex dominant poles.
Those closed‐loop poles that have dominant effects on

the transient‐response behavior are called dominant
closed‐loop poles.
Consider a three-pole system with complex
poles and a third pole on the real axis. Step
response of a such system is
𝐴 𝐵 𝑠 + 𝜁𝑤𝑛 + 𝐶𝑤𝑑 𝐷
𝐶 𝑠 = + +
𝑠 𝑠 + 𝜁𝑤𝑛 2 + 𝑤𝑑 2 𝑠 + 𝛼𝑟
𝑐 𝑡 = 𝐴𝑢 𝑡 + 𝑒 −𝜁𝑤𝑛 𝑡 (𝐵 cos 𝑤𝑑 𝑡 + 𝐶 sin 𝑤𝑑 𝑡) + 𝐷𝑒 −𝛼𝑟 𝑡
𝛼𝑟 = 𝛼𝑟1 𝑎𝑛𝑑 𝛼𝑟 ≅ 𝜁𝑤𝑛 𝛼𝑟 = 𝛼𝑟2 𝑎𝑛𝑑 𝛼𝑟 ≫ 𝜁𝑤𝑛 𝛼𝑟 = ∞

𝐶 𝑠 = + +
The pure exponential will die out

much more rapidly than the
second-order underdamped step
response.
OS%,Ts,Tp can be calculated from

the second-order underdamped
𝛼𝑟 = 𝛼𝑟2 𝑎𝑛𝑑 𝛼𝑟 ≫ 𝜁𝑤𝑛 step respose component.
𝐶 𝑠 = + +
The real pole’s transient respose

will not decay to insignificance at
Assume that the exponential decay
the settling time generated by the
is negligible after five time
second-order pair.
constants.
In this case, the exponential
Thus, if 5𝛼𝑟 > 𝜁𝑤𝑛 then
decay is significant, and the
represent the system by its
system cannot be represented as
dominant second-order pair of poles
𝛼𝑟 = 𝛼𝑟1 𝑎𝑛𝑑 𝛼𝑟 ≅ 𝜁𝑤𝑛 a second order system.
𝐶 𝑠 = + +
 What about the magnitude D of the

pure exponential decay, 𝑒 −𝛼𝑟 𝑡 ?
 Can it be so large that its contribution  In a three-pole system with dominant
is not negligible? second-order poles and no zeros,
the residue of the third pole,D,
will actually decrease in magnitude
as the third pole is moved farther into the
No. Because left half plane. Therefore, D cannot be too
large.
4.8 SYSTEM RESPONSE WITH ZEROS
If we add a zero at –a to the transfer
function(s+a)T(s)
The zero
 affects the amplitude of a response
 but does not affect the nature of the response—

exponential, damped sinusoid, and so on.
The closer the zero is to the dominant poles

(located at −1 ± 𝑗2.8), the greater its effect on
the transient response.
An interesting phenomenon occurs if the zero is in the right half-plane (i.e a is negative.)
 Initially, the response may turn toward the negative direction even though the final value is
positive.
 A system that exhibits this phenomenon is known as a nonminimum-phase system.
 If a motorcycle or airplane was a nonmimimum-phase system, it would initially veer left when
commanded to steer right.
TIME RESPONSE
4.1 INTRODUCTION
(1) We would like to
control/evaluate (2) Can be represented
by transfer function
Linear, Time-invariant System
(3)Should be analyzed for its

• Transient response and
• Steady-state responses
To see if these characteristics yield the
desired behaviour at the output.
4.2 POLES, ZEROS, AND SYSTEM RESPONSE
Output response= Natural response + Forced response
Homogeneous solution.
(If the system is stable, it is also called Particular solution, steady-state response
transient response.) Depends on the input.
Depends only on the system, not the input.
Transfer Numerator The zeros of a transfer function are the values of s

Function Polynomial that cause the transfer function to become zero.
N ( s) Function
G ( s) 
D( s)
Denominator The poles of a transfer function are the values of s
Polynomial that cause the transfer function to become infinite.
Function
Is there a relationship between the pole/zero locations and time response of a system?
Poles and Zeros of a first-order system: An Example
1. A pole of the input generates the form

of the forced response
2. A pole of the transfer function
generates the form of the natural
response
3. The zeros and poles generate the
amplitudes for both the forced and
natural responses.
• A pole at generates an exponential
response of the form
• The farther to the left a pole is on the

negative axis, the faster the
exponential transent response will
decay to zero.
System 1 System 2
System 2 reaches to its steady-state value earlier than the System 1.

(i.e., Its transient response decays faster) . System 2 is faster.
4.3 FIRST-ORDER
1
SYSTEMS
𝑠
𝑎
𝑠(𝑠 + 𝑎)
When /
𝑐(𝑡) = 1−𝑒 = 1 − 0.37 = 0.63

/
Time Constant: Time it takes for the step

response to rise to 63% of its final value.

Rise time: Time for the waveform to go Settling Time: Time for the response to reach, and
from 0.1 to 0.9 of its final value. stay within, 2% of its final value.
 
Laboratory results of an output
FIRST-ORDER TRANSFER FUNCTIONS VIA TESTING for a step input
What about if it is not possible/practical to obtain a
system’s transfer function analytically?
A possible approach:
• Apply a step input
• Measure the time constant and steady-state
value from the response
• Write the transfer function by using
measured values.
Consider a simple first-order

system,  Final Value: 0.72
whose step response is  0.63 x 0.72 = 0.45
 Time constant is the time where response
𝐾 reaches 0.45→ 𝜏 = 0.13 𝑠𝑒𝑐.
𝑠(𝑠 + 𝑎)  Then 𝑎 = = 7.7
.
=
𝐾/𝑎 𝐾/𝑎
−  Forced resonse reaches a steady-state
𝑠 𝑠+𝑎 value of =0.72. Then K=5.54
.
Identify K and a from  The transfer function, 𝐺(𝑠) =
.
laboratory testing
The pole locations determine the form of the response!

Poles: Two real at ,

Natural Response: Two
exponentials with time constants
equal to the reciprocal of the pole
locations
Poles: Two complex at

Natural Response: Damped
sinusoid with an exponential
Poles: Two complex at
Natural Response: Undamped sinuoid
with radian frequency equal to the
imaginary part of the poles
Poles: Two real at

Natural Response: an exponential + the
product of time, t, and an exponential.
4.5 THE GENERAL SECOND-ORDER SYSTEM
Natural Frequency, The frequency of oscillation of the system without damping.
A quantity that compares the exponential decay frequency of the

Damping Ratio,  :
envelope to the natural frequency. It is constant regardless of the time
scale of the response.
𝑏
𝐺 𝑠 =
and 
can be used to 𝑤
describe the 𝐺 𝑠 =
𝑠 + 2𝜁𝑤 𝑠 + 𝑤
characteristics
of the response. The poles of a transfer
function are

,
SECOND-ORDER RESPONSE AS A FUNCTION OF DAMPING RATIO
The poles of a transfer

function are

,
Attenuation Damped frequency

4.6 UNDERDAMPED SECOND-ORDER SYSTEMS
𝑤
𝐺 𝑠 =
𝑠 + 2𝜁𝑤 𝑠 + 𝑤
𝜁
𝑤 𝐾 𝐾 𝑠+𝐾 1 𝑠 + 𝜁𝑤 + 𝑤 1−𝜁
1−𝜁
𝐶 𝑠 = = + = −
𝑠(𝑠 + 2𝜁𝑤 𝑠 + 𝑤 ) 𝑠 𝑠 + 2𝜁𝑤 𝑠 + 𝑤 𝑠 (𝑠 + 𝜁𝑤 ) +𝑤 (1 − 𝜁 )
1
𝑐 𝑡 = 1− 𝑒 cos(𝑤 1 − 𝜁 t − ϕ)
1−𝜁
𝜁
ϕ = tan ( )
1−𝜁
The lower the value of ,

the more oscillatory the response.
Rise time: The time for the waveform to go
from 0.1 to 0.9 of its final value.
Peak time: The time required to reach the

first, or maximum, peak.
Percent overshoot: The amount that the

waveform overshoots the final or steady-state
value at the peak time, expressed as a
percentage of the steady-state value.

Settling time: The time required for the

transient’s damped oscillations to reach and
The poles of a transfer function are stay within ±2% of the steady-state value.

,
Representation of a pole in s-domain. Lines of constant peak time Tp, settling time, Ts, and
percent overshoot %OS.
Peak time:

The poles of a transfer function are Percent overshoot:

, Settling time:
𝑤
𝐺 𝑠 = C s = G s R(s)
𝑠 + 2𝜁𝑤 𝑠 + 𝑤

,
Step responses of different 2nd-order systems having same

(a) real parts; (b) imaginary parts; (c) damping ratios.
Except for certain applications where oscillations cannot be tolerated, it is desirable that the
transient response be sufficiently fast & sufficiently damped.
Thus, for a desirable transient response of a second-order system, the damping ratio must
be between 0.4 and 0.8.
( Small values of  (<0.4) yield excessive overshoot.

Large values of  (>0.8) responds sluggishly. )
Note that, fastest response without oscillation is the critically damped response.
Problem: Given the pole plot shown in figure below, find

,

Problem: Given the system shown in figure below, find J and D to yield 20%
overshoot and a settling time of 2 seconds for a step input torque T(t).

=4=
From the transfer function G(s):

= =
=
From the problem statement :

=0.456
4.7 SYSTEM RESPONSE WITH ADDITIONAL POLES
 Up to now, we analyzed systems with one or two poles.
 Note that the formulae describing OS%, Ts and Tp are derived only for a system with two
complex poles and no zeros!
 What about if a system has more than two poles or has zeros ?
One of the Approaches:

 Approximate it as a second-order system that
has just two complex dominant poles.
Those closed loop poles that have dominant effects on

the transient response behavior are called dominant
closed loop poles.
𝛼 =𝛼 𝑎𝑛𝑑 𝛼 ≅ 𝜁𝑤 𝛼 =𝛼 𝑎𝑛𝑑 𝛼 ≫ 𝜁𝑤 𝛼 =∞
The pure exponential will die out

much more rapidly than the
second-order underdamped step
response.
OS%,Ts,Tp can be calculated from

the second-order underdamped
𝛼 =𝛼 𝑎𝑛𝑑 𝛼 ≫ 𝜁𝑤 step respose component.
The real pole’s transient respose

will not decay to insignificance at
Assume that the exponential decay
the settling time generated by the
is negligible after five time
second-order pair.
constants.
In this case, the exponential
Thus, if then
decay is significant, and the
represent the system by its
system cannot be represented as
dominant second-order pair of poles
𝛼 =𝛼 𝑎𝑛𝑑 𝛼 ≅ 𝜁𝑤 a second order system.
 What about the magnitude D of the

pure exponential decay, ?
 Can it be so large that its contribution  In a three-pole system with dominant
is not negligible? second-order poles and no zeros,
the residue of the third pole,D,
will actually decrease in magnitude
as the third pole is moved farther into the
No. Because left half plane. Therefore, D cannot be too
large.
4.8 SYSTEM RESPONSE WITH ZEROS
If we add a zero at –a to the transfer
function(s+a)T(s)
The zero
 affects the amplitude of a response
 but does not affect the nature of the response—

exponential, damped sinusoid, and so on.
The closer the zero is to the dominant poles

(located at , the greater its effect on
the transient response.
An interesting phenomenon occurs if the zero is in the right half-plane (i.e a is negative.)
 Initially, the response may turn toward the negative direction even though the final value is
positive.
 A system that exhibits this phenomenon is known as a nonminimum-phase system.
 If a motorcycle or airplane was a nonmimimum-phase system, it would initially veer left when
commanded to steer right.
Block Diagram Algebra
5.1 INTRODUCTION
Many systems are composed
of multiple subsytems, as in the
figure.
A graphical tool can help us

to visualize the model of a
system and evaluate the
mathematical relationships
between their subsytems, using
their transfer functions.
We will now
 examine some common
topologies for interconnecting
subsytems and
 derive the single transfer
function representation.
5.2 BLOCK DIAGRAMS
It represents the mathematical relationships between the elements of the system.
The transfer function of each component is placed in box, and the input-output relationships
between components are indicated by lines and arrows.
New schematic
elements:
Summing junction
and pickoff points.
CASCADE FORM
Cascaded
subsytems :
Equivalent transfer
function :
PARALLEL FORM
Paralle
subsytems :
Equivalent transfer
function :
FEEDBACK FORM
Feedback
control system:
𝐶𝐶 𝑠𝑠 = 𝐸𝐸 𝑠𝑠 𝐺𝐺 𝑠𝑠
𝐸𝐸 𝑠𝑠 = 𝑅𝑅(𝑠𝑠) ∓ 𝐶𝐶 𝑠𝑠 𝐻𝐻(𝑠𝑠)
Simplified model:
𝐶𝐶 𝑠𝑠 = (𝑅𝑅(𝑠𝑠) ∓ 𝐶𝐶 𝑠𝑠 𝐻𝐻(𝑠𝑠)) 𝐺𝐺(𝑠𝑠)
1 ± 𝐻𝐻 𝑠𝑠 𝐺𝐺 𝑠𝑠 𝐶𝐶 𝑠𝑠 = 𝐺𝐺 𝑠𝑠 𝑅𝑅(𝑠𝑠)
Equivalent transfer 𝐺𝐺(𝑠𝑠)

𝐶𝐶 𝑠𝑠 = 𝑅𝑅(𝑠𝑠)
function : 1 ± 𝐻𝐻 𝑠𝑠 𝐺𝐺 𝑠𝑠
MOVING BLOCKS TO CREATE FAMILIAR FORMS
Moving a block to
the left past a
summing junction
Moving a block to
the right past a
summing junction
Moving a block to
the left past a
pickoff point
Moving a block to
the right past a
pickoff point
Problem: Reduce the block diagram to a single transfer function
Y (s)
T (s) =
R( s)
2s + 4
2
T (s) = s
2s + 4 2s + 4 Y
1+
s2
s 2 + 2s + 4
2s + 4
T (s) = 2
s + 2s + 4
Problem:
Reduce the
block diagram
to a single
transfer
function
STABILITY
6.1 INTRODUCTION
Homogeneous solution. Particular solution, steady-state

Depends only on the system, not response
the input. Depends on the input.
There are many definitions for stability,

depending upon the kind of system or the point of view.
In this course we limit ourselves to

linear, time-invariant systems.
Stability Definition Using the Natural Response:
STABLE UNSTABLE MARGINALLY

STABLE
• If the natural • If the natural • If the natural
response response grows response neither
approaches zero as without bound as decays nor grows
time approaches time approaches but remains
infinity infinity constant or oscillate
as time approaches
infinity
Stability Definition Using the Total Response (BIBO) :
• A system is stable if every bounded

STABLE input yields a bounded output.
• A system is unstable if any bounded

UNSTABLE input yields an unbounded output.
• A signal is bounded if there is a finite value such that the signal

magnitude never exceeds.
• Physically, an unstable system can cause damage to the system, to

adjacent property, or to human life.
Y(s)
R(s) 1 This system is said to be unstable
𝐻 𝑠 = 2 because the output response y(t)
𝑠 −1
1 1 goes to infinity as time approaches
𝑌 𝑠 =𝐻 𝑠 𝑅 𝑠 = 2 = infinity.
𝑠 −1 𝑠 − 1 (𝑠 + 1)
0.5 0.5
𝑌(𝑠) = − The pole in the right half-plane
𝑠−1 𝑠+1 (RHP) causes unstability.
𝑦 𝑡 = 0.5𝑒 𝑡 − 0.5𝑒 −𝑡
The pole at s=1 yields pure The pole at s=-1 yields pure
exponentially increasing response. exponential decay.
(This natural reponse term (This natural reponse term

approaches infinity as time approaches zero as time
approaches infinity) approaches infinity)
The poles in the left halfplane (LHP) yield either pure exponential
decay or damped sinusoidal natural responses.
(These natural reponses decay to zero as time approaches infinity)

The poles in the RHP yield either pure exponentially increasing or
exponentially increasing sinusoidal natural responses.
(These natural reponses approach infinity as time approaches infinity)

A system that has imaginary axis poles of multiplicity 1 yields pure
sinusoidal oscillations as a natural response.
These responses neither increase nor decrease in amplitude.
Thus, marginally stable systems have closed-loop transfer functions

with only maginary axis poles of multiplicity 1 and poles in the LHP.
NAURAL RESPONSE(/STABILITY) VS. POLE LOCATION
THE STABILITY OF A CONE.
If the cone is resting on its base and is tipped slightly, it returns to its
original equilibrium position. This response is STABLE.
If the cone rests on its side and is displaced slightly, it rolls with no tendency
to leave the position on its side. This response is NEUTRAL.
If the cone is placed on its tip and released, it falls onto its side. This
response is UNSTABLE.
A stability of a dynamic systems is defined in similar manner. The response to

a displacement, or initial conditions, will result in either a decreasing, neutral,
or increasing response.
TACOMA NARROWS BRIDGE
As oscillation begins
At catastrophic failure
on November 7, 1940.
EXAMPLE: BI-BO CRITERION
Q. If the step function is applied at the input of
continuous system and the output remains below a
certain level for all time, is the system stable?
Q.If the step function is applied at the input of
continuous system and the output remains below a
certain level for all time, is the system stable?
A.The system is not necessarily stable since the output must

be bounded for every bounded input.
A bounded output to one specific bounded input does not
ensure stability.
Q. If the step function is applied at the

input of continuous system and the
output is of the form y = t, is the
system stable or unstable?
Q. If the step function is applied at the

input of continuous system and the
output is of the form y = t, is the
system stable or unstable?
A. This system is unstable since a bounded

input produced an unbounded output.
Example:
Determine in each case if the set of roots
represents, stable, marginally stable, or
unstable systems.
a) -1, -2 e) -2 + j, -2 - j, 2j, -2j

b) -1, +1 f) 2,-1,-3
c) -3,-2,0 g) -6,-4,7
d) -1 + j, -1- j h) -2 + 3j, -2 -3j, -2
Determine in each case if the set of roots
represents, stable, marginally stable, or
unstable systems.
a) -1, -2 e) -2 + j, -2 - j, 2j, -2j

b) -1, +1 f) 2,-1,-3
c) -3,-2,0 g) -6,-4,7
d) -1 + j, -1- j h) -2 + 3j, -2 -3j, -2
STABLE SYSTEM
A necessary and sufficient condition for a feedback system to
be stable is that all the poles of the transfer function have
negative real parts.
This means that all the poles are in the left half-plane.
How we can determine the pole locations?

Use computers or Routh-hurwitz criterion.
DETERMINING THE ROUTH TABLE
Consider the denominator of the closed loop transfer function:
q(s)  an  sn  an1sn1  an2sn2  an3sn3   a1s  a0
First arrange the coefficients of the characteristic equation in two

rows, beginning with the first and second coefficients and followed by
the even-numbered and odd-numbered coefficients:
sn : an an2 an4
sn1 : an1 an3 an5

Then add subsequent
rows to complete the sn : an an  2 an  4 
Routh table:
s n 1 : an1 an 3 a n 5 
s n2 : bn1 bn 3 bn5 
s n 3 : cn1 cn 3 cn  5 
   
2
s :
?! s :
1
s :
0
hn 1
Compute elements for the
3rd row: sn : an an2 an4
sn1 : an1 an3 an5
1 an an2 sn2 : bn1 bn3 bn5
bn1  ,
an1 an1 an3 sn3 : c n1 c n3 c n5
1 an an4
bn3  ,
an1 an1 an5 s2 : * *
1 an1 an3 s1 : *
c n1 
bn1 bn1 bn3 s0 : *

EXAMPLE 1: MAKE THE ROUTH TABLE
INTERPRETING THE ROUTH TABLE
Necessary Condition: All the coefficients are

positive and nonzero.
The number of roots of the polynomial that

are in the right half-plane is equal to the
number of sign changes in the first column.
Thus, a system is stable if there are no sign

changes in the first column of the Routh
table.
EXAMPLE:
Given the characteristic equation,
a(s)  s 6  4s5  3s 4  2s3  s 2  4s  4

is the system described by this characteristic equation stable?
Answer:
All the coefficients are positive and nonzero.
Therefore, the system satisfies the necessary condition for

stability.
We should determine whether are there any sign changes in the

first column of the Routh array.
a(s)  s 6  4s5  3s 4  2s3  s 2  4s  4
6
s : 1 3 1 4
s5 : 4 2 4 0
s4 : ? ? ?
s3 : ? ? ?
s2 : ? ?
1
s : ? ?
s0 : ?
a(s)  s  4s  3s  2s  s  4s  4
6 5 4 3 2
s6 : 1 3 1 4
s5 : 4 2 4 0
1 2  3  4  10 5
  
s4 : 5 2 0 4 4 4 2
s3 : ? ? ?
2 1 4  1 4 0
s : ? ?   0
1
4 4
s : ? ?
s0 : ? 1 0  4  4  16
  4
4 4
a(s)  s 6  4s5  3s 4  2s3  s 2  4s  4
6
s : 1 3 1 4
s5 : 4 2 4 0
s4 : 5 2 0 4
4 0  25 2
s 3 : 2  12 5 0 
52
 (2)  2
s2 : ? ?
1
s : ? ? 4 4  45 2 32  20  12
  
0 52 5 5
s : ?
a(s)  s  4s  3s  2s  s  4s  4
6 5 4 3 2
s6 : 1 3 1 4
s5 : 4 2 4 0
The elements of
s4 : 52 0 4 the 1st column
s3 : 2  12 5 0 are not all positive:
s2 : 3 4
the system is unstable
s1 :  76 15 0
s0 : 4
6.3 ROUTH-HURWITZ CRITERION: SPECIAL CASES
Two Special cases:
(1) The Routh table have a zero only in the first column of a row
(2) The Routh table have an entire row that consists of zeros.
Case 1: Zero Only in the First Column
If the first element of a row is zero, division by zero is required !!!!
To avoid this phenomenon, an epsilon, , is assigned to replace the

zero in the first column.
The value of  is then allowed to approach zero, after which the

signs of the entries can be determined.
Problem: Determine the stability of the closed-loop transfer function
10
𝑇 𝑠 =
𝑠 5 + 2𝑠 4 + 3𝑠 3 + 6𝑠 2 + 5𝑠 + 3
If  is chosen positive
very small value then
there are 2 sign changes
(from s3 row to s2 row
and from s2 row to s1
row) in the first column.
Hence the system is

unstable and has two
poles in the Right half-
plane.
Case 2: Entire Row is Zero
Look at the following example that demonstrates how to
construct and interpret the Routh table.
Problem: Determine the number of right half plane poles in the
10
closed-loop transfer function 𝑇 𝑠 = 5
𝑠 + 7𝑠 4 + 6𝑠 3 + 42𝑠 2 + 8𝑠 + 56
 Identify the row immediately above the row of zeros. (Row 4)

 Then, form an auxiliary polynomial, using the entries in that row as coefficients.
( 𝑃 𝑠 = 𝑠 4 + 6𝑠 2 + 8 )
𝑑𝑃(𝑠)
 Next, diffentiate the polynomial with respect to s. ( = 4𝑠 3 + 12𝑠 + 0 )
𝑑𝑠
 Use coefficients of obtained polynomial to replace the row of zeros. (4, 12, 0)
 After calculations, there are no sign changes in the first column. Hence,
the system is stable. There are no right half plane poles.
If all the coefficients in any row are zero,
Whether
(1) the roots are symmetrical and real,

(2) the roots are symmetrical and
imaginary,
(3) the roots are quadrantal.
Problem: For the transfer function 20
𝑇 𝑠 = 8
𝑠 + 𝑠 7 + 12𝑠 6 + 22𝑠 5 + 39𝑠 4 + 59𝑠 3 + 48𝑠 2 + 38𝑠 + 20
tell how many poles are in the RHP, in the LHP, and on the jw-axis.
𝑑𝑃(𝑠)
 𝑃 𝑠 = 𝑠 4 + 3𝑠 2 + 2 ; = 4𝑠 3 + 6𝑠 + 0
𝑑𝑠
 Two sign changes in the first column, therefore 2 poles in the RHP.
 No sign changes from s4 row to s0 row. All four of the poles on the jw-
axis.
 Remaining two poles should be in the LHP.
STABILITY DESIGN VIA ROUTH-HURWITZ
Problem: Find the range of gain, K, that will cause the system to
be stable, unstable, and marginally stable.
1
𝐶(𝑠) 𝐾 𝐾
𝑠(𝑠 + 7)(𝑠 + 11)
𝑇 𝑠 = = = 3
𝑅(𝑠) 1 + 𝐾 1 𝑠 + 18𝑠 2 + 77𝑠 + 𝐾
𝑠(𝑠 + 7)(𝑠 + 11)
𝑄 𝑠 = 𝑠 3 + 18𝑠 2 + 77𝑠 + 𝐾 = 0
Characteristic Equation
(denominator of the closed-loop
transfer function)
𝑄 𝑠 = 𝑠 3 + 18𝑠 2 + 77𝑠 + 𝐾 = 0
• All terms in the first column

0<K<1386 are positive
• 3 poles in the LHP and
stable
• Two sign changes.

K>1386 • 2 RHP and 1 LHP poles and
unstable.
• An entire row of zero.

𝑃 𝑠 = 18𝑠 2 + 1386
𝑑𝑃(𝑠)
K=1386 = 36𝑠 + 0
𝑑𝑠
• 2 poles on the jw axis and
marginally stable
Problem: Consider a feedback system given below. Determine
the range of K over which the system is stable.
𝑠+1
𝑌(𝑠) 𝐾 𝐾(𝑠 + 1)
𝑠(𝑠 − 1)(𝑠 + 6)
𝑇 𝑠 = = = 3
𝑅(𝑠) 1 + 𝐾 𝑠 + 1 𝑠 + 5𝑠 2 + 𝐾 − 6 𝑠 + 𝐾
𝑠(𝑠 − 1)(𝑠 + 6)
𝑄 𝑠 = 𝑠 3 + 5𝑠 2 + 𝐾 − 6 𝑠 + 𝐾 = 0
Characteristic Equation
(denominator of the closed-loop
transfer function)
Q( s )  s 3  5s 2  ( K  6) s  K
The corresponding Routh array is: Therefore, the system is

stable if and only if
s3 : 1 K 6 4K  30
 0 and K  0
s2 : 5 K 5
s1 : (4 K  30) 5  K  7.5 and K  0
s0 : K
 K  7.5

STABILITY VERSUS TWO PARAMETER RANGE
Problem: Consider a Proportional-Integral (PI) control such as:
Find the range of the controller gains ( K , K1 ) so that the PI feedback

system is stable.
Characteristic equation
 K1  1
Q( s)  1   K   0
 s  ( s  1)( s  2)
Q( s )  s 3  3s 2  (2  K ) s  K1  0
s  3s  ( 2  K )s  K1  0
3 2
The corresponding Routh array is: For stability, we must have:
s3 : 1 2 K
s2 : 3 K1
K1
s : ( 6  3K  K1 ) 3
1 K1  0 and K  2
3
s0 : K1
STEADY-STATE ERRORS Assoc.Dr. Güleser K Demir
7.1 INTRODUCTION
Error Signal
Reference Input Output
(Set point) Controller Plant
Control System Objective:

To make the system OUTPUT and the desired REFERENCE as close as possible,
i.e, to make the ERROR as small as possible.
Definition:
Steady-state error is the difference between the input and the output as 𝑡 → ∞.
Consider a position control system where the outout position follows the input
commanded position. Inputs vary with target type (/application).
Inputs
Application to Stable Systems
Steady-state error can be calculated after the system reaches

its steady-state.
Unstable systems represent loss of control in the steady-state

and are not acceptable for use at all.
You should not calculate steady-state errors for unstable

systems. Because the results are wrong.
Evaluating Steady-State Errors
For a Step Output 1 has zero steady- For a Output 1 has zero steady-state
input state error, and ramp error,
input
Output 2 has a finite steady- Output 2 has a finite steady-
state error,𝑒2 ∞ . state error,𝑒2 ∞ .
Output 3 has an infinite steady-

state error.
7.2 STEADY-STATE ERROR FOR UNITY FEEDBACK SYSTEMS
𝐶(𝑠) 𝐺(𝑠)
𝑇 𝑠 = =
𝑅(𝑠) 1 + 𝐺(𝑠)
𝐶 𝑠 = 𝑅 𝑠 𝑇(𝑠)
𝐸 𝑠 =𝑅 𝑠 −𝐶 𝑠 ⇒ 𝐸 𝑠 = 𝑅 𝑠 [1 − 𝑇 𝑠 ] 𝑒 ∞ = lim 𝑒 𝑡 = lim 𝑠𝐸(𝑠)

𝑡→∞ 𝑠→0
𝑒 ∞ = lim𝑠 𝑅 𝑠 [1 − 𝑇 𝑠 ]
𝑠→0
Example: Find the steady-state error if 𝑇 𝑠 = 𝐶(𝑠) = 5
and the input is a unit step.
𝑅(𝑠) 𝑠 2 + 7𝑠 + 10
𝐶 𝑠 = 𝑅 𝑠 𝑇(𝑠)
𝑠 2 + 7𝑠 + 5
𝐸 𝑠 =𝑅 𝑠 −𝐶 𝑠 ⇒ 𝐸 𝑠 =𝑅 𝑠 1−𝑇 𝑠 =
𝑠(𝑠 2 + 7𝑠 + 10)
𝑒 ∞ = lim 𝑒 𝑡 = lim 𝑠𝐸(𝑠)
𝑡→∞ 𝑠→0
𝑠 2 +7𝑠+5 1
𝑒 ∞ = lim𝑠 𝑅 𝑠 [1 − 𝑇 𝑠 ] = lim 𝑠 =
𝑠→0 𝑠→0 𝑠(𝑠 2 +7𝑠+10) 2
Steady-State Error in Terms of G(s)
𝐶 𝑠 = 𝐸 𝑠 𝐺(𝑠)
𝐸 𝑠 =𝑅 𝑠 −𝐶 𝑠
𝐸 𝑠 = 𝑅 𝑠 − 𝐸 𝑠 𝐺(𝑠)
𝐸 𝑠 + 𝐸 𝑠 𝐺(𝑠) = 𝑅 𝑠
𝑅(𝑠)
𝐸 𝑠 =
1 + 𝐺(𝑠)
𝑠𝑅(𝑠)
𝑒 ∞ = lim 𝑒 𝑡 = lim 𝑠𝐸 𝑠 = lim
𝑡→∞ 𝑠→0 𝑠→0 1 + 𝐺(𝑠)
Note that division by s in the freq.

domain integration in the time
domain
Unit Step Input , R(s)=1/s
𝑠𝑅(𝑠)
𝑒 ∞ = lim 𝑒 𝑡 = lim 𝑠𝐸 𝑠 = lim If G(s) has at • 𝐺 𝑠 = 𝑛(𝑠+𝑧1)(𝑠+𝑧2)…
𝑡→∞ 𝑠→0 𝑠→0 1 + 𝐺(𝑠) 𝑛≥1
least one 𝑠 (𝑠+𝑝1 )(𝑠+𝑝2 )…
1 pole at the • 𝑠→0

lim 𝐺 𝑠 = ∞ then 𝑒 ∞ = 0
𝑠( ) 1 origin
𝑒 ∞ = lim 𝑠 = • Zero error
𝑠→0 1 + 𝐺(𝑠) 1 + lim 𝐺(𝑠)
𝑠→0
If G(s) has • 𝑛=0

(𝑠+𝑧1 )(𝑠+𝑧2 )… 𝑧1 𝑧2 …
DC Gain of the forward not pole at • lim 𝐺 𝑠 = lim
𝑠→0 𝑠→0 (𝑠+𝑝1 )(𝑠+𝑝2 )…
=
𝑝1 𝑝2 …
transfer function the origin • Finite error
(𝑠+𝑧1 )(𝑠+𝑧2 )…
If G(s) has at •𝐺 𝑠 =
𝑠 𝑛 (𝑠+𝑝1 )(𝑠+𝑝2 )…
𝑛≥2
least two pole at • lim 𝑠𝐺 𝑠 = ∞ then 𝑒 ∞ =0
𝑠→0
the origin • Zero error
Ramp Input , R(s)=1/s2
𝑠𝑅(𝑠)
𝑒 ∞ = lim 𝑠𝐸 𝑠 = lim • 𝑛=1
𝑠→0 𝑠→0 1 + 𝐺(𝑠) If G(s) has only
(𝑠+𝑧 )(𝑠+𝑧 )… 𝑧 𝑧 …
one pole at the • lim 𝑠𝐺 𝑠 = lim 𝑠 𝑠 (𝑠+𝑝1 )(𝑠+𝑝2
𝑠→0 𝑠→0 1 2 )…
= 𝑝1 𝑝2 …
1 2
1
𝑠( 2 ) origin • Finite error
𝑒 ∞ = lim 𝑠
𝑠→0 1 + 𝐺(𝑠)
•𝑛=0
1 1 If G(s) has no (𝑠+𝑧 )(𝑠+𝑧 )…
• lim 𝑠𝐺 𝑠 = lim 𝑠 (𝑠+𝑝1 )(𝑠+𝑝2 )… = 0
𝑒 ∞ = lim = poles at the origin 𝑠→0 𝑠→0 1 2
𝑠→0 𝑠 + 𝑠𝐺(𝑠) lim 𝑠𝐺(𝑠)
𝑠→0 • Infinite error
(𝑠+𝑧1 )(𝑠+𝑧2 )…
If G(s) has at least •𝐺 𝑠 =
𝑠 𝑛 (𝑠+𝑝1 )(𝑠+𝑝2 )…
𝑛≥3
three pole at the • lim 𝑠 2 𝐺 𝑠 = ∞ then 𝑒 ∞ =0
Parabolic Input , R(s)=1/s3 origin 𝑠→0
• Zero error
𝑠𝑅(𝑠)
𝑒 ∞ = lim 𝑠𝐸 𝑠 = lim
𝑠→0 𝑠→0 1 + 𝐺(𝑠)
If G(s) has only • 𝑛=2
(𝑠+𝑧 )(𝑠+𝑧2 )… 𝑧 𝑧 …
1 two pole at the • lim 𝑠 2 𝐺 𝑠 = lim 𝑠 2 𝑠2 (𝑠+𝑝1
𝑠→0 𝑠→0 1 )(𝑠+𝑝2 )…
= 𝑝1 𝑝2 …
𝑠( 3 ) origin
1 2
𝑒 ∞ = lim 𝑠 • Finite error

𝑠→0 1 + 𝐺(𝑠)
1 1 If G(s) has one or •𝑛≤1

𝑒 ∞ = lim 2 = • lim 𝑠 2 𝐺 𝑠 = 0
𝑠→0 𝑠 + 𝑠 2 𝐺(𝑠) lim 𝑠 2 𝐺(𝑠)
𝑠→0
less poles at the 𝑠→0
origin • Infinite error
For a step 1
7.3 STATIC ERROR CONSTANTS input 𝑒 ∞ =
1 + lim 𝐺(𝑠)
AND SYSTEM TYPE 𝑠→0
Position Constant: 𝑲𝒑 = 𝒍𝒊𝒎 𝑮(𝒔)

𝒔→𝟎
1
𝑒 ∞ =
1 + 𝐾𝑝
For a ramp 1 1
input 𝑒 ∞ = lim =
𝑠→0 𝑠𝐺(𝑠) lim 𝑠𝐺(𝑠)
𝑠→0
Velocity Constant: 𝑲𝒗 = 𝒍𝒊𝒎 𝒔𝑮(𝒔)

𝒔→𝟎
1
𝑒 ∞ =
𝐾𝑣
For a 1
parabolic 𝑒 ∞ =
lim 𝑠 2 𝐺(𝑠)
input 𝑠→0
Acceleration Constant: 𝑲𝒂 = 𝒍𝒊𝒎 𝒔𝟐 𝑮(𝒔)

𝒔→𝟎
1
𝑒 ∞ =
𝐾𝑎
SYSTEM TYPE
(𝑠 + 𝑧1 )(𝑠 + 𝑧2 ) …
𝐺 𝑠 = 𝑛
𝑠 (𝑠 + 𝑝1 )(𝑠 + 𝑝2 ) …
System type:
 The number of pole at the origin, or
 The number of pure integration in the
forward path.
7.4 STEADY-STATE ERROR SPECIFICATIONS
If a control system has The system is stable.
the specification Kv =
1000, we can draw
several conclusions:
The system is of Type 1, since only Type 1 systems have
Kv's that are finite constants. Recall that Kv = 0 for Type
0 systems, whereas Kv =  for Type 2 systems.
A ramp input is the test signal. Since Kv is specified as a

finite constant, and the steady-state error for a ramp
input is inversely proportional to Kv we know the test
input is a ramp.
The steady-state error between the input ramp and the

output ramp is 1/Kv per unit of input slope.
What information is The system is stable.
contained in the
specification 𝐾𝑝 = 1000?
The system is of Type 0, since only Type 0

system has a finite 𝐾𝑝 . Type 1 and Type 2
systems have 𝐾𝑝 = ∞
The input test signal is a step, since 𝐾𝑝 is

specified.
The steady-state error per unit step is

1 1 1
𝑒 ∞ = = =
1 + 𝐾𝑝 1 + 1000 1001
Example: Find the value of K so that there is e()=0.1 in the steady-state
Since the system is Type 1, the error stated in the problem must apply to a ramp input;
only a ramp yields a finite error in a Type 1 system.
1
𝑒 ∞ = lim 𝑒 𝑡 = lim 𝑠𝐸 𝑠 = = 0.1
𝑡→∞ 𝑠→0 𝐾𝑣
5x𝐾
𝐾𝑣 = 10 = lim 𝑠𝐺 𝑠 = ⇒ 𝐾 = 672
𝑠→0 6x7x8
7.5 STEADY-STATE ERROR FOR DISTURBANCES
Disturbance/Unwanted input
𝑒 ∞ = lim 𝑠𝐸 𝑠
𝑠→0
𝑠 𝑠𝐺2 𝑠
𝑒 ∞ = lim 𝑅 𝑠 − lim 𝐷(𝑠)
𝑠→0 1 + 𝐺1 (𝑠)𝐺2 (𝑠) 𝑠→0 1 + 𝐺1 (𝑠)𝐺2 (𝑠)
𝑒 ∞ = 𝑒𝑅 ∞ + 𝑒𝐷 (∞)
C s = R s − E(s)
Steady-state error due
𝐶 𝑠 = 𝐸 𝑠 𝐺1 𝑠 𝐺2 𝑠 + 𝐷(𝑠)𝐺2 (𝑠) to disturbance.
In which conditions this
𝑅 𝑠 − 𝐸(𝑠) = 𝐸 𝑠 𝐺1 𝑠 𝐺2 𝑠 + 𝐷(𝑠)𝐺2 (𝑠) error reduces?
(𝐺1 𝑠 𝐺2 𝑠 + 1)𝐸(𝑠) = 𝑅 𝑠 + 𝐷(𝑠)𝐺2 (𝑠) Assume a step disturbance, D(s)=1/s

𝑠𝐺2 𝑠 1 1
𝑒𝐷 ∞ = lim =−
1 𝐺2 𝑠 𝑠→0 1 + 𝐺1 (𝑠)𝐺2 (𝑠) 𝑠 1
𝐸 𝑠 = 𝑅 𝑠 − 𝐷(𝑠) lim + lim 𝐺1 (𝑠)
𝑠→0 𝐺2 𝑠 𝑠→0
1 + 𝐺1 (𝑠)𝐺2 (𝑠) 1 + 𝐺1 (𝑠)𝐺2 (𝑠)
The s.s error produced by a step disturbance
Transfer function Transfer function can be reduced by increasing the dc gain of
relating E(s) to R(s) relating E(s) to D(s) 𝐺1 (𝑠) or decreasing the dc gain of 𝐺2 (𝑠).
Example: Find the steady-state error component due to a step disturbance for the system given
below.
The system is stable.

𝑠𝐺2 𝑠 1 1
C s = R s − E(s) 𝑒𝐷 ∞ = lim =−
𝑠→0 1 + 𝐺1 (𝑠)𝐺2 (𝑠) 𝑠 1
lim + lim 𝐺1 (𝑠)
𝐶 𝑠 = 𝐸 𝑠 𝐺1 𝑠 𝐺2 𝑠 + 𝐷(𝑠)𝐺2 (𝑠) 𝑠→0 𝐺2 𝑠 𝑠→0
𝑅 𝑠 − 𝐸(𝑠) = 𝐸 𝑠 𝐺1 𝑠 𝐺2 𝑠 + 𝐷(𝑠)𝐺2 (𝑠) 1 1

𝑒𝐷 ∞ = − =−
0 + 1000 1000
(𝐺1 𝑠 𝐺2 𝑠 + 1)𝐸(𝑠) = 𝑅 𝑠 + 𝐷(𝑠)𝐺2 (𝑠)
1 𝐺2 𝑠
𝐸 𝑠 = 𝑅 𝑠 − 𝐷(𝑠)
1 + 𝐺1 (𝑠)𝐺2 (𝑠) 1 + 𝐺1 (𝑠)𝐺2 (𝑠)
𝑒 ∞ = lim 𝑠𝐸 𝑠
𝑠→0
𝑠 𝑠𝐺2 𝑠
𝑒 ∞ = lim 𝑅 𝑠 − lim 𝐷(𝑠)
𝑠→0 1 + 𝐺1 (𝑠)𝐺2 (𝑠) 𝑠→0 1 + 𝐺1 (𝑠)𝐺2 (𝑠)
7.6 STEADY-STATE ERROR FOR NONUNITY FEEDBACK SYSTEMS
If the system feedback is not unity, this signal is

not error!!!
It is actuating signal.
Recall: E s = R s − C(s) (error)
𝐸𝑎 s = R s − H(s)C(s) (actuating signal)
To find error form a unity feedback system

as shown in the next slide.
1 3 5
𝐺𝑒 (𝑠)
Now, write error

E s = R s − C(s)
2 4 and proceed with
𝐺(𝑠)
𝐺𝑒 𝑠 = 1+𝐺 𝑠 𝐻 𝑠 −𝐺(𝑠)
in the forward path.

Example: Find the system type, the appropriate error constant associated with the system type,
and the steady-state error for a unit step input.
𝐺 𝑠 =
100 Do not say the system is TYPE 1 by just looking 𝐺 𝑠 .
𝑠(𝑠 + 10) Convert the system into an equivalent unity feedback system and
check the equivalent forward transfer function.
1
𝐻 𝑠 =
(𝑠 + 5)
𝐺(𝑠) 100(𝑠 + 5) The system is TYPE 0, since there are no
𝐺𝑒 𝑠 = = 3
1 + 𝐺 𝑠 𝐻 𝑠 − 𝐺(𝑠) 𝑠 + 15𝑠 2 − 50𝑠 − 400 pure integrations (/no pole at the origin).
100x5 5
𝐾𝑝 = lim 𝐺𝑒 𝑠 = =−
𝑠→0 −400 4
1 1 The negative value implies that the output is
𝑒 ∞ = = = −4 larger than the input.
1 + 𝐾𝑝 1 − 5/4
7.7 SENSITIVITY
The degree to which changes in system parameters affect system transfer functions, and
hence performance, is called sensitivity.
A system with zero sensitivity is ideal.
𝐾
𝐹= 𝐾 = 10, 𝑎 = 100 ⇒ 𝐹 = 0.091
𝐾+𝑎
300 − 100 0.32 − 0.091
=2 = −0.65
100 0.091
(𝟐𝟎𝟎% fractional change) (−65% fractional change)
𝐾 = 10, 𝑎 = 300 ⇒ 𝐹 = 0.031

The function F has
reduced sensitivity to
changes in parameter a.
Formal Definition:
System sensitivity is the ratio of the fractional change in the function to the
fractional change in the parameter as the fractional change of the parameter
approaches zero.
𝐹𝑟𝑎𝑐𝑡𝑖𝑜𝑛𝑎𝑙 𝑐ℎ𝑎𝑛𝑔𝑒 𝑖𝑛 𝑡ℎ𝑒 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛, 𝐹 ∆𝐹/𝐹

𝑆𝑃𝐹 = lim = lim
∆𝑃→0 𝐹𝑟𝑎𝑐𝑡𝑖𝑜𝑛𝑎𝑙 𝑐ℎ𝑎𝑛𝑔𝑒 𝑖𝑛 𝑡ℎ𝑒 𝑝𝑎𝑟𝑎𝑚𝑒𝑡𝑒𝑟, 𝑃 ∆𝑃→0 ∆𝑃/𝑃
𝑃 ∆𝐹
𝑆𝑃𝐹 = lim
∆𝑃→0 𝐹 ∆𝑃
𝑃 𝜕𝐹
𝑆𝑃𝐹 =
𝐹 𝜕𝑃
Example: Calculate the sensitivity of the closed loop transfer function to changes in the
parameter a. How would you reduce the sensitivity.
𝐾
𝑠(𝑠 + 𝑎) 𝐾
𝑇 𝑠 = = 2
𝐾 𝑠 + 𝑎𝑠 + 𝐾
1+
𝑠(𝑠 + 𝑎)
𝑎 𝜕𝑇 𝑎 −𝐾𝑠 −𝑎𝑠 Increasing K reduces the sensitivity!

𝑆𝑎𝑇 = = =
𝑇 𝜕𝑎 𝐾 (𝑠 2 + 𝑎𝑠 + 𝐾)2 𝑠 2 + 𝑎𝑠 + 𝐾
𝑠 2 + 𝑎𝑠 + 𝐾
In this example, sensitivity is a function of s.

Therefore, sensitivity values are different
for different frequencies.
Example: Find the sensitivity of the steady-state error to changes in parameter K and
parameter a for the system shown below with a step input.
𝐾
𝐾𝑝 = 𝑙𝑖𝑚 𝐺 𝑠 = 𝑙𝑖𝑚
𝑠→0 𝑠→0 (𝑠+𝑎)(𝑠+𝑏)
1 1 𝑎𝑏
𝑒 ∞ = = =
1 + 𝐾𝑝 1 + 𝐾 𝑎𝑏 + 𝐾
𝑎𝑏
𝑎 𝜕𝑒 𝑎 𝑎𝑏 + 𝐾 𝑏 − 𝑎𝑏 2 𝐾
𝑆𝑎𝑒 = = =
𝑒 𝜕𝑎 𝑎𝑏 (𝑎𝑏 + 𝐾)2 𝑎𝑏 + 𝐾
𝑎𝑏 + 𝐾
𝐾 𝜕𝑒 𝐾 −𝑎𝑏 −𝐾
𝑆𝐾𝑒 = = =
𝑒 𝜕𝐾 𝑎𝑏 (𝑎𝑏 + 𝐾)2 𝑎𝑏 + 𝐾
𝑎𝑏 + 𝐾
ROOT LOCUS PLOTS
Goal: Learn a specific technique which shows how
changes in one of a system’s parameter
(usually the controller gain, K)
will modify the location of the closed-loop poles
in the s-domain.
Table1: Pole locations for different
K values
Pole plot from the Table-1.
Pole plot from the Table-1.
Root-locus.
8.3 PROPERTIES OF THE ROOT_LOCUS 𝑌(𝑠) 𝑝(𝑠) 𝐾𝐺(𝑠)
𝑇 𝑠 = = =
𝑅(𝑠) 𝑞(𝑠) 1 + 𝐾𝐺(𝑠)
Closed-loop transfer function
Loop transfer function

𝑞 𝑠 = 1 + 𝐾𝐺 𝑠 = 0 (or Open-loop Tr. Func.)
Characterictic equation
K is a variable parameter,
0≤𝐾≤∞
𝑞 𝑠 = 1 + 𝐾𝐺 𝑠 ⇒ 𝐾𝐺 𝑠 = −1
𝐾𝐺(𝑠) ⌊𝐾𝐺 𝑠 = −1 + 𝑗0
The roots of characteristic equations 𝐾𝐺(𝑠) =1,

must satisfy these magnitude and
angle requirements. ⌊𝐾𝐺 𝑠 = 180 + 𝑘 360°, 𝑘 = 0, ±1, ±2, ±3, …
Example: Check whether -2+3j point can be a pole of this system for some value of K?
If 𝑠! = −2 + 𝑗3 is a closed-loop pole for some value of gain,

Then the angles of the zeros minus the angles of poles must
equal an odd multiple of 180°.
𝜃! + 𝜃" − 𝜃# − 𝜃$ = 56.31° + 71.57° − 90° − 108.43° = −70.55°
Therefore 𝑠! = −2 + 𝑗3 is not a point on the root locus,

or alternatively −2 + 𝑗3 is not a closed-loop pole for any gain.
Example: Check whether −𝟐 +
𝒋 𝟐
𝟐
point can be a pole of this system for some value of K?
𝒋 𝟐
If 𝑠! = −𝟐 + is a closed-loop pole for some value of gain,
𝟐
Then the angles of the zeros minus the angles of poles must
equal an odd multiple of 180°.
𝜃! + 𝜃" − 𝜃# − 𝜃$ = 180°
𝒋 𝟐
Therefore 𝑠! = −𝟐 + 𝟐
is a point on the root locus for some
value of gain K.
2
1 Π 𝑝𝑜𝑙𝑒 𝑙𝑒𝑛𝑔𝑡ℎ𝑠 1.22
𝐾= = = 2 = 0.33
𝐺 𝑠 𝐻(𝑠) Π 𝑧𝑒𝑟𝑜 𝑙𝑒𝑛𝑔𝑡ℎ𝑠 (2.12 )(1.22)
At this value of K, given point is on

the root-locus
8.4 SKETCHING THE ROOT_LOCUS
PROPERTIES OF THE ROOT LOCUS
1. Number of branches of the root locus equals the number of closed-loop poles.
(Branch: the path that one pole traverses as the gain is varied.)
2. Symmetry: The root locus is symmetrical about the real axis.
3. Real-axis segments: The root locus exists to the left of an odd number of real-
axis, finite open-loop poles and/or finite open-loop zeros
4. Starting and ending points: The root locus
Begins (zero gain) at the finite and infinite poles of G(s)H(s) and
Ends (infinite gain) at the finite and infinite zeros of G(s)H(s).
𝑁$ (𝑠)
𝐾𝐺(𝑠) 𝐾 𝐾𝑁$ (𝑠)𝐷% (𝑠)
𝐷$ (𝑠)
𝑇 𝑠 = = =
1 + 𝐾𝐺 𝑠 𝐻(𝑠) 𝑁$ (𝑠)𝑁% (𝑠) 𝐷$ 𝑠 𝐷% 𝑠 + 𝐾𝑁$ (𝑠)𝑁% (𝑠)
1+𝐾
𝐷$ (𝑠)𝐷% (𝑠)
As 𝐾 → 0 𝐾𝑁$ (𝑠)𝐷% (𝑠)

𝑇 𝑠 =
𝐷$ 𝑠 𝐷% 𝑠 + 𝜖
Closed-loop system poles approaches the poles of 𝐺 𝑠 𝐻(𝑠)
Therefore, root locus begins at the poles of 𝐺 𝑠 𝐻(𝑠)
As 𝐾 → ∞ 𝐾𝑁$ (𝑠)𝐷% (𝑠)

𝑇 𝑠 =
𝜖 + 𝐾𝑁$ (𝑠)𝑁% (𝑠)
Closed-loop system poles approaches the zeros of 𝐺 𝑠 𝐻(𝑠)
Therefore, root locus ends at the zeros of 𝐺 𝑠 𝐻(𝑠)

5. Behavior at infinity.
The root locus approaches straight lines as asymptotes as the locus approaches infinity.
Real-axis intercept of ∑ 𝑓𝑖𝑛𝑖𝑡𝑒 𝑝𝑜𝑙𝑒𝑠 − ∑ 𝑓𝑖𝑛𝑖𝑡𝑒 𝑧𝑒𝑟𝑜𝑠

𝜎& =
asymptotes #𝑓𝑖𝑛𝑖𝑡𝑒 𝑝𝑜𝑙𝑒𝑠 − #𝑓𝑖𝑛𝑖𝑡𝑒 𝑧𝑒𝑟𝑜𝑠
Angle of the asymptotes (2𝑘 + 1)𝜋

𝜃& = 𝑘 = 0, ±1, ±2, ±3
with respect to the positive #𝑓𝑖𝑛𝑖𝑡𝑒 𝑝𝑜𝑙𝑒𝑠 − #𝑓𝑖𝑛𝑖𝑡𝑒 𝑧𝑒𝑟𝑜𝑠
extension of the real axis .
6. Real-Axis Breakaway and Break-in Points
These points are the points in

Break-away point the s-domain where multiples
Break-away point roots of the characteristic
equation of the feedback
control occur.
These points correspond to

intersection points on the root
locus.
To find the break-in and break-away points take the derivative of gain and solve it.
89
8:
=0
; ;
where 𝐾 = −< :
= −= : >(:)
7. The jw-Axis Crossings
To find the jw-axis crossing, we can use the Routh-Hurwitz criterion, as follows:
Forcing a row of zeros in the Routh table will yield the gain; going back one row to the
even polynomial equation and solving for the roots yields the frequency at the imaginary-
axis crossing.
ANGLE OF DEPARTURE The angle of locus departure from a pole is
the difference between the net angle due to all
other poles and zeros and the criterion angle of
±180° (2k + 1),
ANGLE OF ARRIVAL
The angle of locus arrival at a zero is
the difference between the net angle due to all
other poles and zeros and the criterion angle of
±180° (2k + 1),
RULE #1
Assuming n poles and m zeros for G(s)H(s):
The n branches of the root locus start at the n poles.

m of these n branches end on the m zeros
The n-m other branches terminate at infinity along asymptotes.
First step: Draw the n poles and m zeros of G(s)H(s) using x and o
respectively
APPLYING STEP #1
Draw the n poles and m zeros of G(s)H(s)
using x and o respectively.
1
G(s )H (s ) =
s(s + 1)(s + 2)
3 poles:
p1 = 0; p2 = -1; p3 = -2
No zeros
RULE #2
The loci on the real axis are to the left of an ODD number of REAL poles
and REAL zeros of G(s)H(s)
Second step: Determine the loci on the real axis.

Choose a arbitrary test point. If the TOTAL number of both real poles and
zeros is to the RIGHT of this point is ODD, then this point is on the root locus
APPLYING STEP #2
Determine the loci on the real axis:
Choose a arbitrary test point.

If the TOTAL number of both real poles
and zeros is to the RIGHT of this point is
ODD, then this point is on the root locus
RULE #3
å p -åzi i
sa= n
n-m
m
± 180o (2l + 1)
fl =
n-m
Third step: Determine the n - m asymptotes of the root loci. Locate s = 𝜎N on the real
axis. Compute and draw angles. Draw the asymptotes using dash lines.
APPLYING STEP #3
Determine the n - m asymptotes:
Locate s = 𝜎& on the real axis:
p1 + p2 + p3 0 - 1 - 2
s a = 3 - 0 = 3 = -1
Compute and draw angles:
± 180 (2l + 1) l = 0,1, 2, !
fl =
n-m
ì ± 1800 (2 ´ 0 + 1)
f
ï 0 = = ± 60 0
ï 3-0
Þí
ïf = ± 180 (2 ´1 + 1) = ±1800
0
ïî 1 3-0
Draw the asymptotes using dash lines.
APPLYING STEP #3
Determine the n - m asymptotes:
Locate s = α on the real axis:
p1 + p2 + p3 0 - 1 - 2
s =a = = = -1
3-0 3
Compute and draw angles:
± 180 (2l + 1) l = 0,1, 2, !
fl =
n-m
ì ± 1800 (2 ´ 0 + 1)
f
ï 0 = = ± 60 0
ï 3-0
Þí
ïf = ± 180 (2 ´1 + 1) = ±1800
0
ïî 1 3-0
Draw the asymptotes using dash lines.
RULE #4
Given the characteristic equation is KG(s)H(s) = -1
The breakpoints are the closed-loop poles that satisfy:

dK
=0
ds
Fourth step: Find the breakpoints. Express K such as:
-1
K= .
G(s )H (s )
Set dK/ds = 0 and solve for the poles.
APPLYING STEP #4
Find the breakpoints.
Express K such as:

-1
K= = - s (s + 1)(s + 2 )
G( s )H ( s )
K = - s 3 - 3s 2 - 2 s
Set dK/ds = 0 and solve for the poles.
- 3s 2 - 6s - 2 = 0
s1 = -1.5774, s2 = -0.4226
It is not chosen because it is not on the root-locus segments of real axis

APPLYING STEP #4
Find the breakpoints.
Express K such as:
-1
K= = - s (s + 1)(s + 2 )
G( s )H ( s )
K = - s 3 - 3s 2 - 2 s
Set dK/ds = 0 and solve for the
poles.
- 3s 2 - 6s - 2 = 0
s1 = -1.5774, s2 = -0.4226
RECALL RULE #1
The n branches of the root locus start at the n poles.

m of these n branches end on the m zeros
The n-m other branches terminate at infinity along
asymptotes.
Last step: Draw the n-m branches that terminate at infinity

along asymptotes
APPLYING LAST STEP
Draw the n-m branches that terminate

at infinity along asymptotes
POINTS ON BOTH ROOT LOCUS & IMAGINARY AXIS?
Points on imaginary axis satisfy:

s = jw jω?
Points on root locus satisfy:
1 + KG(s )H (s ) = 0
Substitute s=jω into the characteristic

equation and solve for ω. - jω
w = 0 or w = ± 2
EXAMPLE: Sketch the root locus and find the following.
a) The exact point and gain where the locus crosses jw axis
b) The breakaway point on the real axis,
c) The range of K within which the system is stable.
𝐾(𝑠 " − 4𝑠 + 20)
𝐺 𝑠 𝐻 𝑠 =
(𝑠 + 2)(𝑠 + 4)
𝑃𝑜𝑙𝑒𝑠: 𝑝! = −2 , 𝑝" = −4
𝑍𝑒𝑟𝑜𝑠: 𝑧! = 2 + 𝑗4 , 𝑧" = 2 − j4
𝑁𝑜 𝑎𝑠𝑦𝑚𝑝𝑡𝑜𝑡𝑒𝑠. ( #𝑝𝑜𝑙𝑒𝑠 − #𝑧𝑒𝑟𝑜𝑠 = 0)

PARAMETER DESIGN
PARAMETER DESIGN EXAMPLE -I
PARAMETER DESIGN EXAMPLE -II
How can we obtain a root locus for

variations of the value of p1?
10
𝐾𝐺 𝑠 𝐻 𝑠 = Problem! P1 is not a multiplying factor of the function.
(𝑠 + 2)(𝑠 + 𝑝! )
Solution: Create an equivalent system whose denominator is 1 + 𝑝! 𝐺 𝑠 𝐻(𝑠)
𝐾𝐺(𝑠) 10 10
𝑇 𝑠 = = =
1 + 𝐾𝐺 𝑠 𝐻(𝑠) 𝑠 " + 𝑝! + 2 𝑠 + 2𝑝! + 10 𝑠 " + 2𝑠 + 10 + 𝑝! (𝑠 + 2)
10
"
𝑇 𝑠 = 𝑠 + 2𝑠 + 10
𝑝 (𝑠 + 2) Now, sketch root locus for this function.
1+ " !
𝑠 + 2𝑠 + 10
10
𝑠 " + 2𝑠 + 10
𝑇 𝑠 =
𝑝 (𝑠 + 2)
1+ " !
𝑠 + 2𝑠 + 10
Open-loop transfer function.

Apply all root-locus rules to this function.
BODE PLOTS
1
BODE PLOTS
Semilog plots of the magnitude (in decibels) and
phase (in degrees) of a transfer function versus
frequency in a simple and oftentimes approximate
manner.
Therefore, it consists of two separate plots:
1. |H(jw)| versus w
2. Phase angle of H(jw) versus w.
2
The gain magnitude is many times expressed in terms
of decibels (dB)
dB = 20 log10 A
where A is the amplitude or gain
a decade is defined as any 10-to-1 frequency range

an octave is any 2-to-1 frequency range
20 dB/decade = 6 dB/octave
3
Real, First-Order Poles and Zeros
Consider the following transfer function where all the poles and zeros
are real and first-order
K ( s  z1 ) K ( j  z1 )
H (s)   H ( j ) 
s ( s  p1 ) j ( j  p1 )
The first step is to put H(jw) in a standard form as:
Kz1 (1  j z1 ) K 0 (1  j z1 )
H ( j )  
p1 ( j )(1  p1 ) ( j )(1  j p1 )
j
K0 1  z1  1 K0 1 
j j
H ( j )   ( 1  900  1 )
z1
 900 1  j p 11
 1  j p 1
4
K0 1 j
H ( j ) 
z1
 1 j
p1
 ( )   1  90 0   1  tan 1 ( / z1 )  90 0  tan 1 ( / p1 )
The Bode diagram consists of plotting |H(jw)| and θ(w) as

functions of w.
5
The amplitude of H(jw) in decibels is
AdB  20 log10 H ( j )
For the transfer function H(s) in the example
K 0 1  j z1
AdB  20 log10
 1  j p 1
 20 log10 K 0  20 log10 1  j z1  20 log10   20 log10 1  j p1
The key point is to plot each term separately and then combine the
separate plots graphically.
6
General Form Of Transfer Function
K ( j ) 1 (1  j / z1 )[1  j 21 / k  ( j / k ) 2 ]
H ( ) 
(1  j / p1 )[1  j 2 2 / n  ( j / n ) 2 ]
K ( s ) 1 (1  s / z1 )[1  21s / k  ( s / k ) 2 ]
H ( s ) s  j 
(1  s / p1 )[1  2 2 s / n  ( s / n ) 2 ]
Four simple terms constitute the transfer function :
(1) Constant K
(2) ( s ) 1
(3)(1  s / a ) 1 includes 1 /(1  s / p1 ) or (1  s / z1 )
(4)[1  a1s  a2 s 2 ]1
7
Constant Terms
The plot of 20log10K is a horizontal line because K is not a function of

frequency. The value of this term is positive if K>1, zero for K=1, and
negative for K<1.
8
Pole/Zero At The Origin
Zero at the origin: The plot of
20log10w is a straight line having a
slope of 20dB/decade that intersects
the 0 dB axis at w=1.
It has constant angle 90.
Pole at the origin: The plot of

-20log10w is a straight line having a
slope of -20dB/decade that
intersects the 0 dB axis at w=1.
It has constant angle -90.
9
Simple Zero
j  0,   0
H dB  20 log10 1   20log101  0 as   0   
z1   tan 1     45,   z1
 z1  90,   

j 
H dB  20 log10 1   20log10 as   
z1 z1
10
Magnitude Approximation Of Bode Plot
ω H(ω)=|1+jω/z|(dB) Bode Plot(dB)

(Exact) (Approximate)
0.1z 0.0432 0
z 3.01 0
10z 20.04 20
11
Phase Approximation Of Bode Plot
ω φ(ω)=tan-1(ω/z)(deg) Bode Plot(deg)

(Exact) (Approximate)
0.1z 5.71 0
z 45 45
10z 84.29 90
12
Quadratic Pole/Zero
2
j 2 2  j 
H dB  20 log10 1      0 as   0
n  n 
2
j 2 2  j  
H dB  20 log10 1      40 log10 as   
n 
 n  n
 0,   0
1 2 2 / n 
   tan    90,   n
1   / n 
2 2
 180, 13 
TABLE 1.A Summary Of Bode Straight-line Magnitude And
Phase Plots
Factor Magnitude phase
K
14
TABLE 1.B Summary Of Bode Straight-line Magnitude And
Phase Plots
( j ) N
15
TABLE 1.C Summary Of Bode Straight-line Magnitude And
Phase Plots
1
( j ) N
16
Table 1.D Summary Of Bode Straight-line Magnitude And
Phase Plots
 1  j 
N
Factor  
 z 
Magnitude Phase
17
Table 1.E Summary Of Bode Straight-line Magnitude And
Phase Plots
1
Factor
(1  j / p ) N
Magnitude Phase
18
Table 1.F Summary Of Bode Straight-line Magnitude And
Phase Plots
2N
 2 j  j 
Factor 1      
 n  n 
Magnitude Phase
19
Table 1.G Summary Of Bode Straight-line Magnitude And
Phase Plots
1
Factor
[1  2 j / k  ( j / k ) ]
2 N
Magnitude Phase
20
Example 1
Construct the Bode plots for the transfer function.
200 j
H ( ) 
( j  2)( j  10)
Solution:
10 j
H ( ) 
(1  j / 2)(1  j / 10)
10 j
 (90  tan 1  / 2  tan 1  / 10)
1  j / 2 1  j / 10
21
The magnitude and phase are
j
H dB  20 log10 10  20 log10 j  20 log10 1 
2
j
 20 log10 1 
10
 
  90  tan 1
 tan 1
2 10
22
EXAMPLE 1
23
Example 2
Obtain the Bode plots for
j  10
H ( ) 
j ( j  5) 2
Solution:
0.4(1  j / 10)
H ( ) 
j (1  j / 5) 2
j j
H dB  20 log10 0.4  20 log10 1   20 log10 j  40 log10 1 
10 5
1  1 
  0  tan  90  2 tan
10 5
24
25
26
Example 3
Draw the Bode plots for
s 1
H(s)  2
s  60 s  100
1 / 100(1  j )
H ( ) 
1  j 6 / 10  ( j / 10) 2
j 6  2
H dB  20 log10 100  20 log10 1  j  20 log10 1  
10 100
1  6 / 10 
1 
  0  tan   tan 
1   / 100 
2
27
28
29
Example 4
Given the Bode plot in Figure 1, obtain the transfer
function H().,
30
Solution:
40  20 log10 K  log10 K  2
K  102  100
100 j
H ( ) 
(1  j / 1)(1  j / 5)(1  j / 20)
j104

( j  1)( j  5)( j  20)
104 s
H( s)  , s  j
( s  1)( s  5)( s  20)
31
GAIN MARGIN AND PHASE MARGIN
Gain margin:
The gain margin is the number of dB that is |G(jw)| below 0 dB

at the phase crossover frequency (ø=-180º).
Phase margin:
The phase margin is the number of degrees the phase of that is

above -180º at the gain crossover frequency. It can also be
defined as the angle where the phase can be increased before the
closed loop system becomes unstable.
32
GAIN MARGIN AND PHASE MARGIN
Magnitude
(dB)
20dB
0dB
-20dB Gain Crossover

Gain margin
-40dB with 0dB line
-60dB Magnitude
curve
Phase
curve
0º Phase Crossover
Phase margin with -180º line
-90º
-180º
-270º
Phase Log ω
(Degree)
33
GAIN CROSS-OVER, PHASE CROSS-OVER
Gain Cross-Over Frequency
The frequency at which the magnitude plot is having the
magnitude of zero dB is known as gain cross over frequency. It
is denoted by ωg or ωgc.
Phase Cross over Frequency

The frequency at which the phase plot is having the phase of -
180 is known as phase cross over frequency. It is denoted by
wp or ωpc.
34
Bode Plot & Stability
1. Stable
If wg<wp => GM & PM are positive
2. Unstable
If wg>wp => GM & PM are negative
3. Marginally stable
If wg=wp => GM & PM are zero
35
WHY is 0dB, -180 DEGREES SO BAD?
R E Y
G(s)
We apply a sine wave and because the system is -180 degrees out of
phase, from the output we get the same sine wave, but inverted!
Because of the substraction, we get an error that is twice the size of
the input! This is no good. This causes the system to create an actual
response from the system that is bigger, causing a bigger error, and
this will spiral out of control until the system aborts or breaks. This is
an unstable system.
36
EXAMPLE 5: A CONTROL SYSTEM HAS THE BLOCK DIAGRAM
R E Y
k G(s)
100
G(s) 
s ( s 2  10 s  400)
Sketch the Bode plot of the open loop system for k = 1 and use it
to find:
a) the smallest value of k that will make the system unstable
b) the value of k that will give a gain margin of 10 dB.
37
METHOD
Find frequency response function in Bode form:
n2 = 400 n = 20
G ( j ) 
100k Bode gain K 0  0.25k
j ( j   10 j  400)
2
100k 0.25k 1
  
    
j  400 j / 20  0.5 j / 20  1
2
 j    
j / 20  0.5 j / 20  1
2
20 log G ( j ) =0 for k = 1
 20 log 0.25  20 log k  20 log   20 log 1   / 20   

2 2
 0.5 / 20 
2
-12 dB
G ( j ) G ( j )
-20
Magnitude (dB)
-20 dB/dec
-40
-60
-60 dB/dec
-80
0 1 2
10 10 20 break frequency 10
= natural frequency
Frequency (rad/sec)
39
 0.5 / 20 
G ( j )  0  90  tan 
 1

2 
 1   / 20  
-90
-135
Phase (deg)
-180
-225
-270
0 1 2
10 10 20 break frequency 10
Frequency (rad/sec)
40
Bode plot for k = 1
Bode Diagram
0
Magnitude (dB)
GM = 32 dB
-20
-20 dB/dec
-40
-60
-60 dB/dec
-80
-90
Phase (deg)
-135
phase = –180
-180
-225
-270
0 1 2
10 10 10
Frequency (rad/sec)
41
Questions
a) the smallest value of k that will make the system
unstable?
b) the value of k that will give a gain margin of 10 dB?
Answers – from Bode plot for k = 1

a) From Bode plot: Gain Margin = 32 dB if k = 1
So, system will be unstable if
20 log k  32 k  1032 / 20  39.8
42
Magnitude should equal to -10 at phase crossover frequency. Then
20 log k  32  10  22 k  10 22 / 20  12.6
Bode Diagram
0
Magnitude (dB)
-20 20 log k
-40
GM = 0 dB, k = 39.8
-60 GM = 10 dB, k = 12.6
-80 GM = 32 dB, k = 1
-90
Phase (deg)
-135
phase = –180
-180
-225
-270
0 1 2
10 10 10
Frequency (rad/sec)
43
Assoc.Dr. Güleser Demir
Asst.Dr. Hatice Doğan NYQUIST PLOTS &
DEU, Electrical and Electronics Engineering
STABILITY CRITERION
1
Frequency Response

WHY NYQUIST PLOTS?
Bode' plots
• use frequency as the horizontal axis and
• use two separate plots
• to display amplitude and phase of the frequency response.
Nyquist plots
• display both amplitude and phase angle
• on a single plot,
• using frequency as a parameter in the plot.
Nyquist plots allow you to see whether a system is stable

or not.
3
WHAT IS A NYQUIST PLOT?
Nyquist plot is a polar plot the
plot of the frequency take frequency imaginary part
response
function. That means as a parameter of G(jw) against
you want to plot and the real part of
G(jw). G(jw).
4
Example
1
•Nyquist plot for system with transfer function G ( s ) 
s2
1
G ( j ) 
j  2
 j  2 1
 . G ( j )   0  
 j  2 j   2
 j  2 Im
 2
 4 0.25
2 = 0 rad/s
Re{G ( j )}  2 = 
 4 Re
0 0.25 0.5

Im{G ( j )}  2
 4
- 0.25
= 2 rad/s 5
5
Example: 25  s  1
Ls 

s  s  2  s 2  2 s  16 
Construction of Nyquist loci
 Loop transfer function
 By hand
 Calculate features such as
 Asymptotes: behavior as  0 and  
 Location of axes crossings
6
25  s  1
Ls 
 
System loop transfer func.
s  s  2  s 2  2 s  16
L  j  
25  j  1

 j   j  2  16   2

 2 j 
  
j  j  2    2 j  16  j   j  2  16   2  2 j
2



 j 25  j  1  j  2  16   2  2 j   
 j 25 2   2  j 16   2  2 j 
  4   2  4  28 2  256 
  
2 
 4    16   2 2 
 4 2 

25 12  3 2   j  800  400 2  25 4 
L  j  
  4   2  4  28 2  256 

25 12 3
Im L  j  
 25 4  400 2  800 
4 28 256  
 4   2  4  28 2  256 
7
Re L  j  

25 12  3 2 
 4   2  4  28 2  256
 25 4  400 2  800 
Im L  j  
  4   2  4  28 2  256 
Quantitative analysis
 Limits Im L 0
300
lim Re L  j    0.293
 0  4  256  Re L
0.3
800
lim Im L  j    
 0   4  256 
800
lim Im L  j    
 0   4  256    0
8
Re L  j  

25 12  3 2 
 4   2  4  28 2  256
 25 4  400 2  800  Im L
  0
Im L  j  
  4   2  4  28 2  256 
 Asymptotes for large 

 Keep dominant terms    Re L
   0.3
75 25
L  j    j
4 3
– For  positive: + imaginary axis
– For  negative:  imaginary axis
  0
9
Re L  j  

25 12  3 2 
 4   2  4  28 2  256
 25 4  400 2  800 
Im L  j  
  4   2  4  28 2  256 
Im L
  0
– Imaginary axis crossing(s)   2
• Real part = 0 0.8

   Re L
Re L  j   0  300  75 2 ⇒   2   
0.3
25x16 400x4 800

2
0.8
Im L jw 0.7815
2x8 16 28x4 256
  0
10
Re L  j  

25 12  3 2 
 4   2  4  28 2  256
 25 4  400 2  800 
Im L  j  
  4   2  4  28 2  256 
  0
Im L
– Real axis crossing(s)
• Imaginary part = 0
0  Im L  j   25 4  400 2  800
0.64   Re L
  4.2
  4  16 2  32 0.3
  2  8  64  32  4 2  6   2
0.8

   4 2  6  4.22 
Re L  j 4.22   0.638   0
11
The plot of the contour L for the range - < w < 0_ will be
 the complex conjugate of the plot for the range 0+ < w < +, and
 the polar plot of L(s) = Gc(s)G(s)H(s) will be symmetrical in the L(s)-plane about
the x-axis.
Therefore, it is sufficient to construct the contour L for the frequency range 0+ <
w < +  in order to investigate the stability.
Nyquist Diagram
1
0.5
Imaginary Axis
-0.5
-1
-1 -0.5 0 0.5 1
12
Real Axis
NYQUIST STABILITY CRITERION
Introduction:
To determine the relative stability of a closed-loop system, we must investigate the
characteristic equation of the system:
1 0
where
To ensure stability, we must ascertain that all the zeros of F(s) lie in the left-hand s-
plane.
13
MAPPING CONTOURS ıN THE S-PLANE
We are concerned with the mapping of contours in the s-plane by a function F(s).
A contour map is a contour or trajectory in one plane mapped or translated into
another plane by a relation F(s).
This contour
has been
mapped by
F(s) into a
contour of
an identical
form, a
square, with
the center
shifted by
one unit and
the
Assume magnitude
of a side
multiplied
by
two.
We will consider characteristic equation of the closed loop t.f as F(s) and express it as
The function F(s) is the characteristic equation, and so
where
!!! The zeros of F(s) are the characteristic roots of the system
(i.e. Closed-loop system poles)
Cauchy's theorem (/The principle of the argument)
encircles Z does not pass the traversal is in

If a contour s zeros and P through any the clockwise
in the s-plane poles of F(s) poles or zeros direction along the
and of F(s) and contour,
the
corresponding encircles the origin of the
F(s)-plane N= Z - P times in
contour F in the clockwise direction.
the F(s)-plane
16
NYQUIST STABILITY CRITERION
1
Im
s0
For a system to be stable, all the zeros of F(s) must lie in the left-
hand s-plane.
r  Re
Therefore,
 we choose a contour s in the s-plane that encloses the entire
right-hand s-plane, and
 we determine whether any zeros of F(s) lie within F by Nyquist Contour
utilizing Cauchy's theorem.( Plot F in the F(s)-plane and s
determine the number of encirclements of the origin N. Then
the number of zeros of F(s) within the s contour (and
therefore, the unstable zeros of F(s)) is
17
Alternatively, we may define the function
In this case, the number of clockwise encirclements of the origin of the F(s)-plane
becomes the number of clockwise encirclements of the -1 point in the F'(s) = L(s)-
plane because F'(s) = F(s) - 1.
18
NYQUIST STABILITY CRITERION :
N= number of clockwise encirclements of the (-1,0) point,
P = number of poles of L(s) in right-hand of s-plane.
Z = number of zeros of F(s) in right-hand of s-plane.
If the systems are stable, should be Z = 0.
Nyquist stability criterion :
• A feedback system is stable if and only if the

When P=0 : contour L in the L(s)-plane does not encircle
the (-1,0) point when P = 0.
• A feedback control system is stable if and

only if, for the contour L, the number of
When P 0: counterclockwise encirclements of the (- 1,0)
point is equal to P. (P:the number of poles of
L(s) with positive real parts) 19
Nyquist plot
A  B Plot G ( s 0 )  G ( j )
•To draw a Nyquist plot, the
for  = 0  
contour be can broken into three
pieces Im
G ( j )
Re
G ( s0 )  G ( j )
Im G ( 0)
B
B  C For most systems G(s0) stays fixed
A r  Re
C  A Plot G ( s0 )  G (  j ) for  =   0
Im
C Re
20
20
Example
( s  1) 2
•Nyquist plot for G(s) 
( s  2)( s  4)
( j  1) 2
Nyquist Diagram G ( j ) 
1 ( j  2)( j  4)
G (s )
0.5
Imaginary Axis
-0.5
-1
-1 -0.5 0 0.5 1
Real Axis 21
21
NYQUIST-CRITERION :EXAMPLE
• Note that the number of poles of L{s) in the right-hand s-plane is zero,and thus P = 0.
• Therefore, for this system to be stable, we require N = Z = 0, and the contour must not
encircle the - 1 point.
• Examining figure (b), we find that the contour does not encircle the -1 point, and the
system is stable.
SPECIAL CASE WHEN G(S)H(S) INVOLVES POLES AND/OR ZEROS ON THE JW
AXIS.
Since the Nyquist path must not pass through poles
or zeros of L(s) if the function L(s) has poles or zeros
at the origin (or on the jw axis at points other than
the origin), the contour in the s plane must be
modified.
The usual way of modifying the
contour near the origin is to use a semicircle with
the infinitesimal radius , as shown in figure.
23
Consider, a closed-loop system whose open-loop transfer
function is given by
This detour is a consequence of the condition of Cauchy‘s

theorem, which requires that the contour cannot pass
through the pole at the origin.
Part1 (From B to C):

s  e j
K
G( s) H ( s) se j  , 0    90
e j (e j  1)
K K
lim 0 G(e j ) H (e j )  lim 0 j  lim 0 arg( )
e 
Part2 (From C to D) :
K
s  jw G ( jw) 
jw(1  jwT )
K
G ( jw)  N=Z-P  Z=N+P=0+0=0
w(1  w 2T 2 )1 / 2
Then the system is stable for all gains K,K>0.
arg(G ( jw))  90  tan 1 ( wT )
Part3 (From D to E) : |L(s)|=0

24
The system is stable when
or
The system is stable when
or
This system is unstable irrespective of the gain, K

EXAMPLE Draw the Nyquist plot for the open loop system
k
kG ( s ) 
( s 2  s  1)( s  1)
Will the closed loop system be stable for k = 1?
Will the closed loop system be stable for k = 10?
Nyquist Diagram for k = 1
1
Answers:
The closed loop system will be
stable for k = 1 as the Nyquist
0.5
-1 plot does not circle the -1 point
Imaginary Axis
0.3
0 The system will be unstable for

k > 1/0.3
-0.5
k = 10  system unstable
-1
-1 -0.5 0 0.5 1
Real Axis 29
STABILITY MARGINS
- GAIN AND PHASE MARGINS
Can we see how close to unstable a closed loop system will be
from the Nyquist plot of the open loop system?
How much can we change the gain or phase of kG(j) without

making the closed loop system unstable?
How close is the Nyquist plot to the -1 point?
We can answer these important questions using ‘stability margins’
30
GAIN AND PHASE MARGINS
Gain margin = GM = 20 log (1/b)
Phase margin = PM = 
The gain margin: the increase

in the system gain when
-b phase = -180° that
will result in a marginally stable
system with intersection of the
- 1 + j0 point
on the Nyquist diagram.
Phase Margin: the phase angle

through which the L(jw) locus must
be rotated so that the unity
magnitude |L(jw)| = 1 point will
pass through the ( - 1, 0) point in
the L(jw) plane.
31
EXAMPLE – HOW TO FIND GM AND PM
Nyquist plot for open loop system:
Im
1
GM = 20 log (1/0.6)
= 20 log 1.4
= 4.4 dB
-0.6 Re
PM = °
-1 62 1
-1 kG(j)  = 0  
32
Example: The Nyquist plot of the open loop transfer function G(s) of a unity
feedback system is shown in Figure for gain K=1. The system has no poles of G(s) in the
right half-plane. The plot intersects the negative real axis at 0, -a, -b and -c points.
a) Assume ‐1  [‐b,‐a]. Determine whether

the system is stable and find the number of
roots (if any) in the right‐half s‐plane.
b) Assume ‐1  [‐c,‐b]. Determine whether
the system is stable and find the number of
roots (if any) in the right‐half s‐plane.
c) Find the values of K that make the system
stable in terms of a, b and c.
33
a) Assume ‐1  [‐b,‐a]. Determine whether the
system is stable and find the number of roots (if
any) in the right‐half s‐plane.
P=0, N=+1-1=0 so Z=N+P=0 system stable
b) Assume ‐1  [‐c,‐b]. Determine whether c) Find the values of K that make the
the system is stable and find the number of system stable in terms of a, b and c.
roots (if any) in the right‐half s‐plane. For stability 1/b<K<1/a and k<1/c
P=0, N=2, Z=N+P=2 system unstable
34
Question: Nyquist plots of four stable open loop transfer functions are shown. Find the number of
zeros of closed loop transfer function in right-hand of s-plane.
35
EED 3016 CONTROL SYSTEMS Assoc.Dr. Güleser K Demir
Assist. Dr. Hatice Doğan
PID CONTROLLERS 𝐺 𝑠 , PID controller Plant that we want to control its output
PID controller takes the error e(t), as the input

Transfer function of PID controller is
and computes its proportional, derivative and
U ( s)  K  integral terms. Summation of these terms is
Eq.1 Gc ( s)    K p  i  Kd s 
E (s)  s  outputted as the control signal u(t). This control
signal is inputted to the plant, and a new output
In the time domain; y(t) is obtained. This new output y(t) is measured
t de(t ) by the sensor to find the new error signal e(t).
Eq.2 u (t )  K p e(t )  K i  e(t )dt  K d The controllers takes this new to produce control
0 dt signal. This process goes on and on.
Proportional gain Integral gain Derivative gain
t de(t )
Eq.3 u (t )  K p e(t )  K i  e(t )dt  K d
0 dt
Proportional term Integral term Derivative term
𝐺 𝑠 , PID controller Plant that we want to control its output
Transfer function of PID controller is

U ( s)  K 
Gc ( s)    K p  i  Kd s 
E (s)  s 
If Proportional (P) controller.
If Proportional Integral (PI) controller.
If Proportional Derivative (PD) controller.

PROPORTIONAL (P) CONTROL ACTION
For a controller with proportional control action, the
relationship between the output of the controller
u(t) and the error signal e(t) is
Eq.4
When the control signal 𝑢 𝑡 is linearly proportional to the

system error 𝑒(𝑡), we call the result proportional controller.
𝑈(𝑠)
𝐺 𝑠 = =𝐾
𝐸(𝑠)
𝑒(𝑡) 𝑢 𝑡
𝐾
1
t t
(a) (b) (c)
PROPORTIONAL-PLUS-INTEGRAL (PI) CONTROL ACTION
The control action of a proportional plus-integral controller is defined by
t
Eq.5 u (t )  K p e(t )  K i  e(t )dt
0
Then, transfer function of a PI Controller is

( )
Eq.6
( )
Output of the PD controller

Input to the controller
(a) (b)
PROPORTIONAL-PLUS-DERIVATIVE (PD) CONTROL ACTION
The control action of a proportional plus derivative controller is defined by
de(t )
Eq.7 u (t )  K p e(t )  K d
dt
Then, transfer function of a PD Controller is
( )
Eq.8 ( )
Output of the PD controller
Input to the controller
(a) (b)
PROPORTIONAL-PLUS-INTEGRAL-PLUS-DERIVATIVE CONTROL
ACTION
The combination of proportional control action, integral control action, and derivative control action is termed
proportional-plus-integral-plus-derivative control action. This combined action has the advantages of each of
the three individual control actions. The equation of a controller with this combined action is given by
t de(t )
u (t )  K p e(t )  K i  e(t )dt  K d
0 dt
Then, transfer function of a PID Controller is

( )
( )
EXAMPLE: CRUISE CONTROL SYSTEM
( ) /
A model for the crusie-control system is given by . Assume that the mass of
( ) /
the car is 3000kg and the friction coefficient Design a PI controller such that the speed of
the car will reach the desired speed 90 km/h in 10 seconds (i.e., the settling time is 10s) and the
maximum overshoot is less than 25%.
1
𝐾
(𝐾 + 𝑠 )( 3000 1 )
𝑌(𝑠) 𝐺 𝑠 𝐺(𝑠) 𝑠 + 3000 (𝐾 𝑠 + 𝐾 )/3000
𝑇 𝑠 = = = =
𝑅(𝑠) 1 + 𝐺 𝑠 𝐺(𝑠) 1 1+𝐾 𝐾
𝐾 𝑠 + 𝑠 +
1 + (𝐾 + 𝑠 )( 3000 1 )
3000 3000
𝑠 + 3000
2𝜁𝑤 𝑤
𝑌(𝑠) 𝐺 𝑠 𝐺(𝑠) (𝐾 𝑠 + 𝐾 )/3000
𝑇 𝑠 = = =
𝑅(𝑠) 1 + 𝐺 𝑠 𝐺(𝑠) 1+𝐾 𝐾
𝑠 + 3000 𝑠 + 3000
2𝜁𝑤 𝑤

From requirement 1 overshoot less than 25 %  %𝑂𝑆 = 𝑒 x 100 < 0.25 ⟹ 𝜁 > 0.4
Any value 𝜁 > 0.4 is valid. But we have a zero at 𝑠 = . Zero at the left hand-side of the s-
plane can increase the overshoot. To be sure, we choose 𝜁 = 0.6
From requirement 2 settling time of 10s  𝑇 = ⇒𝑤 = = 0.8
Then,
= 2 0.8 0.6 ⇒ 𝐾 = 2879 𝐾 1920 2879𝑠 + 1920

𝐺 𝑠 = (𝐾 + ) = 2879 + =
𝑠 𝑠 𝑠
= 0.8 ⇒ 𝐾 = 1920
G s
𝐺 (𝑠)
1920
2879 +
𝑠
Check the max.

overshoot, if the
design
specification does
not meet than
change the
damping ratio.
EXAMPLE-2 CONTROL OF FIGHTER AIRCRAFT
It can be showed that the vertical position of a fighter aircraft can be approximated by the eq.:
where is the vertical position (in meters) above the sea level and is the trust force. Block diagram of
feedback control system is given below.
We want the aircraft to reach

10000m above the sea level.
We will analyze the output
a) Without any feedback
and controller (i.e.,
open-loop system)
b) Using a PI controller in
the feedback control
system
a) Without any feedback and controller (i.e., open-loop control system)
If we want to the aircraft to reach 10000m above the sea level without a controller, one might guess that we
need to apply an input force r Let us try to see whether this works or not.
. .
.
The vertical position of the aircraft reaches

200000 m instead of the desired 10000 m.
Since the output is 20 times greater than

the desired one, we may think that 20 times
reduction in input force, i.e. r , will
provide desired output. In the next slide, we
will examine this case.
500
𝑅 𝑠 = ℒ 500 =
𝑠
1 500 1 1
𝑌 𝑠 =𝐺 𝑠 𝑅 𝑠 = = 10000( − )
𝑠 + 0.5 𝑠 𝑠 𝑠 + 0.05
𝑦 𝑡 =ℒ 𝑌 𝑠 = 10000(1 − 𝑒 . )
The vertical position of the aircraft reaches

10000 m as desired when r for
open loop control system (with no controller).
However, we have an important limitation

in this set-up: The aircraft will reach the
desired level in about 120 seconds ! Very
slow !
In the following slides, we will design a PI

controller to get satisfactory response
characteristics such as minimum steady-state
error and acceptable settling time.
a) PI controller
We have seen earlier that it would take approximately 2 minutes for the fighter aircraft to reach
10000 m if it is without a feedback controller. Let us design a PI controller for it such that the aircraft
will reach the desired vertical level in 30 seconds and the maximum overshoot is less than 10%.
Solution:

From requirement 1 overshoot less than 10 % 
Any value is valid. In this example, we choose
From requirement 2  settling time of 30s 
𝐾 𝐾 𝑠+𝐾
𝐺 𝑠 =𝐾 + =
𝑠 𝑠
1
𝐺 𝑠 =
𝑠 + 0.5
𝐾 1
𝑌(𝑠) 𝐺 𝑠 𝐺(𝑠) (𝐾 + 𝑠 )( ) 𝐾 𝑠+𝐾
𝑇 𝑠 = = = 𝑠 + 0.05 = Eq.11
𝑅(𝑠) 1 + 𝐺 𝑠 𝐺(𝑠) 1 + (𝐾 + 𝐾 )( 1 ) 𝑠 + (0.05 + 𝐾 )𝑠 + 𝐾
𝑠 𝑠 + 0.05
Note that the formal transfer function of a standard 2nd order system is given by Eq.12
Comparing denominators of Eq.11 and Eq. 12, we get

0.005 + 𝐾 = 2𝜁𝑤 = 2 0.7 0.19 ⇒ 𝐾 =0.216 𝐾 0.036 0.216𝑠 + 0.036
𝐺 𝑠 = (𝐾 + ) = 0.216 + =
𝐾 = 𝑤 = 0.19 ⇒ 𝐾 =0.036 𝑠 𝑠 𝑠
0.216𝑠 + 0.036
𝑠
EXAMPLE 3:
Suppose we have a simple mass, spring, damper problem with the dynamic model of
Let
In this example, we’ll examine the output characteristics by using i) no feedback
controller, ii) P controller, iii) PD controller, iv) PI controller and v) PID controller.
ms 2 X ( s )  bsX ( s )  kX ( s )  F ( s )
X (s) 1 X ( s) 1
 2  2
F ( s ) ms  bs  k F ( s ) s  10 s  20
I) NO CONTROLLER
The (open) loop transfer function is given by:

X ( s) 1
 2
F ( s ) s  10 s  20
The applied force f(t) is unit step. The steady-state value Figure 3.1
for the output is: 1/20=0.05 is the final value of the

F (s) output to an unit step input.
xss  lim x(t )  lim sX ( s )  lim s 2
t  s 0 s 0 s  10 s  20
1 1 1 This corresponds to a steady-state

 lim s 2 
s 0 s s  10s  20 20 error of 95%, quite large!
The settling time is about 1.5 sec.

II) PROPORTIONAL (P) CONTROLLER
The closed loop transfer function is given by:

Kp
X ( s) s 2
 10 s  20 Kp Figure 3.2
  2 Let
F ( s) Kp s  10 s  ( 20  K p )
1 2
s  10 s  20
Comparing Figure 3.1 and Figure 3.2,
note that the proportional controller
reduces both the rise time and the
steady-state error, increases the
overshoot, and decreases the settling
time by small amount.
III) PD CONTROLLER
K p  Kd s
X (s) s 2
 10s  20 K p  Kd s Figure 3.3
  2
F (s) K p  Kd s s  (10  Kd )s  (20  K p ) Let and
1 2
s 10s  20
Comparing Figure 3.2 and Figure 3.3,
note that the proportional derivative
controller reduced both the overshoot
and the settling time, and had small
effect on the rise time and the steady-
state error.
IV) PI CONTROLLER
K p  Ki / s
X ( s) s 2
 10 s  20 K p s  Ki Figure 3.4
  3
F ( s) K p  K i / s s  10 s 2  (20  K p ) s  K i Let and
1 2
s  10 s  20
Comparing Figure 3.2, 3.3 and 3.4, note
that we have reduced proportional gain
because the integral controller also reduces
the rise time and increases the overshoot as the
proportional controller does (double effect).
The above response shows that the integral
controller eliminates the steady-state error.
V) PID CONTROLLER
K p  Kd s  Ki / s
X ( s) s 2  10s  20 Figure 3.5

F ( s) K p  Kd s  Ki / s
1 Let and
s 2  10s  20
K d s 2  K p s  Ki

s 3  (10  K d ) s 2  (20  K p ) s  K i Now, we have obtained the system with
no overshoot, fast rise time, and no
steady-state error.
EFFECT OF PROPORTIONAL,
INTEGRAL & DERIVATIVE GAINS ON THE
DYNAMIC RESPONSE
PROPORTIONAL CONTROLLER
Pure gain (or attenuation) since:
the controller input is error and the controller output is a proportional gain
Increase in gain:
Step responses
for different
gains in P
Controller
PROPORTIONAL CONTROLLER
Pure gain (or attenuation) since:
the controller input is error and the controller output is a proportional gain
Increase in gain:
Step responses  Upgrade both steady state
for different and transient responses
gains in P  Reduce steady-state error
Controller
 Decrease rise time
 Increase overshoot
INTEGRAL CONTROLLER
Integral of error with a constant gain
 increase the system type by 1
 eliminate steady-state error for a unit step input
 amplify overshoot and oscillations
Step Increase in integral gain:

responses
for different
integral
gains in PI
Controller
INTEGRAL CONTROLLER
Integral of error with a constant gain
 increase the system type by 1
 eliminate steady-state error for a unit step input
 amplify overshoot and oscillations
Step Increase in integral gain:

responses
for different  Do not upgrade steady-state
integral responses
gains in PI
 Increase slightly settling time
Controller
 Increase oscillations and
overshoot!
DERIVATIVE CONTROLLER
Differentiation of error with a constant gain
 detect rapid change in output
 reduce overshoot and oscillation
 do not affect the steady-state response
Step
responses for
Increase in derivative gain:
different
derivative
gains in PD
Controller
DERIVATIVE CONTROLLER
Differentiation of error with a constant gain
 detect rapid change in output
 reduce overshoot and oscillation
 do not affect the steady-state response
Step
responses for Increase in derivative gain:
different
derivative  Upgrade transient response
gains in PD
Controller
 Decrease the peak and rise time
 Decrease overshoot and settling time!

PID CONTROLLER
Step responses for
different gains in PID
Controller
CONCLUSIONS
Increasing the proportional feedback gain reduces steady-state errors,

but high gains almost always destabilize the system.
Integral control provides robust reduction in steady-state errors, but often

makes the system less stable.
Derivative control usually increases damping and improves stability, but has
almost no effect on the steady state error
ON THE APPLICATION OF PID CONTROL
PID regulators provide reasonable control of most industrial processes, provided that the
performance demands is not too high.
PI control are generally adequate when plant/process dynamics are essentially of 1st-
order.
PID control are generally ok if dominant plant dynamics are of 2nd-order.
More elaborate control strategies needed if process has long time delays, or lightly-
damped vibrational modes
ROOT-LOCUS BASED DESIGN Assoc.Dr. Güleser K Demir
CONTROL SYSTEMS DESIGN BY ROOT-LOCUS METHOD
Compensation is the modification of the system dynamics to satisfy the given specifications.
The requirements imposed on the control system are usually spelled out as performance
specifications. The specifications may be given in terms of transient response requirements
(such as the maximum overshoot and settling time in step response) and of steady‐state
requirements (such as steady‐state error in following ramp input).
Design by Root‐Locus Method is based on reshaping the root locus of the system by adding

poles and zeros to the system's open loop transfer function and forcing the root loci to pass
through desired closed‐loop poles in the s plane.
Commonly used compensators are lead, lag, and lag‐lead compensators and PID controllers.

CONFIGURATIONS
Two configuration of compensators will be discussed
 cascade compensators to improve the steady-state error and/or transient response
 feedback compensators to improve the transient response
Series/cascade Compensation Parallel (feedback) Compensation

INTRODUCTION
Root Locus is a specific technique which shows
how
changes in one of a system’s parameter
(usually the controller gain, K)
will modify the location of the closed-loop poles
.
Closed-loop transfer function

Loop transfer function
(or Open-loop Tr. Func.)
Uncompensated System
Desired pole location
1 0
Cascaded Compensated System
Suppose that our goal is to speed up the response at A to that of B,

without effecting the percent overshoot.
We augment, or compensate, the system with additional poles and zeros,
so that the compensated system has a root locus that goes through the
desired pole location for some value of gain.
PRELIMINARY DESIGN CONSIDERATIONS
We place a compensator in series with the transfer function G(s) to obtain desirable behavior.
The main problem then involves the choice of the pole(s) and zero(s) of the compensator
G(s) to alter the root loci so that the performance specifications will be met.
The addition of a pole to the open-loop

transfer function
Pulls the root locus to the right
Lowers the system's relative stability,

Slows down the settling of the response.
The addition of a zero to the open-loop transfer

function
Pulls the root locus to the left
Makes the system more stable and

Speeds up the settling of the response.
LEAD COMPENSATION
Electronic circuit that is
a lead network if ,
and a
lag network if
1 Pole-zero
0 α 1
1 Im
Configuration of a
lead network.
1
General form of
1 Lead Compensator 1 1
DESıGN PROCEDURE FOR LEAD COMPENSATION
1. From the performance specifications, determine the desired location for the dominant closed‐loop poles.
2. By drawing the root‐locus plot of the uncompensated system (original system), ascertain whether or not the gain adjustment
alone can yield the desired closed loop poles. If not, calculate the angle deficiency . This angle must be contributed by the lead
compensator if the new root locus is to pass through the desired locations for the dominant closed‐loop poles.
3. Assume the lead compensator to be
1
1
, 0 α 1
1 1
Where  and T are determined from the angle deficiency. Kc is determined from the requirement of the open‐loop gain.
4. If static error constants are not specified, determine the location of the pole and zero of the lead compensator so that the lead
compensator will contribute the necessary angle .
5. Determine the open‐loop gain of the compensated system from the magnitude condition.
EXAMPLE It is desired to modify the closed‐loop poles so that
rad/sec and the value of the damping ratio,
ζ=0.5.
Then, the desired locations of the closed‐loop poles
For uncompensated closed‐loop system, we have of this example are
10/ 1 10
10 10
1 1 1.5 2.59
1 ,
10
0.5 3.12 0.5 3.12
Remember that the general form of 2nd order

transfer function is
2
Then
2 1 0.1581
10 3.16
Without any compensation, the closed‐loop

poles are at , 1 0.5 3.12
The desired locations of the closed‐loop poles are , 1 1.5 2.5981 . They
should be the roots of characteristic equation of the compensated system, therefore
1 0 ⇒ 1
. . . .
. . 1
⇒
. . 180° 1 , 0, 1, 2, 3, …
A general procedure :
i) find the sum of the angles at the desired closed‐loop pole
with the open‐loop poles and zeros of the
original(/uncompensated) system,
ii) determine the necessary angle  to be added so that the . . 180 (2k+1)

total sum of the angles is equal to 180° 2 1 .
. . . . 180
The lead compensator must contribute this angle  . MANY 220.89

POSSIBILITIES!!
40.9
One Possible Solution
(The pole and zero location of the lead 40,9
compensator contribute the angle =40.9)
Hence, the lead compensator becomes

1
3
The value of can be determined by use of the magnitude conditon:
1 10
. . 1
3 1 . .
Then
3
0.9
10 . .
1
and our compensator is 0.9
3
The closed‐loop transfer function of the compensated system becomes

LAG COMPENSATION
Electronic circuit of a lag network if
1
1
1 1
Pole-zero
General form of Lag Compensator Im
Configuration of a
lag network.
1 1
LAG COMPENSATION
For the case where the system has transient‐response characteristics but unsatisfactory steady‐state characteristics.
Increases the open loop gain without appreciably changing the transient‐response characteristics
This means that the root locus in the neighborhood of the dominant closed‐loop poles should not be

changed appreciably, but the open‐loop gain should be increased as much as needed.
The angle contribution of the lag network should be limited to a small amount, say 5˚.To assure this,

we place the pole and zero of the lag network relatively close together and near the origin of the s
plane.
If gain Kc of the lag compensator is set
equal to 1, then the transient response
characteristics will not be altered.
An increase in the gain means an increase in the static error constants.
The main negative effect of the

For the uncompensated system the static velocity error constant is lag compensation is that the
compensator zero that will be
generated near the origin creates
a closed-loop pole near the origin.
This closed loop pole and
For the compensated system the static velocity error constant becomes compensator zero will generate a
long tail of small amplitude in the
step response, thus increasing the
settling time.
Velocity error constant is increased. Therefore, steady-state error will decrease !

Design Procedures for Lag Compensation by the Root‐Locus Method:
1. Draw the root‐locus plot for the uncompensated system. Based on the transient‐response
specifications, locate the dominant closed‐loop poles on the root locus.
2. Assume the transfer function of the lag compensator to be
1
1
1 1
3. Evaluate the particular static error constant specified in the problem.
4. Determine the amount of increase in the static error constant necessary to satisfy the specifications.
5. Determine the pole and zero of the lag compensator that produce the necessary
increase in the particular static error constant without appreciably altering the original root loci.
6. Draw a new root‐locus plot for the compensated system. Locate the desired dominant
closed‐loop poles on the root locus.
7. Adjust gain of the compensator from the magnitude condition so that the dominant
closed‐loop poles lie at the desired location. ( will be approximately 1.)
EXAMPLE
•The dominant closed‐loop poles are
, 1 0.33 0.58
•The damping ratio, is 0.491.
•The undamped natural frequency of the dominant

closed‐loop poles, is 0.673 rad/sec.
•The static velocity error constant is 0.53 sec‐1.
It is desired to increase the static velocity error constant Kv to

about 5 sec-1 without appreciably changing the location of the
dominant closed-loop poles.
To increase the static velocity error constant
by a factor of about 10, choose  as 10 and place the zero
and pole of the lag compensator at s = ‐0.05 and s = ‐0.005,
respectively.
If the damping ratio of the new dominant
closed‐loop poles is kept the same, then the
poles are obtained from the new root‐locus
plot as follows:
LAG-LEAD COMPENSATION
If improvements in both transient response and steady‐state
Lead compensation basically speeds up the response and
response are desired,
increases the stability of the system.
then both a lead compensator and a lag compensator may be
used simultaneously.
Lag compensation improves the steady‐state accuracy of
the system, but reduces the speed of the response.
Rather than introducing both a lead compensator and a lag
compensator as separate elements, however, it is economical
to use a single lag‐lead compensator.
1 1
1 1
1 1
1
1 1
...
In designing lag‐lead compensators, we consider two cases where
is approximately unity, where s = s1 is one of the dominant
1. From the given performance specifications, determine the closed‐loop poles, choose the values of T1 and  from the
desired location for the dominant closed‐loop poles. requirement that
2. Using the uncompensated open‐loop transfer function
G(s), determine the angle deficiency  if the dominant
closed‐loop poles are to be at the desired location. The
phase‐lead portion of the lag‐lead compensator must
contribute this angle .
3. Assuming that we later choose T2 sufficiently large so that
the magnitude of the lag portion The choice of T1 and  is not unique. (Infinitely many sets of
T1 and  are possible.)
...
Then determine the value of Kc from the magnitude where Kc and  are already determined in step 3. Hence, the
condition: value of  can be determined from this last equation.
Then, using the value of  thus determined,
choose the value of T2 such that
4. If the static velocity error constant Kv is specified,
determine the value of  to
satisfy the requirement for Kv. The static velocity error
constant Kv is given by
4. For the lag‐lead compensator, we later choose T2
sufficiently large so that
1. From the given performance specifications, determine the
desired location for the dominant closed‐loop poles.
2. The lag‐lead compensator is modified to
is approximately unity, where s = s1 is one of the dominant
closed‐loop poles. Determine the values of TI and  from the
magnitude and angle conditions:
Where  > 1. If the static velocity error constant Kc is specified,

determine Kc from: 5. Choose T2 so that
3. To have the dominant closed‐loop poles at the desired
location, calculate the angle
Contribution  needed from the phase lead portion of the lag‐
lead compensator.
EXAMPLE • Let us assume , in that case the transfer function of
the compensator:
•This system has closed-loop poles at Damping frequency
• From the performance specifications, the dominant
, 0.2500 j1.9843  1  closed‐loop poles must be at
The damping ratio is 0.125, the undamped natural

frequency( ) is 2 rad/sec, and the static velocity error constant • Since
is 8.
•It is desired to make the damping ratio of the dominant closed-

loop poles equal to 0.5 and to increase the undamped natural the phase lead portion of the lag‐lead compensator must
frequency to 5 rad/sec and the static velocity error constant to contribute
80 . 180=(‐235)+ = 55
•Design an appropriate compensator to meet all the

• Determine the location of the zero and pole that will
performance specifications.
give 55˚ contribution
There are many possible choices!!! The phase lag portion:
First the value of P is determined to satisfy the requirement on
We shall choose the zero at s = ‐0.5 the static velocity error constant:
By simple calculation, the pole must
be at s = ‐5.021.
Choose the value of T2 large enough so that
Next we determine the value of Kc from the magnitude
condition:
T2 = 5 (or any number greater than 5) satisfies the above two
requirements.
Choose T2 = 5
EXAMPLE
Assuming the specifications are the same as in previous
example, design a compensator.
The desired locations:
The phase lead portion:
For the phase lag portion, we may choose
Since the closed‐loop pole at s = ‐0.1003 is very close to a zero
The closed‐loop poles are : at s = ‐0.1, they almost cancel each other. Thus, the effect of
this closed‐loop pole is very small.
The remaining closed‐loop pole (s = ‐3.8604) does not quite
cancel the zero at s = ‐2.4. The effect of this zero is to cause a
larger overshoot in the step response than a similar system
without such a zero
the dominant closed‐loop poles
Very close to the desired poles!
(accept as equal)
PARALLEL COMPENSATION
Series Compensation Parallel or Feedback Compensation
(Char. Eq.) (Char. Eq.)
Since Gf is a fixed transfer function, the design of Gc becomes the same as the

case of series compensation.
EXAMPLE
Consider the system shown in above. Draw a root-locus diagram. Then determine the value of k such that the damping ratio
of the dominant closed-loop poles is 0.4.
=K
...
Note that the closed-loop poles with  = 0.4
must lie on straight lines passing through the
origin and making the angles ±66.42 with the
negative real axis.
The intersection of the asymptotes with the real axis:
The angle of departure from the pole at s = j2
... we have two solutions for this problem related to two different K values
the three closed‐loop poles are located at the three closed‐loop poles are located at
Dominant poles
...
Notice that the response of the system with k =
0.4490 is oscillatory. (The effect of the closed‐loop
pole at s = ‐2.9021 on the unit‐step response is
small.)
For the system with k = 1.4130, the oscillations due
to the closed‐loop poles at s = ‐2.1589 ± j4.9652
damp out much faster than purely exponential
response due to the closed‐loop pole at s = ‐0.6823.
k = 0.4490  a faster response with relatively small overshoot

k = 1.4130  a slow overdamped response
We should choose k = 0.4490 for the present system.
FREQUENCY RESPONSE BASED DESIGN Assoc.Dr. Güleser K Demir
FREQUENCY RESPONSE
Output: a sinusoidal response with the same
Input: a general sinusoid
frequency of input.
𝑟 𝑡 = 𝐴 cos 𝑤𝑡 + 𝐵 sin 𝑤𝑡 = 𝑀 cos( 𝑤𝑡 + 𝜙 )
𝑐 𝑡 = 𝑀 𝑀 cos( 𝑤𝑡 + 𝜙 + 𝜙 )
𝑀 = 𝐺(𝑗𝑤)
𝜙 = 𝑎𝑛𝑔𝑙𝑒 𝑜𝑓 𝐺(𝑗𝑤)
Magnitude Response: The ratio of the output

sinusoid’s magnitude to the input sinuoid’s magnitude.
Frequency Response: The combination
of the magnitude and phase frequency
Phase Response: The difference in phase angle
between the output and the input sinuoids.
RELATION BETWEEN CLOSED-LOOP TRANSIENT AND CLOSED-LOOP FREQUENCY RESPONSES
𝑤 𝑤
𝑀 = 𝑇(𝑠) = =
𝑠 + 2𝜁𝑤 𝑠 + 𝑤 (𝑤 − 𝑤 ) +4𝜁 𝑤 𝑤
1
𝑀 =
2𝜁 1 − 𝜁
Log magnitude plot of M

𝑤 =𝑤 1 − 2𝜁
Maximum magnitude on the frequency response curve is related to the damping ratio, 𝜁 and
hence, the percent overshoot.
Bandwidth of the closed-loop frequency response: The frequency , at which the magnitude response
curve is 3 dB down from its value at zero frequency.
The bandwidth of a two-pole system can be found by finding that frequency for which (that is, -3
dB) and if we substitute we can relate to settling time.
Normalized bandwidth versus damping ratio

APPROACHES Frequency-Response Method
Root-Locus Method
Frequency response is the steady-state response
Precise root locations are known and actual time of a system to a sinusoidal input.
response is easily obtained by means of the
inverse Laplace Transform.
In the frequency-response approach, we specify
the transient response performance in an indirect
The root-locus method was shown to be very manner.
useful to reshape the transient response
characteristics of closed-loop control systems.
That is, the transient-response performance is
specified in terms of the
Performance specifications Phase and gain margins,
Peak time; Resonant peak magnitude (it gives a rough estimate of
Overshoot; the system damping),
Settling time for a step input; Gain crossover frequency,
Steady-state error. Resonant frequency,
Bandwidth (it gives a rough estimate of the speed of
These performance specifications can be transient response), and
defined in terms of the desirable location of
the poles and zeros of the closed-loop. Static error constants (it gives the steady-state
accuracy).
BODE-DIAGRAM APPROACH
Although the correlation between the transient response and frequency response is indirect, the frequency
domain specifications can be conveniently met in the Bode diagram approach.
The compensator can simply be added to the original Bode diagram, and thus plotting the complete Bode
diagram is a simple matter.
If the open-loop gain is varied, the magnitude curve is shifted up or down without changing the slope of the
curve, and the phase curve remains the same.
BODE DIAGRAM APPROACH
— First adjust the open-loop gain so that the requirement on the steady-state accuracy is met.
— Then plot the magnitude and phase curves of the uncompensated, but gain-adjusted, open-
loop system.
— Reshape the open-loop transfer function with the addition of a suitable compensator to meet
gain margin and phase margin specifications.
— Try to meet other specifications if any.
The open-loop frequency response indicators:

— Low-frequency region (<< wgc) : indicates the steady-state behavior of the closed-loop system
— Medium-frequency region: indicates relative stability (wgc is the gain crossover frequency)
— High-frequency region (>> wgc ): indicates the complexity of the system
Compensation is a compromise between steady-state accuracy and relative stability. The open-loop
frequency-response curve needs to be reshaped.
Design via Gain Adjustment
Desired phase margin represented by CD is Φ . Find Design Procedure
the required increase in gain.
1. Draw the Bode plots.
2. Find the frequency on the

Bode phase diagram that yields
the desired phase margin CD
3. Change the gain by an amount

AB to force the magnitude curve
to go through 0 dB at The
amount of gain adjustment is the
additional gain needed to produce
the required phase margin.
EXAMPLE:
For the position control system shown in figure below, find the preamplifier gain K so that
the phase margin 59.2.
A Design Procedure
Draw the Bode plots.
.
B Find the frequency on the Bode
phase diagram that yields the desired
phase margin CD
Locate on the phase plot the frequency that
yields a phase margin. This
frequency is found where the phase angle
is the difference between and
, or Then,

Change the gain by an amount AB to

force the magnitude curve to go through
0 dB at
14.8 rad/s.
At , the gain is -44.2dB.
𝑤 = 14.8 𝑟𝑎𝑑/𝑠 This magnitude has to be raised to 0db.
Then,
LEAD COMPENSATION
In designing lead compensators via Bode plots, we want to change phase diagram,
increasing the gain crossover to realize a faster transient response.
Assume the following lead compensator:
= (0< <1)
Characteristics of Lead Compensators High-Pass Filter
1
𝑇𝑠 + 1 𝑠+ 𝑗𝑤𝑇 + 1
𝐾𝛼 =𝐾 𝑇 (0 < 𝛼 < 1) 𝐾𝛼 (0 < 𝛼 < 1)
𝛼𝑇𝑠 + 1 1 𝑗𝑤𝛼𝑇 + 1
𝑠+
𝛼𝑇
Polar plot of a lead comp. (Kc=1) Bode diagram of a lead comp. (Kc=1 =0.1)
(Eq. 1)
VISUALIZING LEAD COMPENSATION
The uncompansated system has a small phase margin (B-(-180))
and low phase margin frequency.
Lead compensator, increases bandwidth by increasing the gain

cross over frequency (C). Higher bandwidth leads a reduction in
settling time
After compensation, the phase diagram is raised at higher

frequencies. The result is larger phase margin (D-(-180)) and
higher gain crossover frequency,wgc.
After compensation, the system is more susceptible to high-

frequency noise because of increase in high-frequency gain due to
the increase in bandwidth
Design Procedure
1. Assume the following lead compensator:
The open-loop transfer function of the compensated system is
Determine gain K to satisfy the requirement on the given static error constant.
2. Using the gain K thus determined, draw a Bode diagram of G1(jw), the gain adjusted but
uncompensated system. Evaluate the phase margin.
3. Determine the necessary phase-lead angle to be added to the system. Add an additional 5" to 12" to
the phase-lead angle required, because the addition of the lead compensator shifts the gain crossover
frequency to the right and decreases the phase margin.
4. Determine the attenuation factor  by use of Eq (1). Determine the frequency where the magnitude of
the uncompensated system G1(jw) is equal to  . .Since
Select this frequency as the new gain crossover frequency. This frequency corresponds to 
and the maximum phase shift occurs at this frequency.
5. Determine the corner frequencies of the lead compensator as follows:

Zero of lead compensator:
Pole of lead compensator:
6. Calculate constant Kc from
𝐾
𝐾 =

7. Check the gain margin to be sure it is satisfactory. If not, repeat the design process by modifying the
pole-zero location of the compensator until a satisfactory result is obtained.
Example-1
4
The open-loop transfer function is 𝐺 𝑠 =
𝑠(𝑠 + 2)
It is desired to design a compensator for the system so that the static velocity error constant Kv is
20 sec-1, the phase margin is at least 50°, and the gain margin is at least 10 dB.
Step 1:
4𝐾
𝐺 𝑠 = 𝐾𝐺 𝑠 =
𝑠(𝑠 + 2)
𝑇𝑠 + 1 𝑠4𝐾
𝐾 = lim 𝑠𝐺 𝑠 𝐺(𝑠) = lim 𝑠 𝐺 𝑠 = lim = 2𝐾 = 20
→ → 𝛼𝑇𝑠 + 1 → 𝑠(𝑠 + 2)
𝐾 = 10
Step 2:
The phase and gain margins of the uncompensated system

are 17 and  dB, respectively.
(A phase margin of 17 implies that the system is quite

oscillatory.
Step 3: The specification calls for a phase margin of at least

50. We thus find the additional phase lead necessary to
satisfy the relative stability requirement is
𝜙 =𝜙 −𝜙 = 50 − 17 =33.
Considering the shift of the gain crossover frequency, we

may assume that the max phase lead required is
approximately 38. (we added additional 5)
Step 4:
𝜙 = 38
Step 5:
Step 6:
The compensated system has the
following open-loop transfer
function:
EXAMPLE
For the scontrol system shown in the figure, design a lead compensator instead of
preamplifier so that velocity error constant Kv=40 1/sec and phase margin 48.1.
In order to meet the specification K must be set at 1440. We have to plot

146
   

 .
-3.76 dB corresponds to 39 rad/s. 
 
 60.2
=3428.6

146
LAG COMPENSATION
In designing lead compensators via Bode plots, we want to decrease the steady-state
error without appreciably affecting the transient response.
Assume the following lag compensator:

1
𝑇𝑠 + 1 𝑠+𝑇
𝐺 𝑠 =𝐾𝛽 =𝐾 (𝛽 > 1)
𝛽𝑇𝑠 + 1 1
𝑠+
𝛽𝑇
Characteristics of Lag Compensators.
1
𝑇𝑠 + 1 𝑠+
𝐺 𝑠 =𝐾𝛽 =𝐾 𝑇 (𝛽 > 1)
𝛽𝑇𝑠 + 1 1
𝑠+ Low Pass Filter
𝛽𝑇
Polar Plot of a lag compensator Bode diagram of a lag compensator

Kc=1 and 𝛽=10 Kc=1 and 𝛽=10
Visualizing Lag Compensation
The uncompansated system is unstable.
The lag compensator
• reduces the system gain at higher frequencies

without reducing the system gain at low
frequencies,
• reduces the system bandwidth and so the system

has a slower response speed,
• Improves steady-state accuracy since the total

system gain and hence, low-frequency gain, can
be increased because of the reduced high-
frequency gain ,
• is less susceptible to high-frequency noise since

the high-frequency gain is reduced
Design Procedure
1. Assume the following lag compensator:
1
𝑇𝑠 + 1 𝑠+
𝐺 𝑠 =𝐾𝛽 =𝐾 𝑇 (𝛽 > 1)
𝛽𝑇𝑠 + 1 1
𝑠+
𝛽𝑇
𝐾𝛽=𝐾
𝑇𝑠 + 1 𝑇𝑠 + 1
𝐺 𝑠 𝐺(𝑠) = 𝐾𝐺 𝑠 = 𝐺 (𝑠) (𝛽 > 1)
𝛽𝑇𝑠 + 1 𝛽𝑇𝑠 + 1
Determine gain K to satisfy the requirement on the given static velocity error constant.
2. If G1(jw) = KG(jw) does not satisfy the specifications on the phase and gain margins, then find the frequency point
where the phase angle of the open-loop transfer function is equal to -180 plus the required phase margin. The required
phase margin is the specified phase margin plus 5 to 12. Choose this frequency as the new gain crossover frequency.
3. Choose the corner frequency w = 1/T (corresponding to the zero of the lag compensator) 1 octave to 1 decade
below the new gain crossover frequency.
Notice that we choose the compensator pole and zero sufficiently small. Thus the phase lag occurs at
the low-frequency region so that it will not affect the phase margin.
4. Determine the attenuation necessary to bring the magnitude curve down to 0 dB at the new gain
crossover frequency. Noting that this attenuation is -20 log  determine the value of β. Then the other
corner frequency (corresponding to the pole of the lag compensator) is determined from w = 1/(βT).
5. Using the value of K determined in step 1 and that of β determined in step 4, calculate constant K,
from
𝐾
𝐾 =
𝛽
Example
1
𝑠(𝑠 + 1)(0.5𝑠 + 1)
It is desired to compensate the system so that the static velocity error

constant Kv=5 , PM=40 and GM=10 dB.
The lag compensator will be used.
1
𝑇𝑠 + 1 𝑠 + 𝑇
𝐺 𝑠 =𝐾𝛽 =𝐾 (𝛽 > 1)
𝛽𝑇𝑠 + 1 1
𝑠+
𝛽𝑇
Define :
Define also
K
( )( . )
The first step in the design is to adjust the gain K to meet the required static
velocity error constant:
→ → → ( )( . )
With K=5, Plot the Bode diagram of :
( )( . )
Find the required phase:
θ = −180 + 40 + 12 = −128
(we added additional 12)
Find at which frequency the phase equals -128

and take this frequency as the new gain
crossover frequency:
𝑤 = 0.5 𝑟𝑎𝑑/𝑠
At 𝑤 the gain of 𝐺 is 20 dB we have to bring

the magnitude curve down to 0 dB so
1
−20𝑙𝑜𝑔β = 20𝑙𝑜𝑔 = −20
𝛽
−𝟏𝟐𝟖 𝛽 = 10
Find the pole and zero of the compensator:
𝑤
𝑧= = 0.1
5
𝑧
𝑝= = 0.01
𝛽
EXAMPLE
For a unity feedback system, where forward path TF is given as:
Design a lag compensator to yield Kv=162.2 error while phase margin=59.2.
Kv=162.2 therefore K=5839,2.
We have to plot Bode diagrams for K=5839,2.

110.8
-180+59.2+10=-110.8 at 9.8 rad/s. Chose wgc=9.8rad/s
At 9.8 rad/s the magnitude 24 dB,

lag compensator must provide -24 dB
So -24dB=-20log   =15.8
The zero of the compensator : 9.8/10=0.98 rad/s. 𝐾 5839,2

𝐾 = = = 369,57
0.98 𝛽 15,8
The pole of the compensator: 9.8/15.8=0.062rad/s
𝑠 + 0.98
𝐺 (𝑠) = 369,57
𝑠 + 0.062
LEAD-LAG COMPENSATİON
Generally = is chosen in design, although this is not necessary.

The term
produces the effect of the lead network and the term
produces the effect of lag network.

LEAD-LAG COMPENSATİON
Based on the combination of the design techniques for the lead and lag compensators.
Assume = :
The phase lead portion : The phase lag portion:
alters the frequency-response curve by provides attenuation near and

adding phase lead angle and above wgc and thereby allows an
increasing the phase margin at the gain increase of gain at the low-
crossover frequency, wgc. frequency range to improve the
steady-state performance.
EXAMPLE
Consider the unity-feedback system whose open-loop TF is:
( )( )
It is desired that the static velcoity error constant be 10 1/sec, the PM=50 and the GM be
10dB or more.
The open loop TF of the compansated system: Gc(s)G(s). Assume Kc=1. Then
→
From the requirement on the static velocity error constant:
→ → ( )( )
Hence K=20.
1. Draw Bode diagram for K=20
The gain-adjusted but uncompansated system
PM=-32dB
2. Choose the new wgc.
From the phase-angle curve for G(jw),
G(jw)=-180 at w=1.5 rad/sec so that the
phase lead angle required at w=1.5 rad/sec is
about 50.
3. Choose the corner frequency w=1/T2
(corresponds to the zero of the phase-lag
portion of the compensator) to be 1 decade
below the new wgc or at w=0.15 rad/sec.
Notice that  = 10 corresponds to m = 54.9°. Since we need a 50° PM, we may choose  = 10 . (Note that
we will be using several degrees less than the maximum angle. 54.9°) Thus,  = 10.
Then the corner frequency w = 1/T2 (which corresponds to the pole of the phase-lag portion of the
compensator) becomes w = 0.015 rad/sec. The transfer function of the phase-lag portion of the lag—lead
compensator then becomes
The phase-lead portion can be determined as follows:
Since the new gain crossover frequency is w= 1.5 rad/sec, from figure G(j1.5) is found to
be 13 dB. Hence, if the lag-lead compensator contributes -13 dB at w = 1.5 rad/sec, then
the new wgc frequency is as desired.
From this requirement, it is possible to draw a straight line of slope 20 dB/decade,

passing through the point (1.5 rad/sec, -13 dB). The intersections of this line and the 0-
dB line and -20 dB line determine the corner frequencies. Thus, the corner frequencies
for the lead portion are w=0.7 rad/sec and w=7 rad/sec.
Thus the TF of the lead portion of the lag-lead compensator becomes

Since we choose Kc=1,
. . . .
. . .
the open loop TF of the compensated system is
. .
. ( )( )
( . )( . )
( . )( . )( )( . )
PM=50 and GM=16 dB and the static error velocity constant is 10 for the
compensated system. All the requirement are met and the design has been
completed.
LEAD - LAG COMPENSATORS:
— can result in both fast response and good static accuracy

— can result in an increase in low-frequency gain (which improves steady-
state accuracy) while at the same time the system bandwidth and stability
margins can be increased.
Assoc.Dr. Güleser K Demir
CANONICAL FORMS Assist.Dr. Hatice Doğan
Consider the system defined by a differential equation
where y is the output and u is the input.
Transfer function
State space representation
There are essentially an infinite number of choices of A, B and C to produce the same transfer function (or

to give the identical input/output dynamics)
There are certain standardized state space model structures: these are the so‐called canonical forms. If a

transfer function of a system is given then each canonical models can be obtained.
1. PHASE-VARIABLE CANONICAL FORM Transfer function
x1  y x1  x2
x2  y  x2  x3
let x3  y 
 xn  y ( n )   a1 y ( n 1)  a2 y ( n  2 )    an y  u
  an x1  an 1 x2    a2 xn 1  a1 xn  u
xn  y ( n 1)
y  x1
 x1   0 1 0 0   x1   0 
 x   0 0 1 0   x 2   0 
 2    u
        
      
 x n    a n  a n 1   a1   x n   1 
 x1 
x 
y  1 0  0  2 

 
 xn 
x1  y x1  x2
x2  x3
x2  y 

x3  y
xn  y ( n )   a1 y ( n 1)  a2 y ( n  2 )    an y  u

  an x1  an 1 x2    a2 xn 1  a1 xn  u
( n 1)
xn  y y  x1
y ( n1)  xn
u (t ) y (n ) y  x2 y  x1
…
∫ ∫ ∫ ∫
＋
＋
y ( n2)  xn1
 a1
 a2

 an
Block diagram of phase variable canonical form
Example: Consider the system given below. Obtain phase-variable canonical state space model of this system
and draw an all-integrator block diagram.
d3y d 2 y dy
3
 5 2   2y  u
dt dt dt
x1  y
let x2  x1  y
x3  x2  y
y ' '  x3 y  x2
y  x1
 x1   0 1 0   x1  0 u (t ) y' ' '
 x    0   x   0  u ∫ ∫ ∫
0 1 ＋
 2   2    ＋
 x3   2  1  5  x3  1
5
 x1 
x 
y  1 0  0 2  1

 
 xn  2
EXAMPLE:
2. CONTROLLABLE CANONICAL FORM
Y (s) b1s n 1  b2 s n  2    bn
 G (s)  n
U (s) s  a1s n 1  a2 s n  2    an
Transfer function can be decomposed into cascade form:
U(s) V(s) Y(s)
Re-arranging equation yields:
⋯ ⋯
⋯ ⋯
Y (s) b1s n 1  b2 s n  2    bn ⋯
 G (s)  n
U (s) s  a1s n 1  a2 s n  2    an ⋯
x1c  v x1c  x2c

x2 c  v  x2c  x3c
let x3c  v 
 xnc  an x1c  an1 x2c   a1 xnc  u
xnc  v ( n 1) y  bn x1c  bn1 x2c   b1 xnc
 x1c   0 1 0 0   x1c   0 
 x   0 0 1 0   x  0 
 2c      2c    u
          
      
 x nc    a n  a n 1   a1   x nc   1 
 x1c 
x 
y  bn b n 1  b1  2 c 
  
 
 x nc 
Block diagram of contollable canonical form
b1
b2

bn 1
xnc x( n 1) c
u (t ) y (n )
x1c y
… x2 c
∫ ∫ ∫ ∫ bn +
＋
＋
 a1
 a2

 an
Example: Consider the transfer function given below. Obtain controllable canonical form.
4 s 3  25s 2  45s  34
G (s)  3
2 s  12 s 2  20 s  16
s 2  5s  2 0.5s 2  2.5s  1
G ( s)  2  3  2 3
2 s  12 s  20 s  16
2
s  6 s 2  10 s  8
 x1   0 1 0   x1  0
 x    0 0 1   x   0u (t )
 2   2   
 x3   8  10  6  x3  1
 x1 
y  1 2.5 0.5 x2   2u (t )
 x3 
3. OBSERVABLE CANONICAL FORM
Y (s) b1s n 1  b2 s n  2    bn
 G (s)  n n 1 n2
b 0
U (s) s  a1s  a2 s    an
Note the relationship between the

observable and controllable forms:
Aobs  Acont
T
Bobs  Ccont
T
Cobs  Bcont
T
3. OBSERVABLE CANONICAL FORM
Y (s) b1s n 1  b2 s n  2    bn
 G (s)  n n 1 n2
b 0
U (s) s  a1s  a2 s    an
Note the relationship between the

observable and controllable forms:
Aobs  Acont
T
Bobs  Ccont
T
Cobs  Bcont
T
Block diagram of observable canonical form
u(t )
bn bn1 … b1 b0
y(t )
(n)
y x1o x2o xno
+ ∫ + ∫ … + ∫ +
 a1  a2 …  an
Example: Obtain observable canonical form of the given transfer function.
4 s 3  25s 2  45s  34 s 2  5s  2 0 .5 s 2  2 . 5 s  1
G (s)  3  2 3  2 3
2 s  12 s  20 s  16
2
2 s  12 s  20 s  16
2
s  6 s 2  10 s  8
 x1  0 0  8   x1   1 
 x   1 0  10  x   2.5u (t )
 2   2   
 x3  0 1  6   x3  0.5
 x1 
y  0 0 1 x2   2u (t )
 x3 
4. DIAGONAL CANONICAL FORM
Y (s) b0 s n  b1s n 1  b2 s n  2    bn c c c
 G (s)   b0  1  2    n
U (s) ( s  p1 )( s  p2 )  ( s  pn ) s  p1 s  p2 s  pn
 x1    p1 0 0 0   x1  1
 x   0  p2 0 0   x 2  1
 2    u
   0       
      
 x n   0 0   p n   x n  1
 x1 
x 
y  c1 c2  c n  2   b0 u

 
 xn 
5. JORDAN CANONICAL FORM
Y (s) b1s 5  b2 s 4    bn
 G (s) 
U (s) ( s  p1 ) 3 ( s  p2 ) 2 ( s  p3 )
c11 c12 c13 c21 c22 c3
     
( s  p1 ) 3 ( s  p1 ) 2 ( s  p1 ) ( s  p2 ) 2 ( s  p2 ) s  p3
 x1    p1 1 0   x1  0  x1 
 x   0 x  0 x 
 p1 1 0  2
 2   2   
 x 3   0 0  p1   x3  1   x3 
 x 4    p2   x4  0 x 
y  c11 c12 c3  4 
1
  0     u c13 c21 c22
 x 5   0  p2   x5  1   x5 
 x 6   0  p3   x6  1   x6 
        
        
         
Example: Consider the system given below. Obtain Jordan canonical form.
1 0 1 0.5 0.5
G ( s)     
( s  1) ( s  1) ( s  1) s s  2
3 2
 x1    1 1 0   x1   0 
 x   0 1 1 0   x  0 
 2   2   
 x 3   0 0 1   x3  1 
 x    0   x   1  u
 4   4   
 x 5   0  2   x5  1 
      
      
      
 x1 
x 
 2
 x3 
y   1 0  1 0 . 5 0 . 5  x 4 
 
 x5 
 
 
 
Example: Consider the system given by
Obtain state-space representations in the controllable canonical form, observable

canonical form and diagonal canonical form.
Controllable Canonical Form:
0 1 0
2 3 1
3 1
Observable Canonical Form:
0 2 3
1 3 1
0 1
Diagonal Canonical Form:
1 0 1
0 2 1
2 1
DESIGN VIA STATE SPACE Assoc.Dr. Güleser K Demir
Mathematical Models
12.1 INTRODUCTıON
Frequency Domain Approach State-Space Representation
 Converts system’s differential equation into a
transfer function, thus gives a model that  Provides a unified method for modeling,
algebraically relates output to input. analyzing, and designing a wide range of
systems using matrix algebra.
 Applicable only to Linear, Time-Invariant (LTI)
systems or their close approximations !  Useful to represent nonlinear systems that have
backlash, saturation, and dead zone.
Design by using the  Useful for Nonlinear, Time-Varying, Multiple-

desired location of the input, multiple-output systems
dominant second-order
Design by specifying all
pair of poles.
closed loop poles of the
We hope that higher higher order system.
order poles do not
affect the second-order
approximation!
12.2 CONTROLLER DESıGN
R(s) C(s)
Nth-order feedback
control system
…
⋯
⋯ 0
 Goal: Setting all of the poles of the closed loop system to any desired location.
 Approach: Introduce n adjustable parameter and relate them to the coefficients

of characteristic equation, Q(s).
Plant that we want to control is modelled by using
the state-space representation:
Input vector State vector
⋮ ⋮
Output vector
⋮
State-space representation of a plant
New Control Topology:

Plant with state feedback Instead of feeding back y,
feed back all of the state
variables.
New Controller Design:

Determine the values of
the feedback gains, ,
CONTROLLABıLıTY
If the control signal, , can control the behaviour of each state variable in then, we can place the
poles of the system where we desire. If not, a state feedback design is not possible.
Definiton: If an input to a system can be found that takes every state variable from a desired inital state
to a desired final state, the system is said to be controllable; otherwise the system is uncontrollable.
Uncontrollable system:
behaves in its own
way regardless of the
Controllable system:
control signal, .
A . CONTROLLABıLıTY BY ıNSPECTıON: ( If the system matrix, A, is diagonal)
System 1 System 2
 x1   a1 0 0  x1  1  x1   a4 0 0  x1  0

x    0  a 0 x   1u(t ) x    0  a 0 x   1u(t )
 2  2  2     2  5  2   
 x3   0 0  a3  x3  1  x3   0 0  a6  x3  1
x1  a1x1  u x1  a4 x1

x2  a2 x2  u x2  a5 x2  u
x3  a3 x3  u x3  a6 x3  u
Each state variable is independent and decoupled The state variable is not controlled by the control .
from the rest, the control affects each of the state Thus, the system is said to be uncontrollable.
variable. This system is controllable.
A system with distinct eigenvalues and a diagonal system matrix is controllable

if the input coupling matrix B does not have any rows that are zero
B . CONTROLLABıLıTY BY USING CONTROLLABILITY MATRIX :
An n‐th order plant whose state equation is
is completely controllable if the matrix
… .
is of rank, , where is called the controllability matrix.
Recall: The rank of a matrix equals the number of linearly indepedent rows or columns.
The rank can be found by finding the highest-order square submatrix that is nonsingular.
EXAMPLE: Determine the controllability of the systems given below
System 1 System 2
 x1  1 1 0  x1  0  x1  1 1 0  x1  0

x    0 1 0 x   1u(t ) x    0 1 0 x   0u(t )
 2   2     2   2   
 x3   0 0  2 x3  1  x3   0 0  2 x3  1
0 1 2 0 0 0
1 1 1 0 0 0
1 2 4 1 2 4
1 0 ⇒ 3, is nonsingular. 0 ⇒ , is singular.
The system is controllable since the rank of equals The system is uncontrollable since the rank of is 1,
the system order. which is less than the order,3, of the system.
EXAMPLE:
U (s) 1 s-1 x1
 1 0  1
x    x    u 1
 0  3  0 Y (s)
1
y  1 2x -1 x2 2
s
3
※ State x2 is uncontrollable.
OBSERVABILITY
Definiton: If the initial state vector, , can be found from and measured
over a finite interval of time from , the system is said to be observable; otherwise the
system is said to be unobservable
Unobservable system
Observable system ( is not connected to
(Each state variable the output and could
can be observed at not be estimated from a
the output since each is measurement of the
connected to the output)
output.)
A . OBSERVABILTY BY ıNSPECTıON: (for a diagonalized system)
For systems represented in diagonalized form, if any column of the output matrix C is zero, the system is not
observable
Observable system
1 1 1 Unobservable system
0 1 1
B . OBSERVABILITY by USING OBSERVABILITY MATRIX:

An n‐th order plant whose state and output equations are, respectively,
is completely observable if the matrix
is of rank, , where is called the observability matrix.

EXAMPLE: Determine if the given systems are observable
System 1 System 2
0 1 0 0 0 1 0
0 0 1 0 5 21/4 1
4 3 2 1
5 4
0 5 1
0 5 1 5 4
4 3 3 20 16
12 13 9
344 0 ⇒ 3, 0⇒ 3,
The system is observable. The system is unobservable.

EXAMPLE:
0 1   0
x    x  1u
 0  2    4
y  0 4x
U (s) 1 s-1 x2 s-1 x1 Y (s)
0 4
0 8 2
※ State x1 is unobservable.
0⇒ ,
The system is unobservable.

EXAMPLE: Consider the system given by
1 1 1
0 1 0
Check the system controllability.
1 1
0 0
0 ⇒ , is singular.
The system is uncontrollable since the rank of is 1, which is less than the
order,2, of the system.
EXAMPLE: Consider the system given by
1 1 0
2 1 1
1 0
Is this system controllable and observable?

0 1
1 1
1 ⇒ , is nonsingular. The system is controllable.
1 0
1 1
1⇒ 2,
The system is observable.

Assoc.Dr. Güleser K Demir
POLE PLACEMENT Assist.Dr. Hatice Doğan
13.2 CONTROLLER DESIGN
R(s) C(s)
Nth-order feedback
control system
(… )
𝑇 𝑠 =
𝑠 +𝑎 𝑠 + ⋯+ 𝑎 𝑠+𝑎
Characteristic equation.
𝑄 𝑠 =𝑠 +𝑎 𝑠 + ⋯+ 𝑎 𝑠+𝑎 =0
The roots Q(s) are the poles of the system.
 Goal: Setting all of the poles of the closed loop system to any desired location.
 Approach: Introduce n adjustable parameter and relate them to the coefficients of characteristic equation, Q(s).
(Since the coefficient of the highest power of s is unity in Q(s), there are n coefficients whose values determine the system’s closed-loop
pole locations. Thus, if we can introduce n adjustable parameters into the system and relate them to the coefficients in Q(s), all of the poles
of the closed-loop system can be set to any desired location.)
State-space representation of a plant
New Control Topology:

Plant with state feedback Instead of feeding back y,
feed back all of the state
variables.
New Controller Design:

Determine the values of
the feedback gains,
ℎ
13.3 POLE PLACEMENT FOR PLANTS IN
PHASE-VARIABLE/CONTROLLABLE CANONICAL FORM
To apply pole-placement methodology to plants represented in phase-
variable/contrallable canonical form, we take the following steps:
1. Check the system controllability. If the system is not controllable you can not design
a controller.
2. Represent the plant in phase-variable / controllable-canonical form.
3. Feed back each state variable to the input of the plant through a gain, .
4. Find the characteristic equation for the closed-loop system represented in Step 2.
5. Decide upon all closed-loop pole locations and determine an equivalent
characteristic equation.
6. Equate like coefficients of the characteristic equations from Steps 4 and 5 and
solve for .
Following these steps, the controllable canonical representation of the plant is given by
 x1   0 1 0 0   x1  0
 x   0 0 1 0   x2  0
 2       u
        
      
 xn    an  a n 1   a1   x n  1 
 x1 
x 
y  bn bn  1  b1  2 

 
 xn 
The characteristic equation of the plant is thus
𝑄 𝑠 =𝑠 +𝑎 𝑠 + ⋯+ 𝑎 𝑠+𝑎 =0
Now form the closed-loop system by feeding back each state variable to u, forming
𝑢 = −𝑲𝒙
where
𝑲= 𝑘 𝑘 … 𝑘 𝑘 ’s are the feedback gains.
The system matrix, A-BK, for the closed-loop system is
the characteristic equation of the closed-loop system can be written by inspection as
det(𝑠𝐈 − 𝐀 − 𝐁𝐊 ) = 𝑠 + 𝑎 + 𝑘 𝑠 + 𝑎 +𝑘 𝑠 + ⋯ + (𝑎 +𝑘 )𝑠 + (𝑎 +𝑘 ) = 0 (*)
Now assume that the desired characteristic equation for proper pole placement is
𝑠 +𝑑 𝑠 + ⋯+ 𝑑 𝑠+𝑑 =0 (**)
where the ’s are the desired coefficients. Equating Eqs. (*) and (**), we obtain
𝑑 =𝑎 +𝑘 𝑖 = 1,2, … , 𝑛
From which
𝑘 =𝑑 −𝑎
Example: Consider the plant
By using the state feedback control, it is desired to have closed loop poles at
, and . Determine the state feedback gain
vector.
Solution: Lets find the controllable canonical representation of the system
,
Check the controlability matrix
𝟐
𝑴
Since 𝐌 the system is controllable.

Next feed back all state variables to the control, u, through gains , write the closed
loop system’s state equations:
This equation must match the desired characteristic equation
(*) The roots are 𝑠 , = and
(**)
Equating the coefficients of Eqs (*) and (**), we obtain
we obtain the following state-space representation of the closed-loop

system:
13.4 POLE PLACEMENT DESİGN FOR GENERAL PLANT
Controller Design by Matching Coefficients: If the system given in the phase-
variable/controllable-canonical form or the system order is low (n≤3), the design of state-
variable feedback for closed-loop pole placement consists of equating the characteristic
equation of a closed-loop system, to a desired characteristic equation and then
finding the values of the feedback gains, .
 Controller Design by Transformation: If the system order is high or the system is not
represented in phase-variable or controller-canonical form, the solution for the ’s can be
complex. Thus, it is advisable to transform the system to phase-variable/controllable-canonical
form, design the ’s, and then transform the system back to its original representation.
Controller Design by Matching Coefficients
Example:For the following system design state feedback control to yield 25% overshoot with a settling
time of 0.5 sec.
Example: For the following system design state feedback control to yield 1.5% overshoot with a
settling time of 0.83 sec. It is also desired that one of the three poles of the system is located at −4.8.
0 0 1
𝑪𝑴 = 𝑩 𝑨𝑩 𝟐 = 0 1 −5
𝑨 𝑩
1 −5 22
𝑪𝑴 = −1 ≠ 0 ⇒ 𝑟𝑎𝑛𝑘 𝑪𝑴 = 3, 𝑪𝑴 is nonsingular.
The system is controllable since the rank of 𝑪𝑴 equals the system order.

%𝑂𝑆 = 𝑒 x 100 = 1.5% 𝜁 = 0.8
𝑇 = = 0.83𝑤 = 6
𝐷𝑒𝑠𝑖𝑟𝑒𝑑 𝐶ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐 𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛: 𝑄 𝑠 = (𝑠 + 2𝜁𝑤 𝑠 + 𝑤 ) 𝑠 + 4.8 = 𝑠 + 14.4𝑠 + 82.1𝑠 + 172.8

Example: For the following system design state feedback control to yield 15% overshoot
with a settling time of 0.5 sec.
Solution:
Using the transient response requirements

stated in the problem (settling time and max.
where the characteristic equation is : overshoot), obtain the desired characteristic
equation
𝑠 + 𝑘 + 3 𝑠 + (2𝑘 + 𝑘 + 2) = 0
𝑠 + 16𝑠 + 239.5 = 0
𝑘 +3=16 𝑘 =211.5
2𝑘 + 𝑘 + 2 = 239.5 𝑘 =13

Eed 3016 Control Systems: Assoc - Dr. Güleser K Demir Assist. Dr. Hatice Do Ğan

Uploaded by

Copyright:

Available Formats

Eed 3016 Control Systems: Assoc - Dr. Güleser K Demir Assist. Dr. Hatice Do Ğan

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Eed 3016 Control Systems: Assoc - Dr. Güleser K Demir Assist. Dr. Hatice Do Ğan

Uploaded by

Copyright:

Available Formats

EED 3016 CONTROL SYSTEMS Assoc.Dr.

A control system is an interconnection of components

Input; Desired response : Output; Response :

A control system provides

Flyball governor used on a steam engine in a cotton factory near Manchester.

Invented! by J. Watt 1788 to reject the load disturbances also

● 1868 J. Clerk Maxwell formulates a mathematical model for governor control

● 1927 Harold Black discovers and patents the feedback amplifier

● 1927 Hendrik Bode analyzes feedback amplifiers

● 1932 Nyquist develops methods for analyzing feedback amplifier stability

becomes an engineering discipline

• 1950s Increased use of Laplace transform,

• 1980s Routine use of digital

• 1990s Feedback control in automobiles, automation,

An example of Open-loop system: Toaster

1. Producing the desired transient response

Transient response is also important

Homogeneous solution. Particular solution.

For a control system to be useful à the natural response must

If the natural response grows without bound à the system

´Factors affecting hardware selection (power

´Robustness to system parameter changes

The purpose of this system:

The Laplace transform is defined as The inverse Laplace transform is

Roots of the denominator of F(s) can be

𝑦 𝑡 = 1 − 2𝑒 '*+ + 𝑒 ',+ 𝑢(𝑡) 𝐾) =

𝑓 𝑡 = 𝐾!𝑒 '(" + + 𝐾" 𝑒 '3+ cos 𝛽𝑡 + 𝐵!𝑒 '3+ sin 𝛽𝑡 𝑢(𝑡)

𝑎$ 𝑠 $ 𝐶 𝑠 + 𝑎$'!𝑠 $'!𝐶 𝑠 + ⋯ + 𝑎0𝐶 𝑠 + 𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛 𝑡𝑒𝑟𝑚𝑠 𝑖𝑛𝑣𝑜𝑙𝑣𝑖𝑛𝑔 𝑐 𝑡

𝐶(𝑠) 𝑏# 𝑠 # + 𝑏#'!𝑠 #'! + ⋯ + 𝑏0

𝑉0 𝑠 𝑠 " + 45.95𝑠 + 22.55

This circuit can be used

Analogies between electrical

(𝑅@ +𝐿@ 𝑠)𝑇# (𝑠)

The output response of a system to the Multiplication of an input by a scalar

Linear models are useful because

LINEAR NONLINEAR NONLINEAR NONLINEAR

• Linearize it for small-

Taylor series expansion: Ignore higher-order terms

𝑑𝛿𝑥 Linearized Diff. Eq.

System matrix Input matrix

State Vector: A vector whose elements are the state variables.

The state vector must be chosen according to the following considerations:

The minimum number of state variables is equal to:

3) Substitute the equations in step 2 into the

𝐶(𝑠) 24 ' + 9𝑠 $ + 26𝑠 + 24 C s = 24R(s)

𝑐⃛ + 9𝑐̈ + 26𝑐̇ + 24𝑐 = 24𝑟

1) Seperate the transfer

Ø Take the Laplace transform assuming zero initial conditions

Ø For Single-Input, Single-output (SISO) systems, the transfer function is:

Note that it cannot 𝑓" (𝑥" , 𝑥# , … 𝑥$ , 𝑢" , 𝑢# , … 𝑢% )

𝜕𝑓C 𝜕𝑓C 𝜕𝑓C 𝜕𝑓C

𝐿𝑒𝑡 𝑥! = 𝑦 𝑎𝑛𝑑 𝑥" = 𝑦̇

To find the equilibrium points, we must solve

For the state space model

For the equilibruim point 𝑥(# = 0 and 𝑥($ = 0

The linearized system is

𝛿 𝑥̇ # 0 1 𝛿𝑥# 0 𝛿 𝑥̇ # (𝑡)=𝑥$ (𝑡)

𝑥̇ L(𝑡)=-sin 𝑥C 𝑡 −𝑥L 𝑡 + 𝑢(𝑡)

Nominal (Equilibrium) point;