EE334 Supplementary Notes

Download as pdf or txt
Download as pdf or txt
You are on page 1of 195

 

US  Naval  Academy  
EE334:   E lectrical   E ngineering   I I   a nd   I T   S ystems  
 
 

Supplementary   N otes  
 

Spring  2012–2013  
 

 
 

 
 

 
 
 
 

 
 

 
 
 

 
 

 
 
 

Last  Revision:  January  2013  


CAPT  Kevin  W.  Rudd,  Ph.D.  

2
Table  of  Contents  

Chapter 1: Counters and State Machine Design ........................................................................................... 6


1.1 Introduction......................................................................................................................................... 6
1.2 Sequential Logic Modules Review ..................................................................................................... 6
1.2.1 Clocked SR Flip-Flop .................................................................................................................. 6
1.2.2 The D or Delay Flip-Flop ............................................................................................................ 8
1.2.3 JK and T Flip-Flops ..................................................................................................................... 9
1.2.4 Asynchronous Inputs ................................................................................................................. 11
1.2.5 Timing Diagrams for Interdependent Flip-Flops ...................................................................... 13
1.3 Counters ............................................................................................................................................ 14
1.3.1 Basic Concepts of Digital Counters .......................................................................................... 14
1.3.2 Ripple Counters ......................................................................................................................... 15
1.3.3 Synchronous Counters ............................................................................................................... 18
1.4 Counter Design ................................................................................................................................. 20
1.4.1 State Tables ............................................................................................................................... 21
1.4.2 Flip-Flop Excitation Tables ....................................................................................................... 22
1.4.3 State Table Implementation....................................................................................................... 24
1.4.4 Additional State Machine Design Examples ............................................................................. 27
1.6 Homework Problems ........................................................................................................................ 30
Chapter 2: Digital and Analog Conversion ................................................................................................. 33
2.1 ADC and DAC Concepts .................................................................................................................. 33
2.2 Digital to Analog Conversion ........................................................................................................... 34
2.3 Analog to Digital Conversion ........................................................................................................... 36
2.3.1 The Comparator ......................................................................................................................... 37
2.3.2 Flash ADC ................................................................................................................................. 37
2.4 Homework Problems ........................................................................................................................ 41
Chapter 3: Introduction to Communications ............................................................................................... 42
3.1 Introduction....................................................................................................................................... 42
3.2 Communication Systems .................................................................................................................. 42
Chapter 4: Amplitude Modulation ............................................................................................................. 45
4.1 Introduction....................................................................................................................................... 45
4.2 Amplitude Modulation (AM) ........................................................................................................... 46
4.3 AM Bandwidth ................................................................................................................................. 53
4.4 AM Power ......................................................................................................................................... 54
4.5 Frequency Division Multiplexing (FDM) ........................................................................................ 56
4.6 Homework Problems ........................................................................................................................ 58
Chapter 5: AM Demodulation .................................................................................................................... 63
5.1 Introduction....................................................................................................................................... 63
5.2 Synchronous Demodulation.............................................................................................................. 63
5.3 Envelope Detection ........................................................................................................................... 65
5.4 Homework Problems ........................................................................................................................ 69
Chapter 6: AM Receivers ........................................................................................................................... 71
6.1 Introduction....................................................................................................................................... 71
6.2 TRF Receiver .................................................................................................................................... 71
6.3 The Superheterodyne Receiver ......................................................................................................... 72
6.4 Homework Problems ........................................................................................................................ 79
Chapter 7: Frequency Modulation ............................................................................................................. 81

3
7.1 Introduction....................................................................................................................................... 81
7.2 Description of the Modulation Process............................................................................................. 81
7.3 FM Spectrum .................................................................................................................................... 83
7.4 Advantages and Disadvantages of FM ............................................................................................. 85
7.5 FM Receiver ..................................................................................................................................... 85
7.6 Homework Problems ........................................................................................................................ 91
Chapter 8: Noise in Communication .......................................................................................................... 93
8.1 Introduction....................................................................................................................................... 93
8.2 Expressing Noise – SNR and Noise Ratio/Figure ............................................................................ 94
8.3 Sources of Noise - External Noise .................................................................................................... 96
8.4 Internal Noise.................................................................................................................................... 97
8.5 Overcoming Noise: Filtering .......................................................................................................... 100
Chapter 9: Digital Communications ........................................................................................................ 105
9.1 Introduction..................................................................................................................................... 105
9.2 Pulse Code Modulation................................................................................................................... 105
9.2.1 Sampling .................................................................................................................................. 106
9.2.2 Pulse Amplitude Modulation................................................................................................... 107
9.2.3 Other Analog Pulse Modulation Schemes............................................................................... 108
9.2.4 Quantization ............................................................................................................................ 109
9.2.5 Digital Encoding...................................................................................................................... 110
9.3 Digital Receivers ............................................................................................................................ 113
9.4 Error Detection and Correction ...................................................................................................... 114
9.5 Channel Capacity ............................................................................................................................ 118
9.6 Time Division Multiplexing ........................................................................................................... 123
9.7 Homework Problems ...................................................................................................................... 126
Chapter 10: Networking Overview ........................................................................................................... 129
10.1 Introduction................................................................................................................................... 129
10.2 Basic Networking Components .................................................................................................... 129
10.3 Networking Entities ...................................................................................................................... 129
10.4 Hardware and Software ................................................................................................................ 130
10.5 The OSI Model ............................................................................................................................. 131
10.6 Protocol Stacks ............................................................................................................................. 132
10.6.1 Communication Between Stacks ........................................................................................... 132
10.6.2 Encapsulation ........................................................................................................................ 132
10.7 The Physical Layer ....................................................................................................................... 134
10.7.1 The Data Link Layer ............................................................................................................. 134
10.7.2 The Network Layer ............................................................................................................... 135
10.7.3 Transport Layer ..................................................................................................................... 136
10.7.4 Session Layer ........................................................................................................................ 136
10.7.5 The Presentation Layer .......................................................................................................... 136
10.7.6 The Application Layer........................................................................................................... 136
10.8 Physical Connection of a Network ............................................................................................... 137
Chapter 11: Network Hardware ................................................................................................................ 139
11.1 The Physical Layer ....................................................................................................................... 139
11.2 The Data Link Layer ..................................................................................................................... 146
11.2.1 The Data Link Layer ............................................................................................................. 148
11.2.2 Functions of the Data Link Layer.......................................................................................... 149
11.2.3 Data Link Layer Hardware .................................................................................................... 150
11.2.4 Layer 1 Hardware Revisited .................................................................................................. 151
11.2.5 Layer 2 Hardware .................................................................................................................. 151
11.3 The Network Layer ....................................................................................................................... 153

4
11.3.1 Routed versus Routing Protocol ............................................................................................ 154
11.3.2 TCP/IP ................................................................................................................................... 154
11.3.3 UDP ....................................................................................................................................... 155
Chapter 12: Internet and Addressing ........................................................................................................ 157
12.1 Introduction................................................................................................................................... 157
12.2 IP Addresses ................................................................................................................................. 157
12.2.1 Classes of IP Addresses ......................................................................................................... 158
12.2.2 Reserved Host ID Numbers ................................................................................................... 161
12.3 Network Mask .............................................................................................................................. 162
Chapter 13: Subnetting.............................................................................................................................. 165
13.1 Introduction................................................................................................................................... 165
13.2 Subnet Mask ................................................................................................................................. 165
13.3 Subnetting Example ...................................................................................................................... 168
13.4 Plan for growth ............................................................................................................................. 169
Appendix A: Frequency Spectra and Ideal Filtering ................................................................................ 171
A.1 Amplitude Spectrum ...................................................................................................................... 171
A.2 Ideal Filtering ................................................................................................................................. 172
Appendix B: A Typical CW Communication System .............................................................................. 177
B.1 Introduction .................................................................................................................................... 177
B.2 A Citizen Band (CB) Transceiver .................................................................................................. 177
Appendix C: The Channel......................................................................................................................... 181
C.1 Introduction .................................................................................................................................... 181
C.2 Propagation of Signals in Free Space ............................................................................................ 182
C.3 Radio Waves .................................................................................................................................. 183
C.4 Propagation of Radio Waves.......................................................................................................... 184
C.4.1 Line of Sight (LOS) ................................................................................................................ 185
C.4.2 Surface Wave .......................................................................................................................... 185
C.4.3 Skywave .................................................................................................................................. 186
C.4.4 Forward Scatter ....................................................................................................................... 187
C.4.5 Summary ................................................................................................................................. 188
C.5 Multiple Path Propagation and Skip .............................................................................................. 189
Appendix D: Overview of the USNA SATCOM Communication System .............................................. 193
 

 
 

5
Chapter  1:   Counters  and  State  Machine  Design  
1.1    Introduction  
Up  until  this  point  you  have  been  studying  combinational  logic.    The  circuits  you  assemble  
out  of  AND,  OR  and  NOT  gates  have  definite  limitations.    Most  notably,  such  circuits  have  no  
memory.    Their  output  depends  only  on  their  present  input.    However,  if  you  think  about  it,  most  
complex  computing  functions  require  some  level  of  memory.    This  requires  moving  beyond  
combinational  logic  to  sequential  logic.    Sequential  logic  circuits  have  memory.    This  chapter  begins  
by  reviewing  flip-­‐flops,  which  are  the  building  blocks  for  sequential  logic.  

Using  sequential  logic,  we  will  then  describe  how  to  build  a  state  machine,  which  is  a  system  
that  consists  of  a  finite  number  of  states,  the  transitions  between  those  states,  and  actions  occurring  
as  a  result  of  being  in  or  transitioning  to  a  particular  state.    A  traffic  light  controller  is  an  example  of  
a  simple  state  machine.    Your  computer  is  a  state  machine,  too.    In  fact,  most  systems  can  be  
described  in  terms  of  state  machines.    Counters  are  an  important  sub-­‐category  of  state  machines  in  
which  the  states  progress  in  a  repeating  pattern,  so  we’ll  start  there  and  then  move  on  to  more  
complex  state  machines.    

1.2    Sequential  Logic  Modules  Review  


Flip-­‐flops  are  the  fundamental  building  blocks  from  which  counters  and  state  machines  are  
made.    Before  entering  into  a  discussion  of  counters  and  state  machines,  therefore,  we  need  to  
review  the  behavior  of  these  basic  circuit  blocks.    We  will  review  four  basic  types  of  flip-­‐flops:    the  
Clocked  SR  Flip-­‐Flop,  the  D  Flip-­‐Flop,  the  JK  Flip-­‐Flop,  and  the  T  Flip-­‐Flop.    

1.2.1    Clocked  SR  Flip-­‐Flop  


The  symbol  and  truth  table  for  the  clocked  SR  Flip-­‐Flop  are  shown  below  in  Figure  1-­‐1.  The  inputs  
for  this  flip-­‐flop  are  R  (“reset”),  S  (“set”),  and  C  (“clock”).    The  outputs  are  Q,  the  state  of  the  flip-­‐
flop,  and  “not  Q”,  written  as   Q ,  which  should  always  be  the  opposite  of  Q.    In  the  truth  table,  Qn  
represents  the  present  state  of  the  flip-­‐flop,  and  Qn-­‐1  represents  the  previous  state  of  the  flip-­‐flop.    
The  Xs  in  the  input  columns  of  the  truth  table  indicate  when  an  input  could  be  either  a  1  or  a  0,  
either  of  which  would  have  the  same  effect  (they’re  a  short-­‐hand  way  of  condensing  two  lines  of  the  
truth  table  into  one).      
The  truth  table  tells  us  that  the  flip-­‐flop  will  maintain  the  same  state  as  long  as  the  clock  is  
“off”  (i.e.,  logic  0)  or  both  R  and  S  are  0.    If,  however,  R  is  1  when  the  clock  is  “on”  (logic  1),  the  flip-­‐
flop  will  “reset”  to  0  (i.e.,  the  Q  output  goes  to  logic  0).    On  the  other  hand,  if  S  is  1  when  the  clock  is  
on,  the  flip-­‐flop  will  “set”  to  1  (Q  →  1).    Trying  to  both  reset  and  set  at  the  same  time  is  simply  not  
allowed  (in  reality,  this  would  result  in  Q  being  the  same  as   Q  which  violates  the  definition  of   Q  
and  would  likely  cause  errors  in  the  sequential  logic  circuit).        

6
R S C Qn
R Q 0 0 x Qn-1
C 0 1 1 1
S Q 1 0 1 0
1 1 1 Not allowed

x x 0 Qn-1
 

Figure 1-1: Symbol and truth table for clocked SR flip-flop


 

The  behavior  of  this  flip-­‐flop  can  be  demonstrated  through  the  use  of  a  timing  diagram.    
These  are  plots  which  show  the  behavior  of  the  variables  in  a  sequential  logic  circuit  over  time.    For  
this  example,  arbitrary  inputs  were  assumed  and  the  resulting  output  shown.    Note  that  both  the  
clock  and  the  set  or  reset  input  must  be  logic  1  before  the  output  responds.    

Before  moving  on  to  other  flip-­‐flops,  we  should  pause  to  discuss  the  concept  of  a  clock.    The  
clock  input  is  generally  a  square  wave  like  the  one  shown  in  Figure  1-­‐2.    The  purpose  of  the  clock  is  
to  keep  a  “steady  beat”  throughout  the  system,  keeping  everything  in  step  like  a  drum  for  a  parade.    
But  with  a  clock  response  like  the  SR  flip-­‐flop,  there’s  still  some  wiggle  room  because  of  the  width  of  
the  “on”  part  of  the  clock  (the  time  that    the  clock  is  logic  1).    This  can  cause  different  parts  of  a  large  
sequential  logic  circuit  to  fall  slightly  out  of  step.    We  need  a  crisper  drum  beat,  and  we  get  that  with  
“edge-­‐triggered”  flip-­‐flops.    The  D,  JK,  and  T  flip-­‐flops  are  all  examples  of  edge-­‐triggered  flip-­‐flops.  

These input signals are ignored


because they occur when the
clock is low

C
Inputs

R
Set doesn’t
take effect until
S clock is high

Output Q
Initial state for Q must be given or assumed Time  

Figure 1-2: Timing diagram example for a clocked SR flip-flop

7
1.2.2    The  D  or  Delay  Flip-­‐Flop  
An  edge-­‐triggered  flip-­‐flop  only  responds  to  the  other  inputs  at  points  when  the  clock  is  
transitioning  between  states.    The  transition  time  is  almost  instantaneous  on  the  scale  of  the  
system,  and  therefore  results  in  a  sharper  decision  point.    Such  flip-­‐flops  can  either  be  positive-­‐
edge-­‐triggered  or  negative-­‐edge-­‐triggered.    Positive-­‐edge-­‐triggered  (or  “leading-­‐edge-­‐triggered”)  
flip-­‐flops  respond  to  inputs  when  the  clock  transitions  from  low  to  high  (0  to  1),  while  negative-­‐
edge-­‐triggered  (or  “trailing-­‐edge-­‐triggered”)  flip-­‐flops  respond  when  the  clock  transitions  from  
high  to  low  (1to  0),  as  illustrated  in  Figure  1-­‐3.  

Negative or
Trailing Edge
C

Positive or
Leading Edge  

Figure 1-3: Clock signal


 

The  D  or  “delay”  flip-­‐flop  is  one  example  of  an  edge-­‐triggered  flip-­‐flop.    The  symbol  and  truth  table  
for  a  positive-­‐edge-­‐triggered  D  flip-­‐flop  is  shown  below  in  Figure  1-­‐4.    Note  the  triangle  at  the  “C”  
input  (clock)  on  the  symbol  −  this  denotes  that  the  flip-­‐flop  is  edge  triggered.    Furthermore,  the  
absence  of  a  “bubble”  (    )  at  the  C  input  indicates  that  it  is  positive-­‐edge-­‐triggered.    In  the  truth  
table,  the  arrow  pointing  up  denotes  the  leading  edge  of  the  clock  pulse.  

   

Q C D Qn
D
0 x Qn-1

C 1 x Qn-1
Q
↑ 0 0
This indicates that the flip-flop is ↑ 1 1
positive-edge triggered.  

Figure 1-4: Symbol and truth table for a D flip-flop


 

An  example  of  a  timing  diagram  for  a  D  flip-­‐flop  is  shown  in  Figure  1-­‐5  below.    Note  that  the  state  of  
the  flip-­‐flop  follows  the  input,  but  with  a  delay,  hence  the  name  for  this  flip-­‐flop.  

8
Leading edges

Inputs
D
Output
Q
Q assumed to be initially reset  

Figure 1-5: Example of a timing diagram for a D flip-flop


 

1.2.3    JK  and  T  Flip-­‐Flops  


The  JK  flip-­‐flop  exhibits  a  more  complex  behavior.    The  symbol  and  truth  table  for  a  negative-­‐edge  
triggered  JK  flip-­‐flop  are  shown  below  in  Figure  1-­‐6.    Note  how  the  negative-­‐edge  trigger  is  
indicated  by  the  bubble  and  the  triangle  on  the  clock  input  of  the  symbol.    The  negative  edge  is  
denoted  by  a  downward-­‐pointing  arrow  in  the  truth  table.    This  flip-­‐flop  is  similar  to  an  edge-­‐
triggered  SR  flip-­‐flop  where  K  is  the  reset  input  and  J  is  the  set  input.    Unlike  the  SR  flip-­‐flop,  
however,  the  JK  allows  for  both  reset  and  set  to  be  asserted  simultaneously;  i.e.,  when  both  J  and  K  
are  set  to  logic  1,  the  flip-­‐flop  state  simply  toggles  to  its  opposite  state.  An  example  of  a  timing  
diagram  for  a  JK  flip-­‐flop  is  shown  in  Figure  1-­‐7.  

C J K Qn Label
0 x x Qn-1 memory
J Q
1 x x Qn-1 memory
C
↓ 0 0 Qn-1 memory
K Q
↓ 0 1 0 reset
This indicates that the flip-flop is
negative-edge triggered.
↓ 1 0 1 set
↓ 1 1 Qn-1 toggle
 

Figure 1-6: Symbol and truth table for a JK flip-flop


 

9
Trailing edges

Q
memory reset
memory set toggle toggle  

Figure 1-7: Timing diagram example for a JK flip-flop

One  can  create  a  T  or  “toggle”  flip-­‐flop  by  simply  tying  the  two  inputs  (J  and  K)  of  the  JK  flip-­‐flop  
together.    This  results  in  a  flip-­‐flop  that  will  either  stay  in  the  same  state  when  both  J  and  K  are  logic  
0  (creating  “memory”),  or  toggle  to  the  opposite  state  at  each  trailing  clock  edge  (when  J  and  K  are  
both  logic  1).    The  symbol  and  truth  table  for  the  T  flip-­‐flop  are  shown  below  in  Figure  1-­‐8.    An  
example  of  a  timing  diagram  for  a  T  flip-­‐flop  is  shown  in  Figure  1-­‐9.  

C T Qn Label
T Q 0 x Qn -1 memory
1 x Qn -1 memory
C Q ↓ 0 Qn -1 memory
↓ 1 Qn-1 toggle
 

Figure 1-8: Symbol and truth table for a T flip-flop


 

Q
toggle memory memory
memory toggle toggle  

Figure 1-9: Timing diagram example for a T flip-flop


 

10
1.2.4    Asynchronous  Inputs  
Up  until  now,  all  the  inputs  we  have  examined  have  been  “synchronous”  inputs,  meaning  
that  the  flip-­‐flop  only  responds  to  them  when  they  coincide  with  the  clock.    There  are  occasions,  
however,  where  you’d  like  the  flip-­‐flop  to  respond  regardless  of  the  clock  state.    Such  inputs  are  
called  “asynchronous.”    An  asynchronous  input  that  makes  the  flip-­‐flop  go  to  ‘1’  is  called  a  “preset”  
or  “set”,  and  an  asynchronous  input  that  makes  the  flip-­‐flop  go  to  ‘0’  is  called  a  “clear”  or  “reset”.    
For  example,  the  symbol  and  truth  table  for  the  SR  flip-­‐flop  when  you  add  Preset  and  Clear  
asynchronous  inputs  is  shown  below  in  Figure  1-­‐10.    Note  that  the  asynchronous  inputs  trump  
other  inputs  to  the  system.  An  example  of  a  timing  diagram  with  asynchronous  inputs  is  shown  in  
Figure  1-­‐11.  

Clr Pre R S C Qn
0 0 0 0 x Qn-1
0 0 0 1 1 1
R Pre Q
0 0 1 0 1 0
C
0 0 1 1 1 Not allowed
S Clr Q
0 0 x x 0 Qn-1
0 1 x x x 1
1 0 x x x 0
1 1 x x x Not allowed

Figure 1-10: Symbol and truth table for an SR flip-flop with positive logic asynchronous inputs

This reset is
ignored because
clock is low

R
This set waits until the
clock to take effect
S
This clear takes
effect immediately
Clr
This preset takes
effect immediately
Pre

Q
Preset trumps reset, but when
Clear trumps set.
preset is gone reset is still active.

Figure 1-11: Timing diagram example for SR flip-flop with positive logic asynchronous inputs

11
  Asynchronous  inputs  can  be  added  to  any  of  the  other  flip-­‐flops.    An  additional  wrinkle  is  
that  asynchronous  inputs  often  follow  negative  logic−where  the  active  state  is  ‘0’  instead  of  ‘1’.    An  
example  of  a  JK  flip-­‐flop  with  asynchronous  inputs  using  negative  logic  is  shown  below  in  Figure  
1-­‐12.    Negative  logic  inputs  are  indicated  on  the  symbol  by  bubbles,  and  the  labels  are  usually  Clrn  
and  Prn  for  clear  and  preset,  respectively.    A  timing  diagram  for  this  example  is  shown  in  Figure  
1-­‐13.  

Prn Clrn C J K Qn Label


1 1 0 x x Qn-1 memory
1 1 1 x x Qn-1 memory

J Prn Q 1 1 ↓ 0 0 Qn-1 memory

C 1 1 ↓ 0 1 0 reset
1 1 ↓ 1 0 1 set
K Clrn Q
1 1 ↓ 1 1 Qn-1 toggle
The bubbles
indicate negative 1 0 x x x 0 clear
logic
0 1 x x x 1 preset
0 0 x x x N/A Not allowed
 

Figure 1-12: Symbol and truth table for a JK flip-flop with asynchronous inputs using negative logic
 

Prn

Clrn
clear clear preset

Q
memory set memory reset toggle toggle

Figure 1-13: Timing diagram example for a JK flip-flop using asynchronous inputs with negative
logic

  Negative  logic,  and  the  labeling  system  for  negative  logic  inputs,  is  often  confusing  for  
students.    Pay  special  attention  to  the  fact  that  in  Figure  1-­‐13,  the  asynchronous  inputs  for  “clear”  

12
(Clrn,  meaning  “clear-­‐negative”)  and  “preset”  (Prn)  don’t  affect  the  output  Q  while  set  to  1;  they  
only  affect  the  output  when  they  go  to  0.    Device  designers  generally  make  several  efforts  to  call  
attention  to  negative  logic  conditions.    For  instance,  a  single  input  such  as  Prn  can  be  marked  with  a  
bubble  and  labeled  “ P r n ”,  where  the  bubble,  the  “-­‐n”  suffix,  and  the  overbar  all  serve  as  a  reminder  
that  the  input  uses  negative  logic.    (These  markers  are  not  cumulative!    All  serve  as  a  reminder;  they  
do  not  cancel  each  other  out.)  

1.2.5    Timing  Diagrams  for  Interdependent  Flip-­‐Flops  


Before  moving  on  to  counters,  we  need  a  little  more  practice  with  timing  diagrams  because  
they  get  a  lot  harder  when  you  have  multiple  flip-­‐flops  with  various  inputs  and  outputs  tied  to  one  
another.    The  key  to  analyzing  these  circuits  is  to  understand  that  the  flip-­‐flop  output  decision  is  
based  on  the  input  value  right  before  the  clock  edge.    Let’s  start  with  a  relatively  simple  example.    
Consider  the  logic  circuit  below  in  Figure  1-­‐14  for  which  you  wish  to  determine  the  timing  diagram,  
assuming  that  the  flip-­‐flops  are  initially  reset  (both  Q0  and  Q1  are  logic  0).  
 
Q0 Q1

T0 Q0 T1 Q1

C Q0 C Q1

CLK  

Figure 1-14: Example of circuit with interconnected flip-flops


 
The  first  step  in  analyzing  a  circuit  like  this  is  to  determine  the  inputs  to  the  flip-­‐flops.    For  
this  example,  you  would  note  that  you  have  T  flip-­‐flops,  so  that  the  truth  table  in  Figure  1-­‐8  applies,  
and  you  would  write  down:  
T0 = Q1 T1 = Q0  
 

You  would  then  work  forward  in  time  (left  to  right  in  Figure  1-­‐15),  analyzing  both  flip-­‐flops  
at  each  decision  point  (trailing  clock  edge):  

• Right  before  the  first  decision  point,  Q0  and  Q1  are  both  0  (because  the  problem  
statement  says  “the  flip-­‐flops  are  initially  reset”).    As  CLK  changes:  
o T0  =  Q1  =  0,  so  the  0th  flip-­‐flop  will  stay  the  same            new  Q0  =  0  
o T1  =  NOT  (Q0)  =  1,  so  the  1st  flip-­‐flop  will  toggle.              new  Q1  =  1  
 

• Right  before  the  next  decision  point,  Q0  is  still  0  but  Q1  is  1.  

13
o T0  =  Q1  =  1,  so  the  0th  flip-­‐flop  will  toggle.       new  Q0  =  1  
o T1  =  NOT  (Q0)  =  1,  so  the  1  flip-­‐flop  will  toggle  again.  
st new  Q1  =  0  
 
• Right  before  the  third  decision  point,  Q0  is  1  and  Q1  is  0.  
o T0  =  Q1  =  0,  so  the  0th  flip-­‐flop  will  stay  the  same.     new  Q0  =  1  
o T1  =  NOT  (Q0)  =  0,  so  1  flip-­‐flop  will  stay  the  same.  
st new  Q1  =  0  
 
This  result  is  illustrated  in  the  timing  diagram  below  in  Figure  1-­‐15.    Analyzing  circuits  like  
this  takes  a  little  practice.    For  another  example,  see  the  discussion  of  the  mod-­‐8  synchronous  
counter.    There  are  also  more  opportunities  in  the  problems  at  the  end  of  the  chapter.    Such  circuits  
are  the  basis  for  counters  and  state  machines.    

CLK

T0 = Q1 = 0 T 0 = Q1 = 1
Q0 memory toggle
T 0 = Q1 = 0
memory

T1 = NOT(Q0) = 1 T1 = NOT (Q0) = 0


Q1 toggle
T1 = NOT (Q0) = 1
toggle memory

 
Figure 1-15: Timing diagram for interconnected flip-flop example
 

1.3    Counters  

1.3.1    Basic  Concepts  of  Digital  Counters  


  A  counter  is  a  sequential  digital  circuit  whose  output  progresses  in  a  predictable  repeating  
pattern  with  each  transition  of  the  clock.    For  example,  a  counter  might  count  0,  1,  2,  3,  4,  5,  6,  7,  0,  
1,  2,  3,  4,  5,  6,  7,  0…etc.    In  binary,  the  counter  is  counting  000,  001,  010,  011,  100,  101,  110,  111,  
000…etc.    The  binary  count  state  is  represented  in  this  case  by  a  set  of  three  flip-­‐flops,  with  one  flip-­‐
flop  for  each  binary  digit.    The  states  of  a  counter  can  be  represented  by  a  “state  diagram”  such  as  
the  one  shown  below  in  Figure  1-­‐16.    The  “modulus”  of  a  counter  is  the  number  of  states  it  contains.    
Our  example  has  a  modulus  of  8,  and  would  therefore  be  said  to  be  a  “mod-­‐8”  counter.    Note  that  a  
mod-­‐x  counter  that  starts  at  zero  counts  from  0  to  x-­‐1  (i.e.,  a  mod-­‐8  counter  would  count  0  to  7.  A  
mod-­‐6  would  count  0  to  5,  etc.)    The  maximum  modulus  for  a  counter  consisting  of  n  flip-­‐flops  is  2n.    
A  counter  can  either  count  up  (0,  1,  2,  3,  4,  5,  6,  7,  0,  1,  2,  3…)  in  which  case  it  is  called  an  “up-­‐
counter”,  or  count  down    (7,  6,  5,  4,  3,  2,  1,  0,  7,  6,  5…)  in  which  case  it  is  called  a  “down-­‐counter.”    
The  rest  of  this  section  will  describe  ways  in  which  counters  can  be  implemented.    

14
000 001

111 010

110 011

101 100
 

Figure 1-16: State diagram for mod-8 up-counter which is counting 0,1,2,3,4,5,6,7,0,1,2,…

1.3.2    Ripple  Counters  


A  simple  implementation  of  the  mod-­‐8  up-­‐counter  as  described  by  the  state  diagram  in  Figure  1-­‐16  
is  shown  below  in  Figure  1-­‐17.    The  count  state  is  given  by  Q2  Q1  Q0  (MSB  →  LSB).    Note  how  the  
system  clock  only  attaches  to  the  Q0  flip-­‐flop,  and  Q0  is  the  least  significant  bit  (LSB).  Successive  flip-­‐
flops  are  clocked  by  the  output  from  the  previous  stage.    Note  also  that  the  J  and  K  inputs  are  tied  to  
“1”  so  that  the  flip-­‐flops  are  always  in  the  toggle  mode  (they  could  be  replaced  by  T  flip-­‐flops).        

Q0 Q1 Q2
‘1’ ‘1’ ‘1’
J Prn Q J Prn Q J Prn Q
C C C
K Clrn Q K Clrn Q K Clrn Q

CLK  
Figure 1-17: Mod-8 ripple counter implemented with JK flip-flops
 

The  best  way  to  show  how  this  circuit  works  is  to  examine  the  timing  diagram  for  the  
circuit,  which  is  shown  below  in  Figure  1-­‐18.    Since  only  the  Q0  flip-­‐flop  is  clocked  by  the  system  
clock  (CLK),  the  Q0  output  toggles  with  each  trailing  edge  of  the  system  clock.    This  results  in  the  Q0  
signal  alternating  with  twice  the  period  (half  the  frequency)  of  the  system  clock.    The  Q1  flip-­‐flop  is  
tied  to  Q0,  so  it  toggles  when  Q0  has  a  trailing  edge,  and  the  resulting  signal  has  twice  the  period  as  

15
Q0.    Similarly,  the  Q2  flip-­‐flop  is  tied  to  Q1,  so  it  toggles  when  Q1  has  a  trailing  edge  with  the  result  
that  the  period  doubles  again.    If  you  track  Q2  Q1  Q0,  you’ll  see  that  this  results  in  binary  counting.    
   
The  delay  associated  with  the  flip-­‐flop  response  has  been  exaggerated  in  this  figure  to  
accent  the  “ripple”  effect.    The  delay  is  also  one  downfall  of  ripple  counters.    As  you  scale  up  a  ripple  
counter  to  multiple  bits,  you  incur  more  and  more  delay  between  the  least  significant  bit  (the  
output  of  the  flip-­‐flop  on  the  left)  and  the  most  significant  bit  (the  output  of  the  flip-­‐flop  on  the  
right).    Eventually  this  will  cause  errors  in  the  system.    The  solution  to  this  increasing  delay  is  to  
create  a  “synchronous  counter”  which  is  described  in  a  later  section.  
 

CLK

Q0 ‘0’ ‘1’ ‘0’ ‘1’ ‘0’ ‘1’ ‘0’ ‘1’ ‘0’

Q1 ‘0’ ‘0’ ‘1’ ‘1’ ‘0’ ‘0’ ‘1’ ‘1’ ‘0’

Q2 ‘0’ ‘0’ ‘0’ ‘0’ ‘1’ ‘1’ ‘1’ ‘1’ ‘0’

Q2 Q1Q0 000 001 010 011 100 101 110 111 000

Figure 1-18: Timing diagram for mod-8 ripple counter


 

But  first,  you  might  be  wondering  if  you  can  create  a  ripple  counter  that  isn’t  mod-­‐8,  mod-­‐
16,  or  some  other  modulus  that’s  a  power  of  two.    The  answer  is  that  you  can  by  taking  advantage  of  
the  asynchronous  inputs  on  the  flip-­‐flop.    For  example,  let’s  say  that  you  wish  to  modify  the  counter  
above  so  that  it  just  counts  0-­‐5.    The  way  you  would  do  this  would  be  to  send  an  asynchronous  clear  
signal  to  the  flip-­‐flops  when  the  output  tries  to  go  to  binary  6  (‘110’).    This  is  done  in  Figure  1-­‐19  
below.    Note  that  the  ‘clear’  inputs  have  negative  logic,  so  that  means  you  want  the  input  to  these  
terminals  to  be  ‘1’  most  of  the  time  but  ‘0’  when  the  count  state  goes  to  6.    This  is  accomplished  with  
the  addition  of  the  NAND  gate,  which  will  be  1  all  the  time  except  when  Q1  and  Q2  are  both  1.  So  
when  the  counter  tries  to  go  to  1102  ,  the  clears  will  be  activated  by  the  logic  0  from  the  NAND  gate  
and  the  flip-­‐flops  will  assume  the  000  state.  

16
Q0 Q1 Q2
‘1’ ‘1’ ‘1’
J Prn Q J Prn Q J Prn Q
C C C
K Clrn Q K Clrn Q K Clrn Q

CLK

Figure 1-19: Mod-6 ripple counter (counts 0-5)


 

Figure  1-­‐20  illustrates  the  timing  diagram  for  this  circuit.    Notice  how  the  counter  goes  very  
briefly  to  110,  but  is  quickly  cleared  back  to  000.    The  duration  of  the  “blips”  (the  short  duration  
pulses)  in  the  Q1  output  and  the  Clrn  input  has  been  exaggerated  for  illustration.    Thus,  you  can  
implement  any  modulus  that  is  less  than  the  maximum  modulus  (set  by  the  number  of  flip-­‐flops)  
with  the  prudent  use  of  asynchronous  clears.  

CLK

Q0 ‘0’ ‘1’ ‘0’ ‘1’ ‘0’ ‘1’ ‘0’ ‘1’

Q1 ‘0’ ‘0’ ‘1’ ‘1’ ‘0’ ‘0’ ‘0’ ‘0’

Q2 ‘0’ ‘0’ ‘0’ ‘0’ ‘1’ ‘1’ ‘0’ ‘0’

CLRN

Q2 Q1Q0 000 001 010 011 100 101 000 001

Figure 1-20: Timing diagram for mod-6 counter shown above

To  review,  let’s  say  that  you  are  assigned  the  task  of  designing  a  mod-­‐X  counter  and  wish  to  
use  a  ripple  counter.    First,  you  would  determine  the  number  of  flip-­‐flops  you  need.    Since  the  
maximum  modulus  that  can  be  implemented  with  n  flip-­‐flops  is  2n,  this  means  that  you  should  
determine  the  lowest  power  of  2  that  is  greater  than  or  equal  to  your  desired  modulus  and  use  the  
exponent.    For  example,  let’s  say  you  wish  to  count  0  to  99,  or  mod-­‐100.    The  lowest  power  of  2  that  
exceeds  100  is  128  or  27,  so  you  will  need  7  flip-­‐flops.    For  a  ripple  counter,  these  flip-­‐flops  will  all  
be  T  flip-­‐flops  (or  JK  flip-­‐flops  with  the  input  J  and  K  terminals  set  to  ‘1’  for  toggle),  and  the  clock  for  
each  flip-­‐flop  should  be  tied  to  the  Q  output  of  the  previous  stage,  with  the  clock  input  of  the  first  

17
stage  tied  to  the  system  clock.    The  first  stage  will  always  be  your  least  significant  bit,  and  the  last  
your  most  significant  bit.    
 
Next,  you  need  to  figure  out  the  logic  for  the  clear  so  that  the  count  stops  where  you  want  it.    
For  example,  for  a  0-­‐99  counter,  you  want  the  flip-­‐flops  to  all  clear  when  the  input  reaches  10010,  or  
11001002.    Assuming  that  your  Clear  inputs  use  negative  logic,  that  means  that  you  need  an  
expression  which  is  1  most  of  the  time,  but  0  when  you  reach  the  count  1100100.    This  can  be  
accomplished  by  a  NAND  gate  with  inputs  that  match  the  binary  encoding  of  the  clear  point.    For  
this  example,  you  could  use  the  following  expression  for  CLRN:  
 

(
CLRN = Q6Q5Q4Q3Q2Q1Q0   )
 
In  fact,  you  can  simplify  this  expression  a  bit  more,  because  since  you  plan  to  clear  at  
1100100,  then  you  will  never  reach  1100101,  1100110,  or  1100111,  so  you  don’t  care  about  the  
inputs  that  correspond  to  the  digits  to  the  right  of  the  least  significant  ‘1’  in  your  clear  point.1    Thus  
this  expression  can  be  further  simplified  to:      
 

(
CLRN = Q6Q5Q4Q3Q2   )
 
Armed  with  this  information,  you  could  then  build  the  circuit  to  implement  your  counter.      

1.3.3    Synchronous  Counters  


Ripple  counters  have  problem  of  accumulated  delay,  which  means  they  don’t  scale  well  to  
large  numbers  of  flip-­‐flops.    A  more  elegant  solution  can  be  obtained  by  a  synchronous  counter.    An  
example  of  a  mod-­‐8  synchronous  counter  is  shown  below  in  Figure  1-­‐21.  
 
Q0 Q1 Q2

‘1’ T0 Q0 T1 Q1 T2 Q2

C Q0 C Q1 C Q2

CLK  
Figure 1-21: Mod-8 synchronous up-counter example

                                                                                                                         
1
Actually, you do care a little about these inputs, because you want your system to be able to recover if it
accidentally ends up in one of these unused states. So setting your logic such that the unused states lead to a clear
makes your counter more robust in the face of error.

18
This  circuit  is  called  “synchronous”  because  all  of  the  flip-­‐flops  are  connected  to  the  same  
clock  signal.    To  convince  you  that  this  is  a  counter,  let’s  go  through  the  analysis  of  the  timing  
diagram  using  the  same  process  as  was  previously  introduced.    First,  you  should  note  the  
expressions  for  the  flip-­‐flop  inputs:  

T0 = 1 T1 = Q0 T2 = Q0Q1  

  Next,  you  would  review  the  truth  table  for  the  T  flip-­‐flop,  shown  in  Figure  1-­‐8,  which  tells  us  
that  if  T  is  1  the  output  will  toggle,  and  if  T  is  0  the  output  will  stay  the  same.    Then  you  would  work  
forward  in  time,  analyzing  all  three  flip-­‐flops  at  each  decision  point:      

• At  the  first  decision  point,  assuming  the  flip-­‐flops  are  initially  reset,  Q0,  Q1,  and  Q2  
are  all  0.    
o T0  =  1,  so  the  0th  flip-­‐flop  will  toggle  
o T1  =  Q0  =  0,  so  the  1st  flip-­‐flop  will  stay  the  same.  
o T2  =  Q0Q1  =  0,  so  the  2nd  flip-­‐flop  will  stay  the  same.  
 
• Right  before  the  next  decision  point,  Q0  is  1,  while  Q1  and  Q2  are  still  0.  
o T0  =  1,  so  the  0th  flip-­‐flop  will  toggle  
o T1  =  Q0  =  1,  so  the  1st  flip-­‐flop  will  toggle.  
o T2  =  Q0Q1  =  0,  so  the  2nd  flip-­‐flop  will  stay  the  same.  
 
• Right  before  the  next  decision  point,  Q0  is  0,  Q1  is  1,  and  Q2  is  still  0.  
o T0  =  1,  so  the  0th  flip-­‐flop  will  toggle  
o T1  =  Q0  =  0,  so  the  1st  flip-­‐flop  will  stay  the  same.  
o T2  =  Q0Q1  =  0,  so  the  2nd  flip-­‐flop  will  stay  the  same.  
 
• Right  before  the  next  decision  point,  Q0  is  1,  Q1  is  1,  and  Q2  is  still  0.  
o T0  =  1,  so  the  0th  flip-­‐flop  will  toggle  
o T1  =  Q0  =  1,  so  the  1st  flip-­‐flop  will  toggle.  
o T2  =  Q0Q1  =  1,  so  the  2nd  flip-­‐flop  will  finally  toggle.  
 
The  result  is  shown  below  in  Figure  1-­‐22.  

19
CLK

Q0 ‘0’ ‘1’ ‘0’ ‘1’ ‘0’ ‘1’ ‘0’ ‘1’ ‘0’

Q1 ‘0’ ‘0’ ‘1’ ‘1’ ‘0’ ‘0’ ‘1’ ‘1’ ‘0’

Q2 ‘0’ ‘0’ ‘0’ ‘0’ ‘1’ ‘1’ ‘1’ ‘1’ ‘0’

Q2 Q1Q0 000 001 010 011 100 101 110 111 000
 

Figure 1-22: Timing diagram for mod-8 synchronous up-counter


 

Note  that  this  counter  no  long  exhibits  the  accumulated  delay  in  the  higher  order  digits,  
since  all  the  flip-­‐flops  share  the  system  clock.    As  with  the  ripple  counter,  this  counter  could  be  
modified  to  a  lower  modulus  with  the  use  of  the  asynchronous  clear  inputs.    However,  there  is  a  
general  method  that  can  be  used  for  a  more  elegant  design.    For  that  matter,  the  method  described  
in  the  next  section  can  be  used  to  create  any  state  machine.  

1.4    Counter  Design  


The  focus  of  this  section  is  the  process  by  which  one  can  design  any  state  machine.    We  will  
focus  on  the  design  of  synchronous  counters.    In  brief,  the  design  process  for  a  state  machine  is  as  
follows:  
• Define  the  problem.  
• Draw  a  state  diagram  for  your  system.  
• Make  a  “state  table,”  which  lists  all  possible  Present  States  (in  binary  order),  the  Next  States  
to  which  those  states  should  transition,  and  the  flip-­‐flop  inputs  necessary  to  make  those  
transitions  happen.  
• Use  flip-­‐flop  “excitation  tables”  to  determine  the  inputs  required  to  make  the  desired  
transitions.  
• Determine  combinational  logic  expressions  for  all  of  the  flip-­‐flop  inputs  using  the  present  
state  variables  as  inputs.  
• Implement  the  circuit.  
 
  The  process  is  best  illustrated  through  an  example.    Let’s  say  that  we  wish  to  design  a  mod-­‐
6  synchronous  counter  −  that  is,  a  counter  that  will  count  from  0  to  5.    The  state  diagram  for  this  
counter  is  shown  in  Figure  1-­‐23.  

20
000

101 001

100 010

011

Figure 1-23: State diagram for mod-6 counter design example


 

Furthermore,  let’s  say  that  you  wish  to  build  this  state  machine  using  T  flip-­‐flops  (you  could  use  any  
type).    The  next  step  is  to  construct  the  state  table.  

1.4.1    State  Tables  


  As  described  above,  a  state  table  lists  all  of  the  possible  present  states,  the  next  states  to  
which  you  wish  to  transition  and  the  flip-­‐flop  inputs  necessary  to  make  that  happen.    This  counter  
requires  three  flip-­‐flops,  and  so  you  would  begin  by  making  a  table  like  the  one  shown  in  Figure  
1-­‐24.    Note  that  all  the  possible  present  states  are  listed,  even  though  this  particular  counter  should  
never  be  in  state  110  or  111.    The  table  of  Figure  1-­‐24  is  only  partially  complete,  and  at  this  point,  
this  table  would  be  the  same  for  any  state  machine  that  used  three  T  flip-­‐flops.  

Present  State   Next  State   Flip-­‐Flop  Inputs  

Q2   Q1   Q0   Q’2   Q’1   Q’0   T2   T1   T0  

0   0   0              

0   0   1              

0   1   0              

0   1   1              

1   0   0              

1   0   1              

1   1   0              

1   1   1              

Figure 1-24: Generic state table for any state machine using three T flip-flops

21
  The  next  step  is  to  use  your  state  diagram  to  fill  out  the  “Next  State”  columns,  as  is  shown  in  
Figure  1-­‐25.    For  example,  from  the  state  000,  you  wish  to  progress  to  state  001,  so  you  would  fill  
out  001  in  the  first  row  of  the  Next  State  column.    This  continues  down  the  table.    Note  how  state  
101  goes  to  000.    Finally,  since  you  don’t  expect  to  ever  use  110  and  111,  these  states  lead  to  don’t  
care  conditions.    (This  is  a  little  misleading,  because  in  truth  you  do  care  a  little  about  these  states.    
You  need  to  make  sure  that  if  your  system  inadvertently  lands  in  an  unused  state  −  like  at  power  
start-­‐up  −  that  it  will  resolve  to  a  used  state  and  not  hang  up  in  an  endless  loop.    We  will  revisit  this  
issue  later.)      For  now,  let’s  just  treat  those  states  as  “don’t  cares”  and  mark  them  with  “X”.  

Present  State   Next  State   Flip-­‐Flop  Inputs  

Q2   Q1   Q0   Q’2   Q’1   Q’0   T2   T1   T0  

0   0   0   0   0   1        

0   0   1   0   1   0        

0   1   0   0   1   1        

0   1   1   1   0   0        

1   0   0   1   0   1        

1   0   1   0   0   0        

1   1   0   X   X   X        

1   1   1   X   X   X        

Figure 1-25: State table for mod-6 counter example with Next State columns completed
 

To  complete  the  state  table,  we  now  need  to  figure  out  what  inputs  are  necessary  to  make  
the  flip-­‐flops  behave  as  you  wish.    To  do  this,  we  need  to  reverse  engineer  the  flip-­‐flops.    That  leads  
us  to  excitation  tables.  

1.4.2    Flip-­‐Flop  Excitation  Tables  


An  excitation  table  shows  what  inputs  are  necessary  to  make  a  desired  transition  at  the  next  
clock  edge  for  a  given  flip-­‐flop.    For  example,  with  a  T  flip-­‐flop  if  you  are  in  the  0  state,  and  you  want  
to  go  to  the  1  state,  then  you  need  to  toggle,  requiring  that  the  T  input  be  a  logic  1.    On  the  other  
hand,  if  you’re  in  the  0  state  and  you  want  to  stay  in  the  0  state,  you  want  memory,  which  means  the  
T  input  must  be  logic  0.  The  excitation  tables  for  the  D,  T,  and  JK  flip-­‐flops  are  shown  below  in  
Figure  1-­‐26.  

22
  The  JK  flip-­‐flop  excitation  table  may  appear  confusing  at  first  because  of  the  “don’t  care”  
conditions  in  the  table.    The  transition  from  0  to  0,  for  example,  can  be  accomplished  either  by  
memory  (J  =  0,  K  =  0)  or  by  reset  (J  =  0,  K  =  1),  so  the  value  of  K  doesn’t  matter  as  long  as  J  is  0.    
Similarly,  the  transition  from  0  to  1  can  be  accomplished  either  by  a  toggle  or  a  set,  and  so  forth.  

Desired Desired Desired


D T J K
Transition Transition Transition
0g0 0 0g0 0 0g0 0 X

0g1 1 0g1 1 0g1 1 X

1g0 0 1g0 1 1g0 X 1

1g1 1 1g1 0 1g1 X 0

D Flip-Flop T Flip-Flop JK Flip-Flop  

Figure 1-26: Excitation tables for D, T, and JK flip-flops


 

Armed  with  the  excitation  tables,  we  can  now  complete  the  state  table  for  our  example,  by  
filling  in  the  T  inputs  that  would  give  us  the  desired  state  transitions.    For  example,  for  the  first  row  
in  Figure  1-­‐27,  Q2  must  transition  from  0  to  0,  requiring  “memory”  or  a  T2  value  of  0.    But  for  the  
same  row,  Q0  must  change  from  0  to  1,  requiring  “toggle”  or  a  T0  value  of  1.      

Present  State   Next  State   Flip-­‐Flop  Inputs  

Q2   Q1   Q0   Q’2   Q’1   Q’0   T2   T1   T0  

0   0   0   0   0   1   0   0   1  

0   0   1   0   1   0   0   1   1  

0   1   0   0   1   1   0   0   1  

0   1   1   1   0   0   1   1   1  

1   0   0   1   0   1   0   0   1  

1   0   1   0   0   0   1   0   1  

1   1   0   X   X   X   X   X   X  

1   1   1   X   X   X   X   X   X  

Figure 1-27: Completed state table for mod-6 counter example

23
1.4.3    State  Table  Implementation  
Once  the  state  table  is  complete,  you  have  the  information  you  need  to  determine  the  
necessary  combinational  logic  for  combining  the  flip-­‐flops.    This  requires  determining  a  logic  
expression  for  each  of  the  flip-­‐flop  inputs  as  functions  of  the  present  state.    This  means  that  you  
need  3  K-­‐maps,  one  each  for  T2,  T1  and  T0,  all  as  functions  of  the  present  state  variables,  Q2,  Q1,  and  
Q0.    Note  that  you  don’t  use  the  “Next  State”  columns  of  the  state  table  at  all  in  this  process.    
The  K-­‐maps  and  the  resulting  minimum  sum-­‐of-­‐products  expressions  are  shown  in  Figure  1-­‐28  
below.  
 
Q1Q0 Q1Q0 Q1Q0
Q2 00 01 11 10 Q2 00 01 11 10 Q2 00 01 11 10
0 0 0 1 0 0 0 1 1 0 0 1 1 1 1
T2 T1 T0
1 0 1 X X 1 0 0 X X 1 1 1 X X

T2 = Q1Q0 + Q2Q0 T1 = Q2Q0 T0 = 1


Figure 1-28: K-maps for mod-6 counter example
 

Finally,  we’re  ready  to  draw  our  circuit.    This  is  done  in  Figure  1-­‐29  below.    Note  how  the  3  
flip-­‐flops  all  share  the  same  clock  and  how  the  wiring  for  the  flip-­‐flops  corresponds  to  the  
combinational  logic  expressions  determined  in  Figure  1-­‐28.  

Q0 Q1 Q2

‘1’ T0 Q0 T1 Q1 T2 Q2

C Q0 C Q1 C Q2

CLK
 

Figure 1-29: Implementation of mod-6 counter using T flip-flops


 

Finally,  let’s  return  to  the  subject  of  the  unused  states.    To  have  a  robust  design,  we  need  to  
make  sure  that  if  our  machine  can  recover  if  it  should  inadvertently  fall  into  an  unused  state.    To  do  
this,  we  need  to  look  back  at  our  state  table,  and  determine  what  flip-­‐flop  inputs  result  from  our  
combinational  logic  expressions  for  the  unused  inputs.    For  our  example,  these  inputs  are  shown  in  
italics  in  Figure  1-­‐30  below.  

24
Present  State   Next  State   Flip-­‐Flop  Inputs  

Q2   Q1   Q0   Q’2   Q’1   Q’0   T2   T1   T0  

0   0   0   0   0   1   0   0   1  

0   0   1   0   1   0   0   1   1  

0   1   0   0   1   1   0   0   1  

0   1   1   1   0   0   1   1   1  

1   0   0   1   0   1   0   0   1  

1   0   1   0   0   0   1   0   1  

1   1   0   ?   ?   ?   0   0   1  

1   1   1   ?   ?   ?   1   0   1  

Figure 1-30: State table with flip-flop inputs determined for unused states
 

Then  from  these  inputs,  you  can  determine  to  what  “Next  State”  the  unused  states  would  
transition.    For  example,  state  110  would  set  T2  and  T1  to  0  and  T0  to  1.  This  would  result  in  toggling  
only  the  Q0  bit  so  that  the  state  transitions  to  111.    Once  in  the  111  state,  the  flip-­‐flop  inputs  would  
become  T2  and  T0  =  1,  and  T1  =  0.    These  inputs  would  cause  Q2  and  Q0  to  toggle  on  the  next  clock  
edge,  while  Q1  would  remain  the  same,  making  the  next  state  010,  which  is  within  the  proper  count  
sequence.  

25
Present  State   Next  State   Flip-­‐Flop  Inputs  

Q2   Q1   Q0   Q’2   Q’1   Q’0   T2   T1   T0  

0   0   0   0   0   1   0   0   1  

0   0   1   0   1   0   0   1   1  

0   1   0   0   1   1   0   0   1  

0   1   1   1   0   0   1   1   1  

1   0   0   1   0   1   0   0   1  

1   0   1   0   0   0   1   0   1  

1   1   0   1   1   1   0   0   1  

1   1   1   0   1   0   1   0   1  

Figure 1-31: Complete state table for mod-6 counter example, including unused states

From  Figure  1-­‐31,  one  can  then  determine  a  state  diagram  for  the  system  that  includes  even  
the  unused  states,  and  this  is  shown  in  Figure  1-­‐32.    This  diagram  indicates  that  our  system  is  
robust,  because  if  the  machine  should  land  in  the  110  or  111  state,  it  will  quickly  return  to  the  main  
counting  loop.    Therefore,  no  further  design  modifications  are  necessary.    If  we’d  found,  for  
example,  that  111  transitioned  back  to  110,  then  we’d  have  an  endless  loop,  so  we  would  need  to  
adjust  our  design.  

000
110
101 001

111
100 010

011

Figure 1-32: State diagram for mod-6 counter example, where unused states are shown

26
1.4.4    Additional  State  Machine  Design  Examples  

1.4.4.1    Mod-­‐6  Counter  Using  JK  Flip-­‐Flops  


The  same  state  machine  can  be  made  out  of  any  type  of  flip-­‐flop.    As  an  example,  let’s  repeat  
the  mod-­‐6  counter  design,  but  instead  use  JK  Flip-­‐Flops.    Our  state  table,  shown  in  Figure  1-­‐33  
below,  would  have  the  same  “Present  State”  and  “Next  State”  columns  as  the  previous  example,  but  
would  now  have  6  columns  for  flip-­‐flop  inputs.    This  makes  a  little  more  work,  but  at  least  with  JK  
flip-­‐flops  you  get  a  lot  of  don’t  care  conditions  that  simplify  the  K-­‐maps,  which  are  shown  in  Figure  
1-­‐34.  

Present  State   Next  State   Flip-­‐Flop  Inputs  

Q2   Q1   Q0   Q’2   Q’1   Q’0   J2   K2   J1   K1   J0   K0  

0   0   0   0   0   1   0   X   0   X   1   X  

0   0   1   0   1   0   0   X   1   X   X   1  

0   1   0   0   1   1   0   X   X   0   1   X  

0   1   1   1   0   0   1   X   X   1   X   1  

1   0   0   1   0   1   X   0   0   X   1   X  

1   0   1   0   0   0   X   1   0   X   X   1  

1   1   0   X   X   X   X   X   X   X   X   X  

1   1   1   X   X   X   X   X   X   X   X   X  

Figure 1-33: State table for mod-6 counter implemented with JK flip-flops
 

Q1Q0 Q1Q0 Q1Q0


Q2 00 01 11 10 Q2 00 01 11 10 Q2 00 01 11 10
0 0 0 1 0 0 0 1 X X 0 1 X X 1
J2 J1 J0
1 X X X X 1 0 0 X X 1 1 X X X

J 2 = Q1Q0 J1 = Q2Q0 J0 = 1
Q1Q0 Q1Q0 Q1Q0

Q2 00 01 11 10 Q2 00 01 11 10 Q2 00 01 11 10
0 X X X X 0 X X 1 0 0 X 1 1 X
K2 K1 K0
1 0 1 X X 1 X X X X 1 X 1 X X

K 2 = Q0 K1 = Q0 K0 = 1
 
Figure 1-34: K-maps for mod-6 counter implemented with JK flip-flops

27
We  leave  it  to  the  reader  to  complete  the  circuit  design  from  this  point.    For  more  practice,  
let’s  look  at  another  state  machine.  

1.4.4.2    2-­‐Bit  Gray  Scale  Counter  


In  a  gray  scale  counter,  only  one  digit  changes  with  each  transition.    So,  for  example,  a  2-­‐bit  
gray  scale  counter  could  count  00,  01,  11,  10,  00,  01,…    The  state  diagram  for  this  is  shown  below  in  
Figure  1-­‐35.  

00

10 01

11
 

Figure 1-35: State diagram for 2-bit gray-scale counter


 

We  will  implement  this  counter  with  D  flip-­‐flops.    The  state  table  for  this  counter  is  shown  
below  in  Figure  1-­‐36.    Note  how  the  states  are  listed  in  binary  counting  order.    The  K-­‐maps  and  
resulting  circuit  for  this  example  are  shown  in  Figure  1-­‐37  and  Figure  1-­‐38.  

Present  State   Next  State   Flip-­‐Flop  


Inputs  

Q1   Q0   Q’1   Q’0   D1   D0  

0   0   0   1   0   1  

0   1   1   1   1   1  

1   0   0   0   0   0  

1   1   1   0   1   0  

Figure 1-36: State table for 2-bit gray-scale counter


 

28
Q1 Q1
Q0 0 1 Q0 0 1
0 0 0 0 1 0
D1 D0
1 1 1 1 1 0

D1 = Q0 D0 = Q1  
Figure 1-37: K-maps for flip-flop inputs in 2-bit gray-scale counter example
 

Q0 Q1

D0 Q0 D1 Q1

C Q0 C Q1

CLK  
Figure 1-38: Circuit for 2-bit gray-scale counter
   

29
1.6    Homework  Problems  
 

Problem  1-­‐1.  Complete  the  timing  diagram  below  for  the  D  flip-­‐flop  shown.  You  may  assume  that  
the  flip-­‐flop  is  initially  reset  (Q  =  0).    

D Q
C
C Q
D

Q
 

Problem  1-­‐2.  Complete  the  timing  diagram  below  for  the  SR  flip-­‐flop  shown  below  (and  in  Fig  1-­‐1).  

R Pre Q
C C
S Clr Q
R

Clr

Pre

Q
 

30
Problem  1-­‐3.  Complete  the  timing  diagram  below  for  the  JK  flip-­‐flop  shown  in  Fig  1-­‐6.  

Prn

Clrn

Q
 

Problem  1-­‐4.  Create  a  timing  diagram  covering  6  clock  cycles  for  the  sequential  logic  circuit  below.    
Determine  whether  this  circuit  is  an  up-­‐counter,  down-­‐counter,  or  some  other  state  machine,  and  if  
a  counter  determine  its  modulus.    Assume  that  the  flip-­‐flops  are  initially  reset.  

Q0 Q1

T0 Q0 T1 Q1

C Q0 C Q1

CLK  

31
Problem  1-­‐5.  Create  a  timing  diagram  covering  6  clock  cycles  for  the  sequential  logic  circuit  below.    
Determine  whether  this  circuit  is  an  up-­‐counter,  down-­‐counter,  or  some  other  state  machine,  and  if  
a  counter  determine  its  modulus.    Assume  that  the  flip-­‐flops  are  initially  reset.  

Q0 Q1 Q2

J0 Q0 J1 Q1 J2 Q2
C C C
‘1’ K0 Q0 K1 Q1 ‘1’ K2 Q2

CLK  

Problem  1-­‐6.  Determine  the  state  table  and  state  diagram  for  the  state  machine  shown  above  in    
Problem  1-­‐5,  including  unused  states  (since  you  have  the  circuit  there  should  be  no  ‘x’s  in  your  
table).  
 
Problem  1-­‐7.  Design  a  mod-­‐16  ripple  up-­‐counter.  
 
Problem  1-­‐8.  Design  a  mod-­‐10  ripple  up-­‐counter.  
 
Problem  1-­‐9.  Draw  the  circuit  that  would  complete  the  example  of  a  mod-­‐6  counter  using  JK  flip-­‐
flops,  with  the  state  table  given  in  Figure  1-­‐33.  
 
Problem  1-­‐10.    Draw  the  complete  state  diagram,  including  the  110  and  111  states,  for  the  JK  
implementation  of  the  mod-­‐6  counter,  with  the  state  table  given  in  Figure  1-­‐33.  (This  will  require  
you  to  determine  how  your  logic  handled  the  “don’t  care”  states  in  the  table).  
 
Problem  1-­‐11.  Design  a  mod-­‐6  counter  using  D  flip-­‐flops.  
 
Problem  1-­‐12.  Design  a  3-­‐bit  gray  scale  counter,  which  would  count  000,  001,  011,  010,  110,  111,  
101,  100.    Use  JK  flip-­‐flops.  
 
Problem  1-­‐13.  Design  a  synchronous  mod-­‐10  up-­‐counter  using  T  flip-­‐flops.  
 

32
Chapter  2:   Digital  and  Analog  Conversion  
2.1    ADC  and  DAC  Concepts  
By  now  you  should  have  a  sense  of  how  analog  and  digital  signals  are  different.    An  analog  
signal  is  a  “real-­‐world”  signal.    It  can  take  on  any  value  and  can  change  continuously.    A  digital  
signal,  on  the  other  hand,  is  a  stream  of  binary  numbers.    To  convert  an  analog  signal  into  a  digital  
signal,  the  analog  signal  must  first  be  sampled,  then  quantized,  and  then  encoded  as  a  binary  
number.    The  signal  is  then  in  a  form  where  it  can  be  stored  (like  on  a  compact  disk)  or  manipulated  
using  the  digital  system  techniques  you’ve  already  studied.    To  convert  a  digital  signal  back  to  an  
analog  signal,  the  binary  numbers  making  up  the  signal  must  be  translated  into  an  analog  output  
voltage.    The  figure  below  illustrates  these  processes.    
 
 
Signal (V)

time
ADC 011001110101001001010101011101010001

Signal (V)

011001110101001001010101011101010001
DAC
time

 
Figure 2-1: Illustration of ADC and DAC processes
 
 
Sampling  is  the  first  process  involved  in  the  conversion  of  an  analog  into  a  digital  signal.  
Sampling  is  the  measurement  of  a  signal  at  discrete  and  regular  times.  Hourly  sampling  of  the  
temperature  outside  would  result  in  a  sequence  of  numbers,  one  for  each  hour.  Usually  the  sample  
times  are  uniformly  spaced.  To  avoid  losing  any  information  the  samples  have  to  be  spaced  closely  
enough  together  so  that  the  shape  of  the  analog  input  signal  is  not  distorted  or  lost.  Music  stored  in  
a  CD  would  not  sound  very  good  if  the  sampling  rate  were  1  KHz.  
 
How  fast  is  fast  enough?  The  Sampling  Theorem  states  that  to  avoid  loss  of  information,  a  
band  limited  signal  must  be  sampled  at  a  rate  equal  to  or  greater  than  twice  the  bandwidth  of  the  

33
signal.  If  we  are  dealing  with  a  signal  containing  frequency  components  from  about  zero  on  up  to  
some  maximum  frequency,  fsig,max,  then  the  sampling  rate,  fsample,  must  be  equal  to  or  greater  than  
twice  fsig,max.  

  f sample ≥ 2 f sig ,max     (2.1)  

  This  minimum  rate  is  called  the  Nyquist  rate,  named  after  the  engineer  who  investigated  the  
mathematics  of  the  sampling  process.  The  theoretical  limit  is  never  really  fast  enough.  For  example,  
to  make  music  CD  recordings,  the  input  signal,  which  has  a  maximum  frequency  of  20  KHz,  is  
sampled  at  about  44  KHz.  

Many  signals  have  high  frequency  components  that  do  not  contain  essential  information  but  
that  can  cause  problems  when  sampling  is  done.  The  problem  of  aliasing  occurs  when  the  sampling  
rate  is  lower  than  twice  the  highest  frequency  of  the  signal.  It  results  in  high  frequency  components  
masquerading  as  lower  frequency  values  and  causing  distortion.  Musical  instruments  can  create  
frequencies  higher  than  20  KHz  which  are  not  audible.  To  avoid  aliasing  problems,  a  music  signal  is  
first  low  pass  filtered  to  remove  any  components  greater  than  20  KHz.  (This  filter  is  also  called  an  
anti-­‐aliasing  filter).    This  is  what  is  meant  by  band  limiting  a  signal.  Filtering  is  used  to  remove  all  
but  a  limited  range  of  frequencies  from  a  signal  while  preserving  the  essential  information  content.  
If  all  frequencies  above  about  3  KHz  are  removed  from  a  person’s  voice  before  telephone  
transmission,  the  voice  remains  both  intelligible  and  recognizable  although  they  may  not  sound  
exactly  the  same  as  in  person.  
 
Now  let’s  consider  how  the  number  of  bits  used  to  encode  the  signal  affects  the  signal.  Look  
again  at  Figure  2-­‐1.    Can  you  see  how  the  signal  that  emerged  from  the  DAC  is  different  from  the  
original  signal  sent  into  the  ADC?    It’s  blockier  and  would  sound  differently  to  your  ear  than  the  
original  signal.      That  ‘blockiness’  is  called  quantization  noise,  and  it’s  the  inevitable  result  of  
limiting  the  signal  to  a  finite  number  of  voltage  levels  in  the  quantization  process.      The  more  
voltage  levels  you  allow  in  the  system,  the  less  quantization  noise  you  will  have  and  the  closer  the  
final  signal  will  be  to  the  original.      You  get  more  voltage  levels  by  simply  using  more  bits  to  encode  
each  sample  reading.    Of  course,  the  more  bits  you  use  and  the  faster  you  sample,  then  the  larger  
the  total  bit  rate  for  your  signal  becomes,  making  greater  demands  on  your  processing  system  and  
signal  storage  requirements.    The  bit  rate  for  your  system  is  the  product  of  the  sample  rate  and  the  
number  of  bits  for  each  sample.  
 

2.2    Digital  to  Analog  Conversion  


We’ll  start  with  digital  to  analog  conversion,  because  this  process  is  more  straightforward.    
Let’s  consider  a  system  that  converts  an  n-­‐bit  digital  number  into  an  analog  signal  that  will  range  in  
value  from  va,min  to  va,max.      With  n  bits,  the  number  of  levels  in  the  system  will  be  2n,  and  the  
resolution,  𝛿𝑣  will  be  given  by  

va ,max − va ,min
  δv =     (2.2)  
2n − 1

34
Why  is  the  denominator  2n-­‐1  and  not  2n?    The  key  is  that  the  resolution  is  the  space  between  
levels.    Consider  a  2-­‐bit  signal  where  ‘00’  will  correspond  to  0V,  ‘01’  will  correspond  to  2V,  ‘10’  will  
correspond  to  4V  and  ‘11’  will  correspond  to  6V.      There  are  four  levels  in  this  system  (0,  2,  4,  and  6  
V)  but  the  resolution  is  the  full  range  divided  by  3,  or  2V.      

One  circuit  that  can  be  used  to  implement  a  DAC  is  a  weighted  summing  amplifier.    An  
example  of  a  4-­‐bit  weighted  summer  DAC  is  shown  in  Figure  2-­‐2  below.    This  circuit  converts  the  4-­‐
bit  number  given  by  b3b2b1b0  into  a  voltage.    VHI  is  the  voltage  level  corresponding  to  a  high  for  the  
digital  signal.    Notice  how  the  resistors  in  the  input  branches  of  the  circuit  progress  by  powers  of  
two,  with  the  largest  input  resistor  corresponding  to  the  least  significant  bit.  

+VHI

R0/8
b3

R0/4
b2
RF
R0/2
b1 +VCC

R0
b0
Vout

-VEE  
Figure 2-2: 4-Bit Weighted Summer DAC
 
Analyzing  this  circuit,  you  would  have  the  following  equation  for  the  output  voltage  (with  
the  powers  of  two  explicitly  written  out  for  emphasis):  
 
⎛ R R R R ⎞
  v0 = −VHI ⎜ F 3 b3 + F 2 b2 + F 1 b1 + F 0 b0 ⎟     (2.3)  
⎝ R0 2 R0 2 R0 2 R0 2 ⎠

Simplifying  this  expression,    


 
VHI RF 3
  v0 = −
R0
( 2 b3 + 22 b2 + 21 b1 + 20 b0 )     (2.4)  

So  the  output  voltage  for  this  circuit  is  the  decimal  conversion  of  the  binary  number  with  a  scale  
factor  of  –VHIRF/R0.    The  resolution  for  this  DAC  can  be  related  to  its  component  values  as  follows:  

35
VHI RF
  δv =     (2.5)  
R0

As  an  example,  imagine  that  you  wish  to  convert  a  4-­‐bit  digital  signal  into  an  analog  voltage  
with  a  total  range  of  0  to  –15  V,  and  that  a  logical  ‘1’  in  your  digital  system  is  represented  by  5V.    
You  would  use  (2.2)  to  determine  the  desired  resolution:  
 
0V − (−15V)
  δv = = 1V     (2.6)  
24 − 1
Then  you  would  choose  values  for  the  input  resistors.    For  op-­‐amp  circuits,  its  best  to  keep  
all  resistor  values  between  1  kΩ  and  1  MΩ.    A  good  choice  here  would  be  to  set  R0  to  80  kΩ,  which  
makes  the  smallest  resistor  in  the  input  branches  10  kΩ  (for  the  b3  input).    You  would  then  use  
(2.5)  to  determine  the  necessary  value  for  RF.  

δ v ⋅ R0 1V ⋅ 80kΩ
  RF = = = 16kΩ     (2.7)  
VHI 5V

This  circuit  can  be  easily  modified  for  fewer  or  more  bits.    For  a  3-­‐bit  DAC,  for  example,  you  
would  remove  the  b2  branch  of  the  circuit.    Or  for  a  5-­‐bit  DAC  you  would  add  a  branch  for  b4  with  a  
resistor  value  of  R0/16.    In  either  case,  the  resolution  is  still  given  by  (2.5).    The  output  can  also  be  
inverted  by  a  unity-­‐gain,  inverting  amplifier  if  a  positive  output  voltage  is  desired.  
 

2.3    Analog  to  Digital  Conversion  


How  about  the  analog  to  digital  conversion  process?    It  turns  out  then  when  you  consider  
the  process  of  going  from  analog  to  digital,  the  relationship  between  the  step  size  and  the  full  
voltage  range  is  given  by:  
va ,max − va ,min
  δv =     (2.8)  
2n
Why  the  difference  from  the  digital-­‐to-­‐analog  process?  Because  when  you  are  converting  from  
analog  to  digital  you  are  collapsing  a  voltage  range  into  a  single  value.    Let’s  say  you  are  converting  
an  analog  voltage  ranging  from  0  to  4  V  into  a  2-­‐bit  number.      One  approach  would  be  to  convert  0-­‐1  
V  into  00,  1-­‐2  V  into  01,  2-­‐3  V  into  10,  and  3-­‐4  V  into  11.    Note  that  the  maximum  quantization  error  
for  this  scheme  is  the  same  as  the  step  size.        A  better  approach  is  to  offset  the  voltage  levels  by  half  
a  step  size,  so  in  this  example  0-­‐0.5  would  convert  to  00,  0.5-­‐1.5  to  01,  1.5-­‐2.5  to  10  and  2.5-­‐4.0  to  
11.    The  maximum  quantization  error  for  this  scheme  is  still  the  step  size,  but  for  most  of  the  
voltage  range,  the  quantization  error  is  only  half  a  step.      

There  are  a  number  of  methods  used  to  convert  analog  to  digital.    Here  we  will  look  at  a  
method  used  for  high-­‐speed  conversion  called  “Flash  ADC,”  but  there  are  many  other  ADC  methods.    
First,  however,  we  need  to  understand  an  op-­‐amp  circuit  called  a  comparator.  

36
2.3.1    The  Comparator  
A  comparator  is  an  op-­‐amp  operating  without  feedback.  We  are  sometimes  so  used  to  op-­‐
amps  with  negative  feedback  that  we  forget  that  the  op-­‐amp  is  really  a  very  simple  device,  with  the  
voltage  transfer  characteristic  shown  below  in  Figure  2-­‐3.    With  negative  feedback,  you  keep  in  the  
op-­‐amp  in  the  region  where  ε~  0,  but  when  you  remove  the  feedback  you  no  long  place  any  
constraints  on  ε.  

Vout
VS+ VS+

Vin +
ε
-
Vout ε
VREF
VS- VS-
 
Figure 2-3: A comparator is an op-amp operating without feedback. Its transfer function is shown on the
right.
 

So,  the  comparator  is  actually  a  very  basic  A  to  D  converter.    It  accepts  an  analog  input,  and  
outputs  one  of  two  values  that  are  determined  by  the  power  supplies  to  the  op-­‐amp.      The  output  of  
this  circuit  can  be  summarized  as  follows:  

Vout = VS + ε > 0
      (2.9)  
Vout = VS − ε < 0

VS+  and  VS-­‐  don’t  need  to  be  symmetric,  so  you  can  set  VS+  to  VHI  for  your  logic  system  and  VS-­‐  
to  VLO,  to  create  a  1-­‐bit  A/D  converter.    Comparators  are  used  in  most  multiple-­‐bit  A/D  converters  
as  well,  as  we  shall  see  in  the  Flash  ADC.        

2.3.2    Flash  ADC  


Flash  ADC  is  one  of  the  fastest  ADC  techniques,  because  unlike  other  methods  it  is  fully  
parallel.    We  shall  see,  however,  that  this  method  does  not  easily  scale  to  systems  with  many  bits.    A  
block  diagram  for  a  Flash  ADC  (this  shows  a  3-­‐bit  system)  is  shown  in  Figure  2-­‐4  below.    

37
I6
I5
Series of Op-Amps Encoder
Analog Input (Va) I4 converts B2
as Comparators
I3 comparator B1
I2 outputs into 3-
I1 bit binary B0
number
I0

Resistor network
provides voltage
reference levels
 
Figure 2-4: Block diagram for 3-bit Flash ADC
 
The  input  is  fed  into  a  series  of  comparators,  where  the  reference  voltages  have  been  set  by  
a  resistor  ladder  to  span  the  total  input  voltage  range.      Each  comparator  will  yield  a  low  voltage  if  
the  input  is  less  than  its  reference  value,  and  a  high  voltage  if  the  input  is  greater  than  its  reference  
value.      So  when  the  input  is  at  0V,  all  of  the  comparators  produce  low  outputs,  and  when  the  input  
is  at  its  maximum,  all  the  comparators  produce  high  outputs.    When  the  input  is  somewhere  in  
between  all  of  the  comparators  for  which  the  input  exceeds  the  reference  voltages  will  be  high,  and  
the  remaining  comparators  will  be  low.      The  comparator  outputs  can  then  be  translated  into  the  
appropriate  binary  output  through  a  combinational  logic  circuit  (the  encoder).  The  truth  table  for  
the  encoder  is  given  in  Error!  Reference  source  not  found..  
A  3-­‐bit  Flash  ADC  is  illustrated  in  Figure  2-­‐5.    Note  how  the  bottom  and  top  resistors  of  the  
resistor  ladder  are  different  from  the  others,  this  provides  the  ½  step  offset  that  reduces  average  
quantization  error.    Note  also  how  the  3-­‐bit  Flash  ADC  requires  8  resistors  and  7  comparators.    In  
general,  an  n-­‐bit  Flash  ADC  will  require  2n  resistors  and  2n-­‐1  comparators—so  it  doesn’t  scale  so  
well  when  n  gets  big!      Other  techniques,  such  as  successive  approximation  ADC  scale  better  for  
systems  with  a  large  value  for  n.  

38
+VFull Range

3R/2 +VHI
VA
GND
R +VHI
I6
I5 E
GND N
R +VHI I4 B2
C
I3 B1
O
I2
R
GND D B0
+VHI I1
E
I0 R
GND
R
+VHI

R GND
+VHI

GND
R
+VHI

R/2 GND

GND

 
Figure 2-5: Circuit for a 3-bit Flash ADC

39
Comparator  Outputs   Encoder  Output  

I6   I5   I4   I3   I2   I1   I0   B2   B1   B0  

0   0   0   0   0   0   0   0   0   0  

0   0   0   0   0   0   1   0   0   1  

0   0   0   0   0   1   1   0   1   0  

0   0   0   0   1   1   1   0   1   1  

0   0   0   1   1   1   1   1   0   0  

0   0   1   1   1   1   1   1   0   1  

0   1   1   1   1   1   1   1   1   0  

1   1   1   1   1   1   1   1   1   1  

All  other  input  combinations   X   X   X  

Table 2-1: Truth Table for 3-Bit Flash ADC Encoder

 
   

40
2.4    Homework  Problems  
 
Problem  2.1.    Music  for  a  CD  is  sampled  at  44.1  kHz,  with  16  bits  for  each  sample.  
a. What  is  the  bit  rate  for  a  CD?  
b. How  many  bits  are  required  to  store  a  2  minute  song?  
c. If  the  capacity  of  a  CD  is  700  MB,  how  many  minutes  of  music  can  be  stored  on  a  single  
disk?  
d. If  original  analog  audio  signal  had  a  range  of  5V,  what  is  the  step  size  or  maximum  
quantization  error,  for  the  analog-­‐to-­‐digital  conversion?  
 
Problem  2.2.    In  order  to  definitively  answer  the  question  about  a  tree  falling  in  the  forest  with  no  
one  to  hear  it,  Dr.  Zen  plans  to  record  forest  sounds.    The  frequencies  generated  by  a  falling  tree  
range  from  nearly  DC  to  10  kHz,  and  Dr.  Zen  also  plans  to  capture  the  gentle  call  of  the  birds,  which  
can  go  as  high  as  15  kHz.    
 
a. What  sample  frequency  should  Dr.  Zen  use?  
b. If  he  were  to  use  an  anti-­‐aliasing  filter  at  the  input  to  his  ADC  (to  prevent  higher  
frequency  signals  from  interfering  with  his  experiment),  what  should  the  cut-­‐off  
frequency  for  that  filter  be?  
 
Problem  2.3.  Design  a  3-­‐bit  DAC  with  a  step-­‐size  of  1  V  assuming  a  logic  system  where  5V  is  the  
logic  high.    What  would  be  the  output  voltage  range  for  this  DAC?  
 
Problem  2.4.    Design  a  4-­‐bit  DAC  with  a  step-­‐size  of  0.2  V  assuming  a  logic  system  where  5V  is  the  
logic  high.    What  would  be  the  output  voltage  range  for  this  DAC?  
 
Problem  2.5.    How  many  resistors  would  be  required  for  a  16-­‐bit  Flash  ADC?    How  many  
comparators  would  be  required?        
 
Problem  2.6      Design  a  2-­‐bit  Flash  ADC  for  an  input  voltage  range  of  5V.    You  can  leave  the  encoder  
as  a  block.  
 
Problem  2.7.    Design  the  combinational  logic  circuit  for  the  encoder  in  problem  2.6.  
 
Problem  2.8.    Look  up  the  successive  approximation  ADC  method  on  the  web  and  describe  how  
this  technique  works.      Successive  approximation  ADCs  are  slower  than  Flash  ADCs,  but  scale  better  
to  large  bit  systems.  
 

41
Chapter  3:   Introduction  to  Communications  
3.1    Introduction  
  Electronic  communications  is  the  transfer  of  information  from  one  location  to  another.  
Distances  involved  can  be  as  little  as  a  few  inches  over  bus  lines  within  a  computer  or  as  much  as  
hundreds  of  thousands  of  miles  for  video  information  from  a  deep  space  probe.  It  would  be  difficult  
to  find  another  area  of  technology  which  touches  more  people’s  lives.  Modern  communications  
systems  include  television,  radio,  telephone,  the  global  positioning  system  (GPS),  and  Internet.    This  
list  could  go  on  and  on.  Today’s  communications  systems  are  often  mixes  of  older  technologies  such  
as  amplitude  and  frequency  modulation  (AM  and  FM)  and  newer  technologies  such  as  fiber  optics,  
GPS,  satellites  and  digital  communications.  Communications  is  a  very  dynamic  area  with  advances  
being  made  every  day.  One  area  which  is  very  active  is  digital  communications.  HDTV  will  use  a  
digital  format  and  is  not  too  far  off.  Several  local  TV  stations  already  have  transmitter  facilities  for  
HDTV.  Digital  communications  is  very  important  to  the  military  (for  one  thing  it  lends  itself  well  to  
encryption).  Therefore,  a  section  on  digital  communications  has  been  included  in  these  notes.  To  
master  more  than  a  small  part  of  communications  would  take  years  of  study  (a  career).  These  notes  
are  intended  to  be  a  short  overview  of  several  areas  in  this  field.  

3.2  Communication  Systems    


  The  primary  function  of  a  communication  system  is  to  transfer  information  from  one  
location  to  another.  A  block  diagram  of  a  generalized  communication  system  is  shown  in  Figure3-­‐.1.  

Figure 3-1: Block Diagram of a Communication System.

  We  have  already  studied  many  of  the  circuits  and  the  functions  they  perform  which  are  
used  in  communications  systems.  We  have  studied  filters  and  amplifiers,  both  important  building  
blocks.  

42
  If  we  look  at  our  system  one  block  at  a  time,  the  first  thing  we  see  is  a  pair  of  transducers,  
one  at  the  input  and  one  at  the  output.  Their  function  is  to  convert  non-­‐electrical  signals  into  
electrical  signals  and  electrical  signals  back  into  non-­‐electrical  form.  Typical  communication  input  
transducers  are  microphones,  computer  keyboards  and  TV  cameras;  while  typical  output  
transducers  are  loudspeakers,  printers,  and  cathode  ray  tubes  (CRTs).  

  The  input  and  output  processors  generally  consist  of  electronic  subsystems  which  perform  
basic  functions  that  prepare  the  output  of  a  transducer  for  transmission,  or  the  output  of  a  receiver  
into  a  form  that  the  output  transducer  can  handle.  Typical  processors  include:  filters,  scalers,  
multipliers,  adders,  encoders,  decoders,  code  converters,  transformers,  analog-­‐to-­‐digital  and/or  
digital-­‐to-­‐analog  converters.  The  heart  of  the  communication  system  lies  not  with  these  important,  
though  peripheral,  devices,  but  rather  with  the  three  central  blocks:  the  transmitter,  the  channel,  
and  the  receiver.  Let’s  consider  each  of  these  separately.  

  The  primary  function  of  the  transmitter  is  to  accept  the  information  baring  input  signal  
from  the  transducer  or  the  input  processor  and  make  it  suitable  for  injection  into  the  channel.  The  
primary  signal  processing  which  occurs  in  the  transmitter  is  modulation.  Modulation  is  a  process  
which  encodes  the  lower  frequency  information  (often  audio)  to  be  transmitted  onto  a  radio  
frequency  sinusoidal  carrier  or  a  pulse  train  (which  could  be  digital  in  the  form  of  1’s  and  0’s)  
before  its  insertion  into  the  channel.  

  The  channel  is  the  medium  through  which  the  signal  must  travel  in  going  from  the  
transmitter  to  the  receiver,  for  example,  optical  fibers,  telephone  lines,  coaxial  cable  or  even  the  
open  atmosphere.  The  receiver  must  capture  the  signal  from  the  channel  and  deliver  it  to  the  
output  processor  and  transducer.  As  might  be  expected,  the  primary  function  of  the  receiver  is  to  
demodulate  the  signal  captured  from  the  channel.  Demodulation  is  the  reversal  of  the  modulation  
process  which  occurred  in  the  transmitter.  In  an  ideal  situation,  if  x(t),which  contains  the  
information,  is  modulated  onto  a  higher  frequency  carrier  and  transmitted  to  the  receiver,  the  
output  y(t)  of  a  suitable  demodulator  will  equal  x(t).  Much  effort  is  spent  by  communications  
engineer  in  trying  to  achieve  and  maintain  this  ideal  situation.  In  a  typical  communication  system  
the  source  of  most  of  the  difficulties  in  achieving  this  ideal  communication  is  the  channel.  Occurring  
in  the  channel  are  five  undesired  effects:  spreading,  attenuation,  distortion,  interference  and  noise.  

Figure 3-2: An Ideal Modulator-Demodulator System.

43
Chapter  4:    Amplitude  Modulation  

4.1  Introduction  
  When  we  transfer  information  from  one  system  or  subsystem  to  another  we  want  the  information  to  be  
transferred  with  accuracy  and  speed.  An  important  technique  has  been  developed  and  refined  over  the  past  80  
years  or  so  which  enables  us  to  transfer  information  and  recover  it  with  considerable  ease  and  accuracy.  This  
technique,  called  modulation,  is  the  process  of  superimposing  low  frequency  (voice,  music)  information,  or  
intelligence,  onto  a  high  frequency  carrier.  

  The  motivation  for  modulating  a  signal  is  primarily  two-­‐fold.  One  difficulty  associated  with  signal  
transmission  deals  with  practical  antenna  size.  Suppose  we  want  to  transmit  an  audio  signal  the  way  an  AM  
radio  broadcast  station  does.  The  spectrum  of  x(t)  would  be  from  about  100  Hz  to  5  KHz.  It  is  this  spectrum  we  
wish  to  transmit  if  we  wish  to  convey  all  the  information  of  high  and  low  frequencies  to  our  listeners.    

  From  physics  we  know  that  if  we  wish  to  transmit  our  signal  efficiently,  an  antenna  must  be  used  whose  
length  is  about  equal  to  the  wavelength  of  the  frequency  we  want  to  transmit.  Wavelength,  λ,  and  frequency,  f,  
are  related  by  λ=  c/f,  where  c  is  the  speed  of  light  ( 3.108 m/s).  So  for  our  worst  case  of  the  100  Hz  signal,  we  
would  need  an  antenna   1.87.103 miles  long!  For  the  best  case,  the  required  length  would  still  be  37  miles.  This  
obviously  is  impractical.  

  The  second  difficulty  deals  with  separating  different  stations.  Suppose  that  antenna  efficiency  was  not  a  
problem.  Once  the  Federal  Communication  Commission  (FCC)  granted  a  license  to  one  radio  station  to  transmit  
between  100  Hz  and  5000  Hz  (The  frequency  of  5  KHz  corresponds  to  the  highest  note  with,  no  harmonics,  
output  from  the  highest  pitched  instrument  the  piccolo.),  the  entire  practical  audio  spectrum  would  be  used  up.  
For  example,  suppose  a  second  station  was  allowed  to  transmit  its  audio  signal.  Both  stations  would  be  
transmitting  signals  between  100  Hz  and  5000  Hz.  Receiving  antennas  could  not  differentiate  between  the  two  
and  would  receive  them  both.  The  result  is  that  the  sum  of  the  two  audio  signals  would  be  heard.  Of  course,  if  
one  signal  was  appreciably  stronger  than  the  other,  the  stronger  signal  would  be  heard  and  the  weaker  would  
only  act  as  interference  to  the  stronger.  But  suppose  it  was  the  weaker  signal  you  were  interested  in.  There  
would  be  no  reasonable  way  to  extract  it  from  the  garbled  sum  (Rush  Limbaugh  might  drown  out  some  good  
rock  music!)  

  Modulation  avoids  both  difficulties  because  a  carrier  frequency  represents  the  center  of  the  transmitted  
wave.  In  commercial  AM  radio  stations  this  is  around  1000  KHz  (1  MHz).  At  this  frequency  a  perfectly  matched  
antenna  is  984  feet  long.  This  is  still  quite  long,  but  we  can  reduce  the  antenna  to  one  500th  or  even  one  1000th  
of  its  ideal  size  and  make  up  the  loss  in  signal  strength  with  sufficient  gain  and  selectivity.  Furthermore  even  a  
bandwidth  of 2 f max ,  necessary  for  a  double  sideband  transmission,  is  only  10  KHz  wide  and  many  stations  can  
be  effectively  transmitted  side-­‐by-­‐side  by  simply  placing  their  respective  carriers  at  least  10  KHz  apart.  This  
process  is  called  frequency  division  multiplexing  (FDM).  A  single  channel  can  accommodate  multiple  users  if  the  
frequency  spectrum  is  divided  up.  In  fact,  the  FCC,  in  regards  to  the  commercial  AM  broadcast  band  which  
covers  the  electromagnetic  spectrum  from  535  KHz  to  1705  KHz,  allows  stations  to  have f max  up  to  5  KHz  and  
permits  stations  to  be  located  10  KHz  apart.  

45
4.2  Amplitude  Modulation  (AM)  
  Amplitude  modulation  is  a  form  of  continuous  wave  modulation  in  which  the  amplitude  of  a  sine  wave  
of  some  specified  frequency,  called  the  carrier,  is  varied  in  accordance  with  the  signal  containing  the  
information  which  may  be  voice  or  music.  Another  possibility  is  to  vary  the  frequency  of  the  carrier  in  
accordance  with  the  information  signal.  This  form  of  modulation  is  called  FM.  

  Amplitude  modulation  is  the  basis  of  much  of  our  commercial  and  amateur  broadcast  communications.  
To  understand  how  this  technique  operates,  consider  a  sinusoidal  signal  given  by  

vc (t ) = Vc cos(ωct ) (4.1)
 
                             This  is  called  the  carrier  wave  for  reasons  which  will  become  clear  shortly.   Vc  is  the  amplitude  of  the  
carrier  signal  and   ωc = 2π f c  is  the  angular  frequency  of  the  carrier  which  is  in  the  Radio  Frequency  (RF)  band.  
These  frequencies  are  much  higher  than  audio  frequencies  and  are  typically  in  the  1  MHz  and  above  range  such  
that   f c ? 20 KHz.  In  AM,  the  carrier  is  caused  to  carry  information  by  changing  the  amplitude  according  to  the  
information  signal.  More  specifically  we  shall  write  the  amplitude  of  the  carrier  as  follows:  

Vc (t ) = [Vc + kx(t )] (4.2)

  Where   Vc  is  the  original  carrier  amplitude,  and  x(t)  the  original  signal  containing  the  information  to  be  
transmitted,  i.e.  the  modulating  signal.  The  factor  k  is  for  scaling  or  amplification.  Then  the  resultant  modulated  
wave  can  be  written  as

vAM (t ) = [Vc + kx(t )]cos(ωct ) (4.3)

  Since  x(t)  represents  a  real  signal,  such  as  music,  it  can  be  represented  by  an  infinite  series  of  sines  and  
cosines,  called  Fourier  components.  A  Fourier  component  corresponds  to  a  pure  note  or  a  pure  tone.    It  would  
sound  like  the  signal  broadcast  by  the  emergency  broadcast  system  on  your  radio.  Let  

kx(t ) = kX m cos(ωmt ) = Vm cos(ωmt ) = vm (t ) (4.4)

  Where X m  is  the  original  amplitude  of  the  information  signal,  which  may  need  to  be  scaled  up  or  down  
by  the  factor  k  to  allow  us  to  utilize  it  effectively.  We  can  then  write:

vAM (t ) = [Vc + Vm cos(ωmt )]cos(ωct ) = Vc [1 + m cos(ωmt )]cos(ωct ) (4.5)

Vm
  The  ratio = m  is  called  the  modulation  index  and  indicates  the  extent  of  modulation.  The  time  
Vc
domain  representation  of  an  amplitude  modulated  signal  is  shown  in  Figure  0-­‐1.  

46
Figure 0-1: An Amplitude Modulated Signal.

  The  envelope  of  the  modulated  carrier  amplitude  represents  the  modulating  signal  and  varies  
symmetrically  about  the  carrier  amplitude, Vc .  The  maximum  value  of  the  envelope, Vmax ,  occurs  when  both  
cos(ωct ) and   cos(ωmt ) in  Equation  (4.5)  are  equal  to  1.  Therefore, Vmax = Vc + Vm .  The  minimum  value  of  the  
envelope, Vmin ,  occurs  when   cos(ωct ) = 1 but   cos(ωmt ) = −1 .  Therefore,   Vmin = Vc − Vm .  If   Vmax  is  added  to Vmin ,  
the  result  is 2Vc  and  if Vmin is  subtracted  from Vmax ,  the  result  is 2Vm .  Therefore,  the  carrier  amplitude  and  the  
modulating  amplitude  are  given  by  

Vmax + Vmin V − Vmin


Vc = and Vm = max (4.6)
2 2

Then  the  index  of  modulation  can  be  written  in  terms  of  these  maximum  and  minimum  amplitudes.  

Vmax − Vmin Vm
m= = (4.7)
Vmax + Vmin Vc

  Figure  4-­‐2  shows  the  effect  of  varying  the  value  of  the  modulation  index  on  the  modulated  waveform.  
The  modulation  index  controls  the  amount  or  the  intensity  of  the  modulation.  The  value  of  m  can  range  from  0  
to  almost  any  positive  value,  but  it  is  usually  selected  so  that  the  value  of  the  carrier  amplitude Vc (t )  is  never  
less  than  zero.  If   m > 1 ,  as  in  the  bottom  plot  of  Figure  2-­‐2,  recovery  of  the  information  signal  would  require  
very  complicated  forms  of  demodulation.  Therefore,  for  commercial  AM  broadcasting,  it  is  usually  required  that
m ≤ 1 .  

  The  degree  of  amplitude  modulation  can  be  expressed  as  a  percentage,  and  is  given  by  P  =  (m)(100%).  
When  m  is  equal  to  1,  we  have  100%  modulation.  Modulating  beyond  100%,  as  mentioned  above,  is  
undesirable.  Modulating  much  less  than  100%  simply  reduces  the  desirable  effect  of  modulation.  In  practice,  
systems  are  amplitude  modulated  between  50  and  100%.  

  It  is  useful  to  represent  the  modulation  process  by  a  block  diagram  as  follows  in  Figure  4-­‐3.  This  
diagram  makes  it  clear  that  amplitude  modulation  involves  a  multiplication  process  which  results  in  the  
creation  of  new  frequencies.  Specifically,  if  we  expand  Equation  (4.5)  using  the  following  trigonometric  identity,  

47
1 1
cos A cos B = cos( A + B) + cos( A − B) (4.8)
2 2
we  get:  

mVc
vAM (t ) = Vc cos(ωct ) + [cos(ωc + ωm )t + cos(ωc − ωm )t ] (4.9)
2

  We  see  that  the  resultant  amplitude  modulated  signal  is  composed  of  the  sum  of  three  sinusoidal  
functions,  having  the  frequencies, f c ,   ( f c + f m ) ,  and   ( f c − f m ) .  The  effect  of  the  modulation  process  can  best  be  
seen  by  sketching  the  spectrum  of  the  AM  waveform  of  Equation  (4.9).  This  is  done  in  Figure  4-­‐3.  

  The  two  new  frequencies  created  in  this  process,   ( f c + f m ) and ( f c − f m ) ,  are  called  the  sidebands  of  the  
signal,  and  it  is  they,  not  the  carrier  frequency, f c ,  that  contain  the  information.  This  particular  type  of  AM  
modulation  is  called  Double  Side-­‐Band,  with  Large  Carrier  (DSB-­‐LC)  because  both  sidebands  and  the  carrier  
show  up  in  the  spectrum  as  seen  in  Figure  4-­‐4.  If  the  carrier  is  eliminated,  then  only  the  two  side  bands  show  
up.  This  form  is  called  Double  Side-­‐Band  Suppressed  Carrier  (DSB-­‐SC)  and  if  one  side  band  and  the  carrier  are  
eliminated  then  only  one  of  the  two  side  bands  shows  up  in  the  spectrum  which  results  in  Single  Side  Band  
Suppressed  Carrier  (SSB-­‐SC)  AM.  More  will  be  said  about  SSB  and  its  advantages  and  disadvantages  below.  

  Up  until  now,  we  have  dealt  only  with  the  modulation  of  the  carrier  by  a  single  Fourier  component  (a  
pure  tone).  Now  consider  what  happens  when  we  modulate  (we  are  assuming  DSB  with  Large  Carrier)  a  carrier  
by  an  entire  signal  of  many  frequencies.  This  is  the  more  common  case  and  is  true  for  voice,  music  and  data.  No  
one  speaks  the  English  language  using  a  pure  tone  voice.  Let’s  call  this  signal   vm (t )  and  see  how  the  sidebands  
carry  all  of  the  information.  Suppose   vm (t ) has  the  Fourier  spectrum  shown  in  Figure  4-­‐5a.  Its  highest  frequency  
is  indicated  as f max .  When  we  modulate  a  carrier  wave  of  frequency   f c with vm (t ) ,  each  Fourier  component  is  
multiplied  with  the  carrier.  The  result  is  shown  in  Figure  4-­‐5b.  

48
Figure 0-2: Effect of Varying the Modulation Index.
 

Figure 0-3: Block Diagram of Amplitude Modulation Scheme.

49
Figure 0-4: Spectrum of a Carrier Modulated by a Single Fourier Component.

Figure 0-5: Spectrum of Amplitude DSBWC Modulated Signal.

  Each  and  every  line  in  the  spectrum  shown  in  Figure  4-­‐5a  forms  two  new  frequencies  when  modulated  
onto  the  carrier.  One  line  takes  a  position  in  the  upper  side  band  and  the  other  in  the  lower  sideband  of  Figure  
4-­‐5b.  Those  frequencies  from  the  carrier  to   ( f c − f max ) are  called  the  Lower  Side-­‐Band  (LSB)  of  the  signal,  and  
from  the  carrier  to ( f c + f max ) ,  the  Upper  Side-­‐Band  (USB)  of  the  signal.  

  As  
vm (t ) changes  with  time  (necessary  if  it  is  to  carry  information),  then  the  changes  would  be  reflected  

by  different  spectra   m
v (f) .  This  in  turn  would  show  up  in  the  sidebands.  If  you  view  the  spectrum  of  a  music  
signal  in  real  time,  it  will  keep  dancing  around  and  changing.  If  the  bass  control  were  suddenly  turned  up,  the  
lower  frequencies  would  grow  in  strength  and  those  lines  in  the  spectrum  at  the  lower  frequencies  would  get  
longer.  In  actuality,  there  would  be  so  many  frequencies  present  in  most  voice  signals  that  the  spectrum  of  
Figure  4-­‐5a  could  appear  to  be  nearly  continuous.  The  individual  lines  are  shown  in  Figure  4-­‐5a  to  emphasize  
the  creation  of  two  lines  in  the  DSB-­‐LC  AM  spectrum  for  every  one  line  in  the  modulating  signal.  It  should  be  
obvious  that  the  information  contained  in  a  modulated  signal  is  contained  in  the  sidebands.  Furthermore,  the  
upper  and  lower  sidebands  carry  redundant  information.  If  one  side  band  could  be  extracted  at  the  receiver,  the  
transmitted  information  could  be  recovered.  This  process  is  used  more  and  more  in  modern  communications  
and  will  be  mentioned  below.  If  a  single  side  band  modulation  process  is  used  then  the  spectrum  produced  will  
contain  only  one  of  the  two  sidebands.  The  resulting  spectrum  is  shown  below  in  Figure  4-­‐6  assuming  that  the  
upper  side  band  is  transmitted.  

50
Figure 0-6: Spectrum of Amplitude SSB Modulated Signal.

Example 4.1

An  amplitude  modulated  carrier  is  shown  below.  The  form  of  modulation  is  double  side  band  with  large  carrier.  
The  modulating  signal  is  a  pure  tone.  

(a)  Determine  the  carrier  frequency,  


(b)  Determine  the  frequency  of  the  modulating  signal,  
(c)  Determine  the  modulation  index,  
(d)  Write  the  equation  for  the  signal  shown  below,  
(e)  If  the  modulating  signal  is  recovered  from  the  carrier  and  applied  to  the  speaker,  what  would  be  
audible?  
 

Solution

(a)  If  we  count  the  number  of  cycles  of  the  carrier  starting  at  the  peak  just  to  the  right  of  2  ms  and  
ending  just  to  the  right  of  11  ms,  we  get  18  cycles  occurring  in  9  ms.  The  ratio  is  2  cycles/ms  which  
corresponds  to  2  MHz.  This  result  can  be  checked  by  counting  over  a  different  time  frame.  

(b)  If  we  start  to  the  right  of  2  ms,  one  period  of  the  envelope  is  completed  at  just  to  the  right  of  12  ms  
which  gives  a  total  of  10  ms  for  one  period.  Calculating  this  ratio  gives  100  KHz.  

51
(c)  The  maximum  amplitude  is  estimated  from  the  graph  to  be  about  17  V  and  the  minimum  amplitude  
17 − 3
to  be  3  V,  thus   m = = 0.7  
17 + 3

(d)  The  average  of  the  maximum  and  minimum  amplitudes  will  give  the  amplitude  of  the  unmodulated  
17 + 3
carrier.  Thus,   Vc = = 10  V.  Plugging  this  along  with  the  other  parameters  determined  above  into  
2
Equation  (2.5),  we  get:  

vAM (t ) = 10[1 + 0.7 cos(2π 105 t )]cos(2π 2 ×106 t )

(e)  The  modulating  signal  is  a  pure  tone  of  frequency  100  KHz,  which  is  in  the  ultrasonic  range.  Thus  it  
would  not  be  heard  by  human  ears,  which  can  hear  frequencies  up  to  20  KHz.    

Example 4.2

The  carrier  in  an  AM  signal  (DSB-­‐LC,  double  side  band-­‐  large  carrier)  crosses  zero  every  1  ms.  The  modulating  
signal  is  a  pure  tone  and  the  time  between  a  maximum  in  the  amplitude  and  the  very  next  minimum  is  1ms.  The  
maximum  amplitude  is  10  V  and  the  minimum  amplitude  is  6  V.  

(a)  Determine  the  frequency  of  the  carrier,  


(b)  Determine  the  frequency  of  the  pure  tone  modulating  signal,    
(c)  Determine  the  modulation  index,  
(d)  If  the  modulation  index  were  turned  down  to  zero  such  that  the  amplitude  of  the  carrier  no  longer  
changed  what  would  the  resulting  amplitude  be  for  this  signal?  

Solution

(a)  The  time  between  successive  zero  crosses  is  half  a  period.  Thus,  the  period  of  the  carrier  is  2  ms.    
The  reciprocal  of  the  period  is  the  frequency  which  is  0.5  MHz  or  500  KHz.  

(b)  The  time  from  a  peak  on  the  envelope  to  the  next  valley  is  one  half  the  period  of  the  modulating  
signal.  Thus,  the  period  of  the  modulating  signal  is  2  ms,  which  has  a  reciprocal  of  0.5  KHz  or  500  Hz.  
Note  that  this  tone  would  be  audible  if  applied  to  a  speaker.  

Vmax − Vmin 10 − 6
(c)  We  apply  Equation  (2.7)  and  get   m = = = 0.25  
Vmax + Vmin 10 + 6

(d)  With  no  modulation,  the  amplitude  will  be  that  of  the  carrier  alone.  Using  Equation  (2.6)  we  get  
1
Vc = (Vmax + Vmin ) = 8  V.  Note  that  this  is  halfway  between  the  maximum  and  minimum  amplitudes.  
2

52
4.3  AM  Bandwidth  
  When  modulating  signals  it  is  important  to  know  the  bandwidth  of  the  information  signal.  Generally,  the  
wider  this  bandwidth  (more  frequencies),  the  more  information  can  be  carried.  However,  the  wider  the  
bandwidth  the  more  costly  (dollars,  spectrum  usage,  design  difficulty,  etc.)  the  system  necessary  to  process  the  
signal.  The  bandwidth  of  each  sideband  is  given  by:  

BWUSB = ( f c + f max ) − f c = f max (4.10)


BWLSB = f c − ( f c − f max ) = f max (4.11)
And  for  both  sidebands  together  

BW = ( fc + f max ) − ( fc − f max ) = 2 f max (4.12)

  When  both  sidebands  are  transmitted  along  with  the  carrier,  it  is  called  Double  SideBand-­‐Large  Carrier  
(DSB-­‐LC)  transmission.  Since  both  sidebands  contain  the  information,  it  is  possible  to  eliminate  one  of  the  
sidebands  and  transmit  a  Single  SideBand  -­‐  Large  Carrier  (SSB-­‐  LC).  In  some  amplitude  modulation  systems  the  
carrier  is  eliminated  prior  to  transmission.  This  is  called  suppressed  carrier  (SC)  transmission.  We  can  send  
DSB-­‐SC  as  well  as  SSB-­‐SC.  SSB-­‐SC  systems  are  especially  common  in  amateur  and  citizens  band  systems  where  
either  Upper  SideBand  (USB)  or  Lower  SideBand  (LSB)  can  be  sent.  A  disadvantage  of  suppressed  carrier  
systems  is  that  the  demodulation  process  becomes  more  complicated  and  expensive.  This  disadvantage  was  
more  severe  50  years  ago  than  it  is  today  and  it  is  the  reason  that  commercial  AM  uses  the  DSB-­‐LC  process  
which  allows  the  use  of  simpler  and  less  expensive  receivers.  An  important  advantage  of  not  transmitting  the  
entire  modulated  spectrum  is  the  conservation  of  power.  That  is,  no  energy  is  spent  transmitting  the  carrier  or  a  
redundant  sideband.  Another  important  advantage  of  SSB  transmission  is  the  conservation  of  bandwidth  
allowing  more  channels  and  more  users  for  a  given  range  of  frequency.  

Example 4.3: The spectrum of an AM signal is shown below.

(a)  What  type  of  amplitude  modulation  does  this  represent?  


(b)  What  is  the  value  of  the  carrier  frequency?    
(c)  Determine  the  modulating  frequency.  
(d)  Determine  the  modulation  index.  
(e)  How  much  radio  frequency  bandwidth  is  required  to  transmit  this  signal?    
 
 

53
Solution

(a)  There  is  a  carrier  component  and  two  side  band  components  so  this  is  DSB-­‐LC.  Since  the  upper  and  
lower  sidebands  each  consist  of  only  one  frequency  component  each,  the  modulating  signal  is  a  pure  
tone.  The  spectrum  of  a  narrow  side  band  FM  signal  would  also  look  the  same.  We  would  need  the  phase  
spectrum  to  distinguish  between  them.  

(b)  The  carrier  component  is  the  one  in  the  middle  and  its  frequency  is  3  MHz.  

(c)  The  modulating  frequency  is  the  difference  between  the  upper  sideband  frequency  and  the  carrier  or  
the  difference  between  the  carrier  and  the  lower  sideband  frequency.  Either  calculation  gives  0.005  MHz  
=  5  KHz.  

(d)  The  amplitude  of  each  sideband  is  0.5mVc.  Thus  3  V  =  0.5(m)(12)  V.  Solving  for  m,  gives  m  =  0.5.  A  
common  student  error  is  to  incorrectly  apply  Equation  (2.11).  Note  that  12  V  and  3  V  are  not  the  
maximum  and  minimum  values  of  the  envelope.  The  amplitudes  in  a  spectrum  plot  do  not  directly  give  
the  amplitudes  in  the  time  domain  plot.  

(e)  To  transmit  this  signal  the  whole  range  from  2.995  MHz  up  to  3.005  MHz  must  be  included.  Thus  the  
RFBW  =  3.005  -­‐  2.995  MHz  =  0.01  MHz  =  10  KHz.  

Example 4.4

A  given  audio  baseband  signal  contains  frequency  components  from  50  Hz  up  to  6  KHz  and  is  to  be  amplitude  
modulated  onto  a  10  MHz  RF  (radio  frequency)  carrier.  

(a)  How  much  RF  bandwidth  will  be  required  for  DSB-­‐LC  modulation?  
(b)  How  much  RF  bandwidth  will  be  required  for  SSB-­‐SC  modulation  (assume  USB)?  

Solution

(a)  Both  the  upper  and  lower  sidebands  are  included  in  DSB-­‐LC.  This  range  stretches  from    (10  MHz  -­‐  6  
KHz)  up  to  (10  MHz  +  6  KHz)  for  a  total  of  12  KHz.  

(b)  SSB  transmission  economizes  on  bandwidth  and  will  require  only  6  KHz  -­‐  50  Hz  which  is  practically  
equal  to  6  KHz.  

4.4  AM  Power  


  The  power  necessary  to  transmit  an  AM  signal  can  be  found  in  the  usual  manner  using  AC  circuit  theory.  
The  power  delivered  to  a  resistor  by  a  single  sinusoidal  component  is  given  by  one  half  the  square  of  the  
amplitude  (peak  value)  divided  by  the  resistance  value.  The  square  of  the  RMS  value  can  be  used  instead  of  half  
the  square  of  the  amplitude.  If  multiple  sinusoidal  components  of  different  frequencies  are  applied  to  a  resistor,  
then  superposition  applies  and  the  total  power  is  simply  the  sum  of  the  individual  powers.  Thus  for  DSB-­‐LC  AM  
signals  we  have  

Ptotal = Pc + PLSB + PUSB (4.13)

54
  An  ideal  antenna  has  a  radiation  resistance  R,  though  an  antenna  transforms  the  electrical  power  into  
electromagnetic  radiation  instead  of  heat.  If  we  assume  a  pure  tone  modulation,  then  the  AM  power  transmitted  
by  an  antenna  is:  

Vc2 m2Vc2 m2Vc2 m2 m2 m2 ⎛ m2 ⎞


Ptotal = + + = Pc + Pc + Pc = Pc + Pc = ⎜1 + ⎟ Pc (4.14)
2 R 4(2 R) 4(2 R) 4 4 2 ⎝ 2 ⎠

  For  m  =  1,  the  modulation  is  maximum  and  the  amplitude  of  the  sidebands  is  maximum.  Under  this  
condition,  the  power  transmitted  in  the  sidebands  is  maximum,  and  the  total  power  is  

3
Ptotal = Pc (4.15)
2

The  fraction  of  the  total  power  sent  in  the  carrier  and  in  each  side  band  is:  

1
P
Pc 2 PLSB PUSB 4 c 1
= = 0.67 = 67% and = = = = 0.167 = 16.7% (4.16)
Ptotal 3 Ptotal Ptotal 3 P 6
c
2
  So  even  under  the  best  conditions  of  100%  modulation,  the  power  transmitted  in  each  sideband  is  only  
1/6  of  the  total  power.  The  2/3  of  the  total  power  used  to  transmit  the  carrier  is  a  waste.  Since  the  distance  over  
which  communications  can  be  established  is  a  function  of  the  power  in  the    sideband,  communication  over  the  
same  distance  can  be  accomplished  with  SSB-­‐SC  as  with  DSB-­‐LC  but  with  1/6  the  power.  Power  can  be  
conserved  by  suppressing  the  carrier  and  sometimes  one  of  the  sidebands.  This  also  means  that  SSB-­‐SC  can  be  
transmitted  over  longer  distances  than  can  DSB-­‐LC  for  a  given  transmitter  power.  This  makes  SSB-­‐SC  attractive  
for  portable  transmitters.  

Example 2.5: If  the  following  AM  signal  is  applied  to  an  antenna  having  a  radiation  resistance  of  50  W  find  the  
power  in  the  carrier  and  in  each  side  band  and  then  the  ratio  of  the  power  in  the  information  part  of  the  signal  
to  the  total.  

v AM (t ) = 20[1 + 0.8cos(1000t )]cos(107 t ) V

Solution

The  amplitude  of  the  carrier  is  20  V  and  the  sideband  amplitudes  are  each  0.5(0.8)(20)  =  8  V.  Thus,  
0.5(20)2 0.5(8)2
Pc = = 4 W  and   PUSB = PLSB = = 0.64 W.  The  sidebands  are  the  information  part  of  the  signal  
50 50
and  have  a  total  power  of  2(.64)  =  1.28  W  while  the  total  power  is  4  +  1.28  =  5.28  W.  This  leads  to  a  ratio  of  
1.28/5.28  =  0.2424  or  24.24%.  

55
4.5  Frequency  Division  Multiplexing  (FDM)  
  As  was  discussed  in  section  1.3  one  of  the  primary  reasons  to  modulate  low  frequency  information  onto  
a  high  frequency  carrier  is  that  multiple  channels  are  available.  Many  different  carrier  frequencies  can  be  used  
to  carry  many  different  baseband  information  signals  simultaneously  through  the  same  transmission  medium  
be  it  free  space  or  a  coaxial  cable.  The  AM  or  FM  tuner  has  the  capability  of  tuning  to  many  different  radio  
stations.  In  order  to  be  able  to  separate  different  channels,  they  are  arranged  such  that  they  do  not  overlap.  
They  will  certainly  overlap  in  the  time  domain.  The  signals  from  many  different  radio  stations  will  be  present  on  
an  antenna  at  anyone  time  but  they  do  not  overlap  in  frequency.  None  of  the  frequency  components  from  one  
station  are  normally  allowed  to  overlap  and  interfere  with  those  from  another.  This  separation  in  frequency  
allows  a  receiver  to  use  a  band-­‐pass  filter  to  sort  through  and  select  just  one  station.  Occasionally,  when  
interference  occurs  it  is  because  of  overlap  of  frequency  components  from  different  stations.  Perhaps  a  CB  is  
broadcasting  on  an  incorrect  carrier.  It  is  possible  to  multiplex  several  channels  onto  one  transmitting  antenna.  
This  is  shown  in  Figure  2-­‐6.  Here,  three  different  baseband  information  signals x1 (t ) ,   x2 (t )  and   x3 (t )  with  
corresponding  magnitude  spectra   x1 ( f ) ,   x2 ( f ) and   x3 ( f ) are  AM  modulated  onto  three  different  carriers   f1 ,  
f 2  and   f3 .  The  composite  spectrum  is  also  shown  which  includes  some  guard-­‐band  between  each  channel.  The  
use  of  a  guard-­‐band  makes  separation  of  channels  at  the  receiver  easier  since  band-­‐pass  filters  are  not  perfect.  
The  shapes  of  each  spectrum  in  Figure  2-­‐6  are  not  meant  to  convey  anything  in  particular  about  each  signal.  
They  are  simply  place  holders  for  a  range  of  frequencies  and  they  do  convey  that  each  is  band  limited.  That  is,  
the  frequency  content  of  each  is  limited  in  range  up  to f max .  

Example 4.6

How  many  channels  can  be  frequency  division  multiplexed  between  20  MHz  and  22  MHz  using  DSB-­‐LC  if
f max = 10  KHz  for  each  channel  and  a  2  KHz  guard-­‐band  is  to  be  maintained  between  the  highest  frequency  of  
any  channel  and  the  lowest  frequency  of  the  next  higher  channel.  Assume  that  the  frequency  content  of  none  of  
the  channels  is  allowed  to  be  less  than  20  MHz  or  greater  than  22  MHz.  Repeat  if  SSB  is  used  instead.  

Solution

Starting  at  20  MHz  for  DSB  each  channel  plus  associated  guard-­‐band  requires   2 f max + guardband = 22  KHz  of  
bandwidth.  If  we  divide  2  MHz  by  22  KHz,  we  get  90.9,  and  so  the  answer  is  90  channels.  The  upper  frequencies  
of  the   91st  channel  would  go  past  22  MHz.  If  SSB  is  used  instead  of  DSB-­‐LC  then  we  get  2000  divided  by  12  
which  gives166.6  and  so  the  answer  is  166  channels.  

56
Figure 0-7: Composite Spectrum.

57
4.6  Homework  Problems    

Problem 4.1: An  AM  signal  has  a  frequency  spectrum  shown  below

a. What  is  the  bandwidth  of  the  audio  signal  which  was  modulated  onto  the  carrier?  
b.  What  is  the  RF  bandwidth  of  the  signal  being  transmitted?  
c. What  type  of  AM  modulation  produces  the  signal?  

Problem  4.2:  For  the  signal  shown  below,  determine:  

a. The  type  of  modulation  used  to  produce  the  signal  


b. The  modulation  index  
c. The  carrier  frequency  
d. The  modulating  frequency  
e. The  amplitude  of  the  carrier  if  the  modulating  signal  were  reduced  to  zero  
f. The  amplitude  of  the  USB  and  LSB  components  
g. Write  the  equation  for  the  above  waveform  
h. Plot  the  amplitude  spectrum  of  this  AM  signal.  

58
Problem 4.3

A  235  KHz  carrier  is  amplitude  modulated  (DSB-­‐LC)  by  a  5  KHz  pure  tone.  The  un-­‐modulated  amplitude  of  the  
output  was  250VRMS  and  the  modulation  index  is  80%.  

a. What  are  the  lower  and  upper  sideband  frequencies?  


b. What  is  the  amplitude  of  the  upper  sideband  component  in  the  frequency  spectrum?  
c.  If  this  AM  signal  is  applied  to  an  antenna  which  has  a  radiation  resistance  of  80  W,  determine  the  total  
power  in  the  two  side  bands  together  and  the  power  in  the  carrier.  
d. How  much  radio  frequency  bandwidth  does  this  AM  signal  occupy?  

Problem  4.4:  An  amplitude  modulated  waveform  is  given  by  the  equation:    

  v AM (t ) = 100 cos(1.130 ×106 t ) + 45cos(1.0986 ×10 6 t ) + 45cos(1.16 × 106 t )  V  

a. What  is  the  percentage  of  modulation?  


b. For  an  RF  filter  to  pass  the  waveform,  what  must  its  minimum  bandwidth  be?  
c. Sketch  the  envelope  of  the  waveform.  

Problem 4.5

Assuming  AM  DSB-­‐LC  modulation,  determine  the  percentage  modulation  for  each  of  the  following  conditions.  
The  same  carrier  is  used  for  each  condition.  Determine  the  peak  value  of  that  carrier.  

Maximum Output Signal Peak-Peak Minimum Output Signal Peak-Peak


Voltage Voltage

a. 100 60

b. 125 35

c. 160 0

Problem 4.6

A  signal   v AM (t ) = 40{1 + 0.7 cos(2π × 500t ) + 0.5cos(2π × 800t )}cos(2π ×10 6 t ) V  contains  not  one  but  two  
different  modulating  frequencies.  Thus,  there  are  two  different  indices  of  modulation.  

a. Determine  each  index  of  modulation  and  the  corresponding  frequency  that  it  goes  with.  
b.  Plot  the  magnitude  spectrum  and  find  the  RF  bandwidth.  
c. Use  a  computer  plotting  program  to  sketch  the  envelope.    

59
Problem 4.7 Given  the  spectrum  of  an  AM  signal  shown  below:

a. Determine  the  frequency  of  the  modulating  signal.  


b. Determine  the  RF  bandwidth  required  to  transmit  this  signal.  
c. Write  the  equation  for  this  AM  signal.  
d. If  this  AM  signal  is  applied  to  an  antenna  that  has  a  radiation  resistance  of  200  W  find  the  fraction  of  the  
total  power  in  the  information  part  of  the  signal.  

Problem 4.8

A  particular  baseband  audio  signal  has  frequency  components  from  50  Hz  to  8  KHz.  Determine  the  range  of  
frequencies  present  and  the  RF  bandwidth  for  

a. DSB-­‐LC  
b. SSB  (Upper  sideband)  
c. DSB-­‐SC  if  the  baseband  signal  is  amplitude  modulated  onto  a  2  MHz  carrier.  

Problem 4.9

When  a  certain  DSB-­‐LC  AM  signal  (pure  tone  modulating  signal)  is  applied  to  an  antenna,    
400  W  are  transmitted  at  the  carrier  frequency  and  80  W  in  each  of  the  two  sidebands.    
 
a. Determine  the  modulation  index.  
b. If  the  amplitude  of  the  un-­‐modulated  carrier  remains  the  same  but  the  modulation  index  is  changed  to  
0.6  for  the  same  antenna,  find  the  power  in  the  carrier  and  the  two  sidebands.  

Problem 4.l 0

Given  the  total  transmitted  power  in  a  DSB-­‐LC  AM  wave  is  3  KW  when  the  modulation  index  is  0.7,  how  much  
total  power  should  a  SSB-­‐SC  wave  contain  in  order  to  have  the  same  power  content  as  that  contained  in  the  two  
sidebands  together  of  the  DSB-­‐LC  wave?  

 
60
Problem 4.11

Four  different  messages  are  to  be  transmitted  simultaneously  by  the  same  antenna  by  using  frequency  division  
multiplexing  with  four  different  carriers.  Each  message  has  a  frequency  range  from  100  Hz  to  3  KHz  and  a  
guard-­‐band  of  2  KHz  is  to  be  inserted  between  adjacent  channels  to  help  prevent  interference  and  make  
demultiplexing  easier.  If  the  lowest  frequency  carrier  is  1  MHz,  sketch  the  total  magnitude  spectrum  
transmitted  by  the  antenna  and  find  the  total  RF  bandwidth  for  the  Transmitter/Antenna.  Assume  DSB-­‐  LC  
modulation  for  each  channel.  

Problem 4.12

Frequency  division  multiplexing  is  used  in  the  communications  system  shown  below  in  block  diagram  form.  

a. Determine  the  highest  frequency  present  at  point  A.  


b. Determine  the  lowest  frequency  present  at  point  B.  
c. Find  the  range  of  frequencies  present  at  points  C  and  D.  
d. This  system  provides  how  much  guard-­‐band  between  the  audio  and  instrumentation  channels?    
e. Sketch  the  frequency  spectrum  at  the  output  of  the  transmitter,  using  nominal  amplitudes.    
f. What  is  the  minimum  bandwidth  required  for  the  amplifiers  in  the  transmitter?  

61
Chapter  5:    AM  Demodulation  
 

5.1  Introduction  
  In  this  section,  AM  demodulation  will  be  discussed  followed  by  the  description  of  a  superheterodyne  
receiver  in  the  next  section.  The  demodulator  (sometimes  called  a  detector)  is  a  subsystem  within  a  receiver.  A  
receiver  contains  other  subsystems  such  as  a  mixer,  RF  (radio  frequency),  and  AF  (audio  frequency)  amplifiers.  
The  process  of  demodulation  extracts  and  retrieves  the  information  signal  from  the  modulated  carrier.  The  
objective  of  demodulation  is  to  undo  modulation  and  wind  up  with  the  transmitted  information.  Two  different  
types  of  detectors  will  be  examined,  first  the  synchronous  detector  shown  in  block  diagram  form  in  Figure3-­‐.1  
and  then  the  envelope  detector  which  is  useful  for  DSB-­‐LC  signals.

5.2  Synchronous  Demodulation    


  The  synchronous  demodulator  is  very  useful  for  the  detection  of  suppressed  carrier  signals  but  can  also  
be  used  for  DSB-­‐LC  signals.  

Figure 0-1: Block Diagram of a Synchronous Demodulator.

  We  will  not  study  any  particular  circuits  which  implement  the  multiplication  or  mixing  process  
represented  in  Figure  5-­‐1  by  the  multiplication  symbol.  Instead,  our  objective  is  to  understand  the  concept  and  
mathematics  of  the  multiplication  process.  Assume  that  v(t)  is  the  DSB-­‐LC  AM  signal  shown  in  the  equation  
directly  below  and  further  assume  that  the  modulating  signal  is  a  pure  tone.  

vAM (t ) = [Vc + vm (t )]cos(ωct ) (5.1)

  If  the  input  signal, vAM (t ) ,  is  multiplied  by  a  sinusoid  of  exactly  the  same  frequency  and  phase  as  the  
carrier,  the  original  tone  signal  can  be  recovered.  The  signal  y(t)  is  the  product  of   vAM (t ) and   A cos(ωct ) which  
becomes:  

y (t ) = v AM (t ) A cos(ωct ) = Vc [1 + m cos(ωmt )]cos(ωct ) A cos(ωct ) = A cos 2 (ωct )Vc [1 + m cos(ωmt )] (5.2)

Applying  a  trigonometric  identity  for  double  angles  to  the  cosine  squared  term:  
63
⎡ 1 1 ⎤
y (t ) = A ⎢ + cos(2ωct ) ⎥ Vc [1 + m cos(ωmt )] (5.3)
⎣ 2 2 ⎦
 

AVc AVc mAVc mAVc


y (t ) = + cos(2ωct ) + cos(ωmt ) + cos(2ωct ) cos(ωmt ) (5.4)
2 2 2 2

Now,  the  last  term  can  be  expanded  (remember:  Sum  and  Difference  Frequencies)  and  y(t)  becomes:  

AVc AVc mAVc mAVc mAVc


y (t ) = + cos(2ωct ) + cos(ωmt ) + cos(2ωc − ωm )t + cos(2ωc + ωm )t (5.5)
2 2 2 4 4

  The  first  term  of  the  mixer  output  is  a  DC  component  and  is  not  part  of  the  original  signal  which  was  
modulated  onto  the  carrier.  The  second,  fourth  and  fifth  terms  represent  an  AM  signal  centered  at  twice  the  
original  carrier  and  are  unwanted.  The  third  term  is  the  original  signal  to  be  retrieved.  A  plot  of  the  spectrum  
suggests  how  the   cos(ωmt ) part  can  be  selected.  The  spectrum  of  our  synchronously  detected  signal  is  plotted  in  
Figure  5-­‐2.  

Figure 0-2: Spectrum of Synchronously Demodulated Signal.


 

  The  pure  tone  signal  is  located  at f m .  A  band  pass  filter  which  passes  all  frequencies  up  through   f m  but  
blocks  the  DC  component  can  be  used  to  recover  the  desired  signal.  The  output  of  the  detector,  z(t),  will  be  
proportional  to  the  pure  tone  modulating  signal  placed  on  the  carrier  at  the  transmitter.  Actually,  the  BPF  does  
not  need  to  be  very  narrow.  It  can  be  a  low  pass  filter  followed  by  a  blocking  capacitor  to  remove  the  DC  
component.  This  filter  shape  for  this  low  pass,  DC  block  combination  would  extend  almost  all  the  way  down  to  
zero  frequency  as  indeed  it  must  for  audio  signals  because  the  audio  range  extends  from  about  20  KHz  down  to  
about  50  Hz.  

64
  The  biggest  difficulty  with  synchronous  detection  is  the  exact  reproduction  of  the  carrier  at  the  location  
of  the  receiver.  This  method  will  not  tolerate  any  error  in  phase  or  frequency  of  the  reproduced  carrier.  This  
used  to  be  a  bigger  problem  than  it  is  now  with  today’s  modern  integrated  circuitry.  

  Envelope  detection  can  be  used  for  DSB-­‐LC  AM  but  not  for  suppressed  carrier  transmission.  Envelope  
detectors  (to  be  discussed  below)  are  very  inexpensive  and  easy  to  build.  This  is  one  reason  that  DSB-­‐LC  was  
chosen  for  the  early  commercial  AM  radio  system.  The  home  radio  receiver  had  to  be  practical  and  inexpensive.  
The  DSB-­‐LC  commercial  AM  system  persists  to  this  day.  If  a  new  commercial  AM  system  were  to  be  built  from  
the  ground  up  today,  single  side  band  signals  would  be  used.  As  stated  before,  the  simple  and  inexpensive  
envelope  detector  cannot  be  used  for  SSB  transmission,  but  now,  high  quality  synchronous  detectors  are  cheap  
and  easy  to  build  thanks  to  more  stable  oscillators  and  phase  locked  loops.  Most  modern  AM  communications,  
such  as  citizen’s  band,  is  by  SSB  transmission  and  uses  synchronous  detection.  Let  us  consider  what  happens  if  
v(t)  in  Figure3-­‐.2  is  a  single  side  band  signal.  We  will  assume  USB  and  pure  tone  modulation.  Then,  for
v(t ) = VUSB cos(ωc + ωm )t ,  y(t)  becomes:

y(t ) = VUSB [cos(ωc + ωm )t ] A cos(ωct ) (5.6)

And  which  contains  two  frequency  components,  the  sum  and  difference  frequencies  at   (2ωc + ωm ) and ω m .  
Thus,  y(t)  can  be  rewritten  as:  

AVUSB AV
y (t ) = cos(2ωc + ωm )t + USB cos(ωmt ) (5.7)
2 2

  The  first  component  is  at  a  high  frequency  and  is  unwanted.  The  second  is  proportional  to  the  original  
modulating  tone  and  is  the  part  we  want.  A  simple  low  pass  filter  will  reject  the  high  frequency  component  and  
retrieve  the  information  signal.  Synchronous  detection  will  also  work  for  LSB  or  DSB  transmission  when  the  
carrier  is  suppressed.  Of  course  the  same  synchronization  problems  are  present  as  are  for  synchronous  
detection  of  DSB-­‐LC  but  there  are  several  benefits  to  be  had  with  SSB-­‐SC  transmission.  First,  less  bandwidth  is  
used  per  channel  and  more  channels  can  be  frequency  division  multiplexed  together  than  can  be  for  DSB-­‐LC.  
Also,  no  power  is  wasted  in  transmission  of  the  carrier  and  more  range  can  be  achieved  for  a  transmitter  of  the  
same  size  and  power  as  a  DSB-­‐LC  transmitter.  These  factors  are  important  in  portable  and  personal  
communications  systems  such  as  citizen’s  band  radio  (CB).  

5.3  Envelope  Detection    


  While  synchronous  demodulation  is  easy  to  understand  mathematically,  in  practice  it  can  be  difficult  to  
achieve.  An  oscillator  in  the  receiver  must  exactly  reproduce  the  sinusoidal  carrier.  Building  a  stable  oscillator  
which  is  precisely  tunable  to  the  carrier  frequency  over  a  wide  range  of  possible  carriers  can  be  difficult.  The  
receiver  oscillator  must  also  be  in  precise  phase  alignment  with  the  carrier  or  else  significant  loss  of  detected  
power  will  occur.  Stability  of  the  oscillator  is  required  to  avoid  a  station  fading  out  over  time.  Nevertheless,  
many  modern  communications  systems  do  use  some  form  of  synchronous  detection  in  their  demodulation  
schemes.  

65
  A  common  demodulation  technique  which  is  used  for  the  demodulation  of  commercial  AM  is  envelope  
detection.  This  is  the  simplest  and  most  economical  technique  for  detecting  DSB-­‐LC  amplitude  modulated  
waves.  An  envelope  detector  generally  consists  of  a  diode  detector  and  an  RC  low-­‐pass  filter  as  shown  in  Figure  
5-­‐3.  

  In  the  discussion  to  follow  the  amplitude  of  x(t)  is  assumed  to  be  much  larger  than  the  forward  turn  on  
voltage  of  the  diode.  We  will  assume  that  the  diode  is  essentially  ideal.  This  does  bring  up  a  practical  point,  
however.  If  the  signal  directly  from  the  antenna  were  used  as  input  without  any  amplification,  it  would  
generally  be  too  weak  for  the  envelope  detector  to  work  properly.

Figure 0-3:Envelope Detector Circuit.

  The  envelope  detector  works  as  follows.  During  a  positive  half  cycle  of  the  input  x(t),  the  capacitor  
charges  up  to  the  peak  value  of  the  carrier  at  that  time.  As  the  input  signal  falls  below  this  value,  the  diode  
becomes  reversed  biased  ( vc (t ) > input)  and  turns  off.  Until  the  next  positive  peak  of  x(t)  comes  along,  the  
capacitor  decays  exponentially  through  the  resistor.  If  the  RC  time  constant  is  large  enough,  not  much  decay  in  
capacitor  voltage  occurs  before  the  next  positive  peak  of  x(t).  At  some  point  in  time  near  the  next  positive  peak  
the  diode  again  becomes  forward  biased  and  the  capacitor  charges  up  to  the  new  peak  value.  For  the  right  RC  
time  constant  the  result  is  a  capacitor  voltage  which  basically  connects  the  peaks  of  the  input  and  therefore  
yields  the  upper  envelope  of  the  DSB-­‐LC  AM  signal.  The  detector  output  is  shown  in  Figure  5-­‐4  for  several  
different  RC  time  constants.  As  long  as  the  modulation  index  remains  below  l00  %  this  upper  envelope  is  the  
original  modulating  signal  (plus  some  DC).  If  the  time  constant  is  too  short,  too  much  decay  in  the  capacitor  
voltage  will  occur  between  peaks  in  the  input  and  the  output  will  be  too  jagged.  Therefore,  we  want  the  time  
1
constant  to  be  much  greater  than  the  period  of  the  carrier  or RC ? Tc = .  If  the  time  constant  is  too  long  the  
fc
output  will  not  follow  the  fastest  variations  in  the  envelope.  The  fastest  variations  in  the  envelope  are  due  to  the  
1
highest  frequency  components  in  the  modulating  signal.  Therefore  we  want   RC = Tm,min =  where   f m ,max
f m,max
is  the  maximum  frequency  of  the  modulating  signal.  It  turns  out  that  the  geometric  mean  of  these  two  periods  is  
a  good  choice,  or RC = TcTm,max .    

  The  most  important  attribute  of  an  envelope  detection  system  is  that  synchronization  is  never  a  
problem.  Envelope  detection  is  also  much  simpler  and  cheaper  than  synchronous  detection.  The  effect  of  change  
in  value  of  the  RC  time  constant  is  illustrated  in  Figure  5-­‐4  below.  In  this  diagram  the  output  appears  as  a  heavy  
black  line.  For  the  first  case  the  approximation  to  the  envelope  is  very  good.  For  the  second  case  the  RC  time  
constant  is  too  long  and  the  envelope  changes  too  fast  for  the  output  of  the  detector  circuit  to  follow.  For  the  
third  case  the  time  constant  is  too  short  and  the  output  is  too  jagged.  It  should  be  pointed  out  that  the  carrier  
frequency  is  typically  so  high  that  many  more  cycles  would  appear  than  are  shown  in  Figure  5-­‐4,  where  the  

66
period  of  the  carrier  is  shown  too  long  for  clarity.  Thus,  any  jagged  edge  appearing  in  the  detector  output  is  
overemphasized  in  Figure  5-­‐4.  Also,  any  rough  edges  on  the  output  can  easily  be  smoothed  by  additional  low  
pass  filtering.  

  The  output  of  the  envelope  detector  of  Figure  5-­‐4  differs  in  another  way  from  the  original  baseband  
signal.  It  has  a  DC  component.  Many  baseband  signals  such  as  audio  do  not  have  any  DC  component.  The  DC  
component  is  easy  to  remove  from  the  output  of  the  envelope  detector  by  simply  following  it  with  a  capacitor  
which  blocks  DC  current  (potentially  harmful  to  speakers).  A  blocking  capacitor  in  series  with  the  detector,  
along  with  the  load  (e.g.  speakers),  forms  a  high  pass  filter.  The  only  requirement  on  this  blocking  capacitor  is  
that  its  value  must  be  chosen  large  enough  to  avoid  attenuation  of  the  lower  audio  frequency  components  
(about  50  Hz).  

Figure 0-4:Effect of RC Time-Constant on Diode Detector Output Voltage.

Example 5.1

The  block  diagram  of  an  AM  modulator  is  shown  below.  The  signal  x(t)  is  an  information  signal  which  has  the  
spectrum  shown.  Baseband  frequencies  from  100  Hz  up  to  3  KHz  are  included  in  x(t).  The  carrier  frequency  is  5  
MHz.  

(a)  Determine  the  type  of  AM  modulation  produced  by  this  system.  
(b)  Sketch  the  magnitude  spectrum  of  the  output  of  this  modulator.  
(c)  What  type  of  AM  detector  is  needed  to  retrieve  x(t)  from  the  AM  signal?  Characterize  this  detector.  
 

Solution

(a)  Note  that  this  modulator  is  very  similar  to  the  DSB-­‐LC  modulator  shown  in  Figure3-­‐.3  except  that  the  
carrier  itself  is  not  added  to  the  output  of  the  multiplier.  Thus,  the  carrier  will  not  be  included  in  y(t),  
only  the  upper  and  lower  sidebands  corresponding  to  x(t).  The  product  circuit  results  in  the  sum  and  

67
difference  frequencies  for  each  and  every  component  of  x(t}.  Therefore  the  type  of  AM  modulation  is  
double  side  band  suppressed  carrier,  DSB-­‐SC.    

(b)  The  resulting  spectrum  which  includes  all  the  sum  and  difference  frequencies  is  shown  below.  

(c)  Because  the  carrier  is  missing  we  must  use  a  synchronous  detector  like  the  one  shown  in  Figure3-­‐1.  
The  local  oscillator  of  this  detector  must  be  set  at  exactly  5  MHz  and  at  the  same  cosine  phase  as  the  
carrier.  Because  the  carrier  is  suppressed  in  the  AM  signal,  there  will  be  no  DC  component  in  the  output  
of  the  multiplier  of  the  detector.  Hence,  the  band  pass  filter  can  be  replaced  by  a  simple  low  pass  filter  
with  a  cutoff  frequency  a  little  larger  than  3  KHz.  This  points  out  a  potential  advantage  of  suppressed  
carrier  transmission.  If  the  baseband  signal  had  information  content  down  to  zero  frequency  (DC)  (this  
might  be  the  case  if  the  signal  were  originating  from  a  transducer)  then  suppressed  carrier  AM  
transmission  would  be  a  better  choice  than  large  carrier  because  the  demodulation  process  can  use  a  
low  pass  filter  and  therefore  preserve  any  information  down  to  0  Hz.  

Example 5.2

If  the  same  information  signal  x(t)  from  Example  3.1  is  modulated  onto  a  5  MHz  carrier  by  a  DSB-­‐LC  process,  
determine  a  suitable  time  constant  for  the  envelope  detector  at  the  receiver.  

Solution

The  inverse  of  the  carrier  frequency  is  the  carrier  period  and  is  equal  to  0.2  ms  and  the  inverse  of  3  KHz,  which  
is  the  highest  frequency  in  the  baseband,  is  0.333  ms.  The  geometric  mean  of  these  two  periods  is  
(0.2 ×10−6 )(0.333×10−3 ) = 8.165  ms.  If  we  choose  this  as  our  RC  time  constant,  it  satisfies  the  criteria  that  it  
be  much  greater  than  the  period  of  the  carrier  and  much  smaller  than  the  period  for  the  highest  frequency  in  
the  information  signal.  

Example 5.3

Given  that   v AM (t ) = 10 cos(2π 106 t ) + 4 cos(2π 1.0002 ×106 t ) + 4 cos(2π 0.9998 ×106 t ) V  is  input  to  an  envelope  
detector,  like  the  one  shown  in  Figure3-­‐3,  what  is  the  value  of  the  DC  content  at  the  output  of  the  detector?  
What  size  capacitor  should  be  placed  in  series  with  the  detector  if  the  expected  resistive  load  to  the  right  of  this  
capacitor  is  10  KΩ  and  frequency  content  down  to  50  Hz  is  to  be  preserved  by  the  high  pass  filter  which  results  
from  addition  of  the  blocking  capacitor?  

Solution

 Assuming  an  ideal  diode  in  the  detector,  the  DC  level  at  the  output  will  equal  the  carrier  amplitude  which,  in  
this  case,  is  10  V  (the  average  value  of  the  envelope  is  equal  to  the  carrier  amplitude,  see  Equation  (2.7).  For  a  
practical  diode,  the  DC  level  will  be  a  few  tenths  of  a  volt  less.  To  choose  a  blocking  capacitor  properly,  we  must  
1
have  the  cutoff  frequency  of  the  high  pass  filter  approximately  equal  to 2π (50) = .  This  results  in  
RC
1
C= = 0.318  µF.  Any  value  larger  than  this  will  do,  so  the  requirement  on  the  blocking  capacitor  is  not  
π 106
very  difficult  to  meet.  

   

68
5.4  Homework  Problems  

Problem 5.l

The  waveform  shown  below  is  applied  to  the  input  of  the  peak  detector  circuit.  Assume  an  ideal  diode.  For  parts  
a  and  b  assume  the  proper  choice  has  been  made  for  the  RC  time  constant  

a. Determine  the  maximum  value  of  the  detector  audio  output  voltage.  
b. Calculate  the  DC  (average)  voltage  at  the  detector  output  (Assume  the  peak  detector  follows  the  
envelope  ideally).  
c. Determine  an  appropriate  value  for  the  capacitor  if  R  =  5  KΩ.  
d. Add  a  component(s)  to  the  detector  circuit  below  which  will  remove  the  average  value  found  in  part  b.  
Indicate  where  the  new  output,  with  zero  DC,  is  located.  
e. If  the  output  is  applied  to  a  speaker,  after  the  DC  is  removed  (DC  is  not  good  for  speakers),  what  will  be  
audible?  

Problem 5.2

Repeat  Problem  5.l  if  the  diode  in  the  envelope  detector  is  reversed  and  points  the  other  way.  Some  of  the  
answers  will  stay  the  same  and  some  will  change.  

Problem 5.3

Given   vin (t ) = 15[1 + 0.3cos(2π 300t )]cos(2π 3 ×106 t ) V  as  the  input  to  the  envelope  detector  shown  in  Problem  
5.l:  

a. Determine  the  value  of  the  capacitor  for  detection  of  the  information  if  R  =  20  KΩ.  
b. Write  an  equation  which  approximates  the  output  of  the  detector  given  that  the  time  constant  has  been  
correctly  chosen.  Ignore  the  jagged  edge  on  the  output  in  your  expression.  
 
 

69
Problem 5.4

Determine  the  output  of  the  product  modulator  shown  in  Example  5.1  if     A cos(ωc t ) = 2 cos(2π ×106 t )  and  
x(t ) = 1 + 0.5cos(2π 500t ) .What  type  of  modulation  does  this  result  in  and  how  would  it  be  most  easily  detected  
at  a  receiver?  

Problem 5.5

Shown  below  is  a  simplified  speech  scrambler  that  can  convert  a  voice  signal,  x(t),  into  an  unintelligible  signal  
z(t).  

a. If  x(t)  is  a  pure  tone  test  signal  of  700  Hz  with  an  amplitude  of  1,  determine  z(t).  
b. If  x(t)  is  now  taken  to  be  a  full  voice  signal  with  frequency  components  from  50  Hz  on  up,  sketch  the  
spectrum  of  z(t)  and  describe  why  it  will  be  unintelligible.  Assume  a  convenient  symmetrical  shape  for  
the  spectrum  at  the  output  of  the  first  LPF.  
c. Explain  in  words  and  using  a  diagram  how  z(t)  can  be  changed  back  into  x(t)  and  hence  unscrambled.  
 

70
Chapter  6:    AM  Receivers  

6.1  Introduction  
  Simply  connecting  an  antenna  to  an  envelope  detector  will  not  make  a  very  good  radio.  The  signal  from  
the  antenna  would  be  very  weak  and  unable  to  produce  much  if  any  sound  from  a  speaker.  This  means  that  a  
good  radio  receiver  requires  amplification  in  addition  to  demodulation.  Also,  the  signals  from  many  stations  are  
gathered  at  once  by  an  antenna  so  that  simply  amplifying  and  then  detecting  would  result  in  much  interference  
between  stations.  A  good  radio  must  provide  a  method  for  selecting  one  and  only  one  station  at  a  time.  This  can  
be  done  by  band  pass  filtering  because  commercial  AM  stations  are  frequency  division  multiplexed.  Each  station  
is  assigned  its  own  distinct  carrier  and  the  frequencies  associated  with  each  station  are  band-­‐limited  so  that  
they  don't  overlap  the  frequencies  of  an  adjacent  station.  Any  good  radio  receiver  must  have  the  property  of  
sensitivity,  the  ability  to  pull  in  weak  signals,  and  the  property  of  selectivity,  the  ability  to  select  one  station  and  
reject  the  rest  without  any  interference.  Selectivity  is  provided  by  a  band  pass  filter.  It  is  useful  to  think  of  a  
band  pass  filter  as  a  window,  which  if  centered  at  the  right  location,  will  allow  one  particular  station  to  pass  
through,  but  no  others.  

  For  a  radio  receiver  to  be  useful,  it  must  be  tunable  to  different  channels.  If  we  think  of  a  bandpass  filter  
as  a  window,  then  one  way  to  change  to  a  different  channel  would  be  to  move  the  window.  This  can  be  done  by  
changing  the  center  frequency  of  a  bandpass  filter.  (For  example,  a  knob  can  be  turned  that  changes  the  value  of  
a  capacitor.)  This  type  of  tuning  is  used  in  a  tuned-­‐radio-­‐frequency  (TRF)  receiver.  To  provide  historical  
perspective  and  an  alternative  to  the  superheterodyne  receiver,  the  TRF  receiver  will  be  discussed  briefly.  A  
problem  with  the  TRF  receiver  is  that  it  is  difficult  to  build  a  high  quality  band  pass  filter  at  RF  (very  high)  
frequencies  that  is  tunable  (moveable  window)  over  a  wide  range  of  carrier  frequencies.  The  TRF  receiver  will  
work  but  there  is  a  better  solution.  The  superheterodyne  receiver  can  provide  superior  sensitivity  and  
selectivity.  With  this  method,  the  window  of  the  bandpass  filter  is  kept  fixed  at  the  same  frequency  and  stations  
are  moved,  one  at  a  time,  into  this  window  by  the  process  of  mixing  (multiplication).  The  process  of  mixing  
results  in  moving  the  locations  of  radio  stations  along  the  frequency  axis.  Outside  the  receiver  each  station  has  
its  same  familiar  carrier  but  inside  the  receiver  a  new  “carrier”  is  produced  by  the  frequency  shift  resulting  from  
multiplication.  Frequency  shifting,  by  the  multiplication  of  signals,  is  called  heterodyning  and  is  used  in  FM  
receivers  and  radar  as  well  as  AM  receivers.  The  superheterodyne  receiver  is  discussed  in  more  detail  in  a  
separate  section  below.  

6.2  TRF  Receiver  


  The  block  diagram  of  a  tuned  radio  frequency  receiver  is  shown  in  Figure  6-­‐1.  The  signal  picked  up  by  
the  antenna  is  both  amplified  and  filtered  by  several  stages  which  are  simultaneously  tuned  to  a  particular  
carrier.  It  is  these  stages  which  provide  both  the  sensitivity  and  selectivity.  An  example  of  a  spectrum  input  to  
the  RF  Amplifier/Filter  is  shown  in  Figure  6-­‐2  in  which  the  receiver  is  tuned  to  the  station  which  is  shown  
shaded  at  carrier ωc .  A  small  amount  of  guard-­‐band  is  shown  in  Figure  6-­‐2  in  order  that  the  sides  of  the  filter  
window  can  be  seen.  In  practice  no  guard-­‐band  is  provided  for  commercial  AM  and  the  filter  sides  are  less  ideal  
than  portrayed  in  the  diagram.  How  then  is  interference  avoided?  Typically  radio  stations  that  are  assigned  
adjacent  carrier  frequencies  are  in  different  geographical  locations.  This  keeps  stations  with  adjacent  carriers  
weak  enough  to  minimize  interference.  Occasionally,  under  the  right  atmospheric  conditions,  stations  located  
far  away  can  skip  off  certain  layers  in  the  atmosphere  and  cause  interference.  

71
Figure 0-1: Block Diagram of a Tuned Radio Frequency Receiver (TRF).

Figure 0-2: Signal Spectrum Input to RF Amplifier/Filter.

  Tunability  can  be  achieved  in  this  type  of  receiver  by  using  a  variable  capacitor  in  the  resonant  circuit  of  
the  RF  Amplifier/Filter.  A  knob  is  used  to  vary  the  capacitor  value  and  hence  the  frequency  to  which  the  center  
of  the  band  pass  filter  is  tuned.  It  is  technically  difficult  to  build  several  stages  of  RF  Amplifier/Filter  which  track  
well  together  as  the  tuning  knob  is  turned.  The  superheterodyne  receiver  is  a  much  better  solution  and  it  gives  
better  selectivity,  sensitivity  and  overall  performance.  

6.3  The  Superheterodyne  Receiver  


  There  are  several  kinds  of  AM  receivers,  but  none  combine  the  qualities  of  efficiency  and  low  cost  as  
well  as  the  superheterodyne  (superhet)  receiver.  To  allow  tuning  to  one  particular  station  and  filter  out  the  rest,  
a  receiver  must  provide  a  filter  window  onto  the  input  spectrum.  The  superheterodyne  provides  tuning  in  a  
different  manner  than  the  TRF  receiver.  The  bandpass  filter  window  of  a  superheterodyne  receiver  is  kept  at  a  
fixed  frequency  location  and  is  not  moved  by  the  tuning  knob.  This  allows  the  bandpass  filter  to  be  of  very  high  
quality  and  to  have  very  steep  sides  on  its  filter  window  function.  This  window  appears  as  the  band  pass  filter  
(BPF)  of  the  IF  Section  (Intermediate  Frequency)  in  Figure  6-­‐3  below.  The  tuning  capability  is  accomplished  by  
moving  a  selected  carrier  and  its  side  bands  down  into  the  window.  The  shifting  of  a  carrier  is  done  by  mixing  
with  the  local  oscillator.  The  mixer  provides  signal  multiplication  of  the  output  of  the  RF  stage  with  the  sinusoid  
produced  by  the  LO  (Local  Oscillator)  and  sum  and  difference  frequencies  result  at  its  output.  For  example,  if  
the  input  s(t)  to  the  mixer  is  a  cosine  wave  at  1000  KHz  and  the  local  oscillator  is  a  cosine  wave  at  1500  KHz,  
then  the  output  will  be  cosine  waves  at  2500  KHz  (the  sum)  and  500  KHz  (the  difference).  If  one  of  these  
frequencies  is  within  the  window  of  the  IF  section  and  the  window  is  relatively  narrow,  then  only  that  one  
component  will  get  through  the  window  and  travel  on  to  the  detector.  A  superheterodyne  receiver  shifts  the  
information  at  the  desired  carrier  frequency  to  a  fixed  value  intermediate  frequency  so  that  the  desired  side  

72
bands  can  be  bandpass  filtered  with  a  sharp  resonance.  The  superhet  receiver  is  used  in  most  AM  systems.  A  
block  diagram  of  a  superhet  receiver  is  shown  in  Figure  6-­‐3.  

Figure 0-3: Block Diagram of a Superheterodyne Receiver.


 

  The  description  which  follows  refers  to  the  block  diagram  of  Figure  6-­‐3.  In  the  superhet  receiver  there  
are  three  distinct  amplifying  sections  (RF,  IF  and  AF),  a  mixer,  a  detector,  an  antenna  and  an  output  transducer,  
usually  speakers.  The  very  weak  signal   vAM (t )  arrives  at  the  antenna  along  with  many  other  radio  signals,  some  
stronger  and  some  weaker.  The  first  subsystem  the  signal  encounters  is  a  broad  bandpass  RF  filter.  The  
operator  includes  the  desired  signal  by  tuning  the  center  frequency  of  the  RF  filter  to  the  carrier  value.  After  
filtering,  the  signal  or  signals  within  the  bandwidth  of  the  RF  filter  are  all  amplified.  The  amplification  in  the  
block  diagram  is  represented  by  the  scale  factor K1 .  The  signals  in  the  system  are  still  weak  but  are  no  longer  
significantly  affected  by  noise  or  other  interference.  

  After  exiting  from  the  RF  section  as  s(t),  the  signals  are  mixed  by  multiplying  them  by  a  sinusoid  
generated  by  a  built-­‐in  oscillator  known  as  the  local  oscillator  (LO).  The  purpose  of  the  mixer  is  to  down-­‐
convert  or  frequency  shift  the  carrier  of  the  desired  signal  to  an  intermediate  frequency  (IF),  so  that  further  
filtering,  amplifying  and  detection  can  take  place.  

  Our  signal  is  now  u(t),  the  IF  section  is  fixed.  Since  a  wide  range  of  input  signals  is  available  for  
reception,  from  535  KHz  to  1705  KHz  in  the  standard  AM  broadcast  band,  it  is  important  that  the  local  oscillator  
be  tuned  such  that  the  difference  in  frequency  between  the  desired  carrier  and  the  local  oscillator  always  equals  
the  IF  frequency.  In  equation  form  this  becomes:  

f LO − f c = f IF or fc + f LO = f IF (this second choice is rarely seen) (6.1)

  For  commercial  AM,   f IF  is  455  KHz  and  the  local  oscillator  is  tuned  to  a  higher  frequency  than  the  
desired  carrier  as  expressed  in  the  first  part  of  Equation  (6.1).  The  window  of  the  IF  filter  is  always  centered  at  
this  constant  IF  value,  and  provides  consistency  in  filtering.  The  filter  is  also  sufficiently  narrow  to  permit  only  
one  radio  station  signal  to  pass  through.  Its  bandwidth  is  about  10  KHz  (2×5  KHz).  This  bandwidth  is  consistent  
with  the  RF  bandwidth  of  a  commercial  AM  radio  station.  The  AM  audio  range  is  limited  to  50  Hz  up  to  5  KHz  
and  so  the  RF  bandwidth  for  a  DSB-­‐LC  signal  is  twice  5  KHz,  or  10  KHz,  which  is  the  same  as  the  width  of  the  IF  
window.  The  primary  reason  for  the  intermediate  frequency  stage  is  to  provide  a  degree  of  selectivity  which  
would  be  difficult  and  expensive  to  make  in  the  higher  frequency  RF  section.  

73
  Once  the  desired  signal  has  been  isolated,  it  is  amplified  and  sent  to  the  detector.  The  detector  is  usually  
of  the  envelope-­‐type.  The  output  of  the  detector,  y(t),  has  been  filtered  to  remove  all  high  frequency  
components  and  any  DC  components  generated  during  the  detection  process.  Finally,  the  detected  signal  is  sent  
to  the  audio  amplifier  whose  gain  is  adjusted  for  the  listener’s  ear.    

  Figure  6-­‐4  illustrates  the  signal  spectrum  at  various  locations  within  a  superheterodyne  receiver.  In  this  
figure,  each  AM  station  is  represented  by  a  different  shape  so  that  the  tuning  and  selecting  process  can  be  
followed.  The  particular  shape  of  a  station  spectrum  has  no  significance.  Different  shapes  are  used  so  that  each  
station  can  be  followed  through  the  process.  The  station  to  which  the  receiver  is  tuned  is  shown  in  shadow.  By  
the  shape  of  its  spectrum,  perhaps  a  soprano  is  hitting  some  high  notes.  

  Many  stations  incident  upon  the  antenna  are  shown  in  part  (a)  of  Figure  6-­‐4,  along  with  the  shape  of  the  
RF  filter  window.  The  RF  filter  is  typically  broad  and  allows  many  stations  to  pass.  It  is  shown  idealized.  In  
practice,  the  sides  of  the  filter  will  not  be  very  steep  but  rather  will  fall  off  gradually.  Part  (b)  shows  the  mixer  
input  and  reflects  some  filtering  by  the  RF  section.  The  mixer  forms  sums  and  differences  of  the  LO  with  each  
and  every  carrier  at  its  input.  For  the  example  depicted  in  Figure  4-­‐4,   f LO − f c = 455  KHz,  the  IF  frequency.  
Thus,  the  desired  carrier  along  with  its  associated  sidebands  is  shifted  to  a  new  frequency  centered  at  455  KHz.  
It  is  as  though  a  new  intermediate  carrier  has  been  substituted  for  the  RF  carrier.  Actually  the  output  of  the  
mixer  is  not  as  clean  as  depicted  in  Figure  6-­‐4.  All  the  sum  frequencies,  the  LO,  and  the  carriers  are  all  present  to  
some  degree  at  the  output  of  the  mixer  but  these  other  components  are  not  important.  This  is  because  only  the  
one  station,  shown  shaded,  passes  through  the  very  sharp  bandpass  filter  window  of  the  IF  section.  This  filter  
window  is  shown  in  part  (c)  of  the  diagram  and  the  IF  output  in  part  (d).  The  IF  section  also  provides  some  
more  gain  represented  by  the  scale  factor K 2 .  The  detector  section  then  removes  the  information  from  the  
intermediate  carrier  resulting  in  the  spectrum  shown  in  part  (e).  Note  that  DSB-­‐LC  AM  detection  is  equivalent  to  
removing  the  carrier  and  shifting  the  upper  side  band  back  down  to  the  baseband  where  it  originated  at  the  
transmitter.  Finally,  the  resulting  audio  signal  is  further  amplified  by  the  audio  amplifier  to  drive  those  big  
speakers.  

74
|  v(f  )|  
BPF of the RF section  

f  
f  c  
a. Input to RF Filter/Amp  

|  s(f  )|  

f  
f  c  
b. RF Amplifier Output/ Input to Mixer  
|  u(f  )|  
BPF of the IF section  

f  

f  IF  =  f  c-     fLO


    f  c  +f  LO  
c. Mixer Output/ IF Filter Input  

|  z(f  )|  

f  

f  IF  
d. IF Amplifier Output/ Detector Input  

|  y(f  )|  

f  

e. Detector Output/ Audio Amplifier Input  

Figure 0-4: Signal Spectra in a Superheterodyne Receiver

  There  remains  a  possible  source  of  interference.  A  carrier  located  at  455  KHz  higher  than  the  local  
oscillator  will  also  form  a  difference  of  455  KHz  at  the  mixer  output.  This  image  carrier  will  therefore  fall  within  
the  IF  window.  This  potential  problem  is  solved  by  filtering  out  the  image  carrier  at  the  RF  filter  stage.  The  
requirement  is  that  the  bandwidth  of  the  RF  filter  be  less  than  2×910  KHz  =  1820  KHz  which  is  not  difficult  to  
meet.  The  image  location  is  shown  in  Figure  6-­‐5.  The  term  image  is  derived  from  the  fact  that  the  image  carrier  

75
and  the  RF  carrier  are  located  equidistant  from  the  local  oscillator  frequency  just  as  your  face  and  its  image  are  
equidistant  from  the  mirror.  

Figure 0-5: Radio and Image Station Locations.

Example 6.1

A  superheterodyne  receiver  is  tuned  to  a  commercial  AM  radio  station  which  is  broadcasting  music  on  a  carrier  
frequency  of  1  MHz  

(a)  Determine  the  frequency  to  which  the  local  oscillator  of  the  receiver  is  tuned.  

(b)  Determine  the  image  frequency  and  the  maximum  bandwidth  of  the  RF  filter  such  that  this  image  
will  not  interfere.  

(c)  What  range  of  frequencies  is  contained  in  this  DSB-­‐LC  AM  signal?  

(d)  What  range  of  frequencies  is  present  at  the  output  of  the  IF  stage?  

(e)  What  would  be  audible  if  the  output  of  the  IF  stage  were  connected  directly  to  the  speakers?    

(f)  What  range  of  frequencies  is  present  at  the  output  of  the  detector?  

Solution

(a)  A  commercial  AM  receiver  is  tuned  at  455  KHz  higher  than  the  carrier.  This  gives   f LO = 1.455 MHz.  

(b)  The  image  frequency  is  higher  than  the  LO  by  455  KHz,  so   f IMAGE = 1.455 + 0.455 = 1.910 MHz.  If  we  
assume  that  the  RF  filter  is  centered  at  1  MHz  and  that  its  upper  cutoff  frequency  must  be  less  than  1.91  
MHz  then  half  of  its  bandwidth  must  be  less  than  1.91  -­‐  1  =  0.91  MHz  and  its  full  bandwidth  must  be  less  
than  twice  this  or  less  than  1.82  MHz.  

(c)  The  baseband  for  commercial  AM  is  from  about  50  Hz  up  to  5  KHz.  For  DSB-­‐LC  both  sidebands  are  
present  so  in  addition  to  the  1  MHz  carrier,  the  upper  side  band  contains  components  from  1  MHz  +  50  
76
Hz  up  to  1  MHz  +  5  KHz  and  the  lower  sideband  contains  frequency  components  from  1  MHz  -­‐  5  KHz  up  
to  1  MHz  -­‐  50  Hz.  

(d)  The  new  carrier  at  the  IF  output  is  455  KHz.  In  addition  to  the  455  KHz  carrier  there  will  be  the  
upper  side  band  ranging  from  455  KHz  +  50  Hz  up  to  460  KHz  (455  +  5)  and  the  lower  side  band  ranging  
from  450  KHz  (455  -­‐  5)  up  to  455  kHz  -­‐  50  Hz.  

(e)  The  listener  would  hear  nothing  because  all  the  frequencies  listed  in  part  (d)  are  well  above  the  
audible  range  for  humans,  which  ends  at  about  20  KHz.  

(f)  The  original  baseband,  which  was  an  audio  signal  containing  components  from  50  Hz  up  to  5  KHz,  
appears  at  the  output  of  the  detector.  

Example 6.2

When  the  local  oscillator  of  a  superheterodyne  receiver  is  adjusted  to  1365  KHz,  a  1000  Hz  pure  test  tone  is  
heard  from  the  speakers,  indicating  that  this  commercial  AM  station  is  broadcasting  a  test  signal.  

(a)  Determine  the  RF  carrier  frequency  if  the  transmitter  is  a  DSB-­‐LC  commercial  transmitter.  

(b)  Determine  the  frequency  spectrum  of  the  signal  at  the  output  of  the  IF  stage  of  the  receiver.  

Solution

(a)  The  local  oscillator  frequency  is  455  KHz  higher  than  the  carrier,  so    

fc = 1365 − 455 = 910  KHz.  

(b)  For  pure  tone  modulation  (this  is  what  is  heard  at  the  speaker  output,  so  it  must  be  what  was  
modulated  onto  the  carrier  at  the  transmitter)  each  side  band  consists  of  only  one  component  
Therefore,  the  frequencies  are  454  KHz,  455  KHz  and  456  KHz.  

Example 6.3

If  the  knob  of  a  superheterodyne  receiver  is  tuned  to  receive  the  station  to  the  immediate  right  of  the  shaded  
station  shown  in  Figure  6-­‐4,  draw  the  changes  which  will  occur  in  the  spectra  of  Figure  6-­‐4.  

Solution

The  new  station  will  now  appear  in  the  IF  window  and  show  up  at  the  output  of  the  detector  as  shown  below.  

77
   )|  
|v(f
BPF of the RF section  

f  
fc    
a. Input to RF Filter/Amp  

   )|  
|s(f

f  
fc    
b. RF Amplifier Output/ Input to Mixer  
   )|  
|u(f
BPF of the IF section  

f  

f  IF  =  f  c-    fLO


    fc    +f  LO  
c. Mixer Output/ IF Filter Input  

   )|  
|z(f

f  

f  IF  
d. IF Amplifier Output/ Detector
Input  
   )|  
|y(f

f  

e. Detector Output/ Audio Amplifier Input  

78
6.4  Homework  Problems  
 

Problem 6.1

A  commercial  AM  tuner  which  is  tuned  to  one  of  the  stations  shown  in  the  amplitude  spectrum  below  has  the  
following  characteristics:  

Current Frequency of Local Oscillator, f LO 1885 KHz


RF Amplifier Bandwidth 500 KHz
Intermediate Frequency, f IF 455 KHz
Bandwidth of IF Amplifier, BWIF 10 KHz

A  spectrum  analyzer  is  to  be  connected  at  several  test  points  of  a  superheterodyne  radio  receiver,  the  block  
diagram  of  which  is  shown  below.  Key  characteristics  of  the  receiver  settings  are  listed  in  the  table  above.  The  
carrier  frequencies  in  the  amplitude  spectrum  are:   f1 = 1390  kHz,   f 2 = 1430  kHz,   f3 = 1440  KHz  and  
f 4 = 1470  KHz.  Determine  and  plot  the  expected  spectrum  display  at  points  B,  C,  D,  and  E,  given  the  spectrum  
as  shown  at  point  A.  Include  both  sum  and  difference  generated  spectra  at  point  C.  

79
Problem 6.2

A  superheterodyne  receiver  operates  on  a  set  of  AM  carriers  that  range  from  600  KHz  to  

2.5  MHz  and  which  are  separated  by  20  KHz.  It  is  also  known  that  the  IF  frequency  is  500  KHz,  the  RF  amplifier  
has  a  bandwidth  of  200  KHz,  the  audio  baseband  signal  is  band  limited  to  a  maximum  frequency  of  7  KHz  and  
the  LO  is  tuned  to  a  frequency  higher  than  the  carrier  (Note,  this  is  not  the  typical  commercial  AM  system).  

a. Determine  the  frequency  range  over  which  the  local  oscillator  must  be  tunable.  
b. Determine  the  bandwidth  of  the  IF  amplifier  such  that  all  the  side  band  frequencies  but  no  more  are  
passed  through  the  IF  window.  
c.  Determine  the  minimum  bandwidth  of  the  audio  power  amplifier  such  that  none  of  the  information  
passed  to  it  by  the  detector  is  lost.  
d. Determine  the  guard-­‐band.  
e. If  the  receiver  is  tuned  to  a  1.2  MHz  carrier,  find  the  value  of  the  local  oscillator  frequency.  
f. Determine  if  image  frequencies  are  a  problem.  Use  a  quantitative  argument.  

Problem 6.3

The  spectrum  of  WCRP  is  shown  below.  We  wish  to  receive  and  demodulate  this  station  with  a  superheterodyne  
receiver.  The  IF  filter  is  centered  at  700  KHz  and  the  LO  is  tuned  to  a  lower  frequency  than  the  RF  carrier  (Note,  
this  is  not  the  typical  commercial  AM  system).  

a. To  what  frequency  must  the  LO  be  tuned  to  receive  WCRP?  
b. What  is  the  value  of  the  upper  corner  frequency  of  the  IF  filter  such  that  the  band  width  is  just  enough  to  
include  the  full  RF  bandwidth  of  WCRP  but  no  more?  
c. What  is  the  value  of  the  image  frequency  and  how  can  this  potential  interference  be  rejected?  

Problem 6.4

A  superheterodyne  receiver  in  a  hypothetical  communications  system  can  tune  to  RF  carriers  ranging  from  100  
MHz  up  to  101  MHz  while  the  corresponding  image  frequencies  range  from  120  MHz  to  121  MHz.  Determine  
the  range  over  which  the  LO  is  tunable  and  the  value  of  the  IF.  

Problem 6.5

Calculate  the  image  frequency  when  a  commercial  DSB-­‐LC  AM  receiver  is  tuned  to  a  540  KHz  carrier.  Is  this  
image  in  the  AM  band?

80
Chapter  7:    Frequency  Modulation  
 

7.1  Introduction  
  So  far  we  have  investigated  the  effect  of  varying  the  amplitude  of  a  sinusoidal  carrier  in  order  to  
transmit  information.  We  know  that  it  takes  three  quantities  to  specify  a  sinusoid:  amplitude,  frequency  and  
phase.  Modulation  is  defined  as  the  process  of  varying  some  characteristic  of  the  carrier  wave  in  accordance  
with  the  instantaneous  value  of  an  input  signal.  Since  a  sinusoid  has  other  parameters  which  can  be  varied,  
amplitude  modulation  is  not  the  only  means  of  modulating  a  sinusoidal  carrier.  The  instantaneous  frequency  of  
the  carrier  can  also  be  varied  in  accordance  with  the  baseband  information  signal.  This  type  of  modulation  is  
called  frequency  modulation  or  FM.  It  is  possible  to  vary  the  phase  angle  instead,  which  results  in  PM  which  will  
not  be  discussed  here.  Suffice  it  to  say  that  it  is  very  similar  to  FM.  

7.2  Description  of  the  Modulation  Process  


  FM  results  when  the  frequency  of  the  carrier  is  made  to  vary  in  response  to  an  information  signal.  Let’s  
start  with  the  carrier vc (t ) = Vc cos[θ (t )] ,  where  normally   θ (t ) = ωct + φ  and  the  instantaneous  angular  
dθ (t )
frequency  is  the  time  derivative  of  the  angle,  or ω (t ) = = ωc .  (If  you  are  driving  your  car  at  a  constant  
dt
speed  then  the  time  derivative  of  your  location  is  equal  to  that  speed.)  

  When  the  carrier  is  frequency  modulated,  the  instantaneous  frequency  becomes  a  function  of  time  and  
dependent  on  the  information  signal  and  is  given  by   ωinst (t ) = ωc + kx(t ) where x (t )  is  the  information  signal  
and   k is  a  conversion  parameter  resulting  from  the  circuit  which  converts  changes  in  the  amplitude  of  the  
information  signal  into  changes  in  the  carrier  frequency.  Therefore,   k  has  a  unit  of  Hz/V.  To  make  more  
progress  in  the  analysis  of  FM  we  will  assume  pure  tone  modulation  such  that   x(t ) = Vm cos(ωmt ) and  the  
instantaneous  frequency  becomes   ωinst (t ) = ωc + kVm cos(ωmt ) = ωc + Δω cos(ωmt ) where   Δω  is  called  the  
frequency  deviation  and  represents  the  amplitude  of  the  changes  in  the  carrier  frequency.    

  The  maximum  value  of  the  carrier  is   ωc + Δω and  the  minimum  value  is   ωc − Δω and  the  carrier  value  
is  right  in  between  these  two  values.  (Note,  if  we  are  measuring  and  discussing  frequency  in  Hz  rather  than  
rad/s  then f ’s  replace  the ω ’s,  remembering  that ω = 2π f ).  The  result  of  frequency  modulation  by  a  square  
wave  is  shown  below  in  Figure  5-­‐1.  When  the  square  wave  has  its  low  value,  the  period  of  the  carrier  is  
maximum  and  its  frequency  minimum.  When  the  square  wave  has  its  maximum  value  the  period  of  the  carrier  
is  minimum  and  the  frequency  is  maximum.  The  rate  at  which  these  changes  to  the  carrier  occur  is  the  same  as  
the  frequency  of  the  modulating  signal.  Stated  in  a  different  way,  the  length  of  time  for  the  frequency  to  go  
through  one  complete  cycle  of  change  is  the  same  as  the  period  of  the  modulating  signal.  The  amount  of  
frequency  change  in  the  carrier  is  proportional  to  the  strength  of  the  modulating  signal  (its  amplitude).  For  
Figure  7-­‐1  the  difference  between  the  maximum  frequency  and  the  minimum  frequency  is 2Δω ,  twice  the  
frequency  deviation.  

81
Figure 0-1: FM Modulated Carrier.

  To  determine  the  functional  form  of  our  FM  signal,  the  instantaneous  frequency  must  be  integrated  with  
respect  to  time  to  get  the  total  angular  displacement  and  that  total  angular  displacement  placed  in  the  argument  
of  the  cosine  function.  This  is  analogous  to  integrating  velocity,  which  may  be  changing  with  time,  to  obtain  total  
distance  traveled,  just  as  your  car  odometer  does.  In  equation  form  this  becomes    

⎛ t ⎞
vFM (t ) = Vc cos ⎜ ∫ [ωc + Δω cos(ωmτ )]dτ ⎟ (7.1)
⎝ 0 ⎠
which,  upon  integration,  gives    

⎛ Δω ⎞
vFM (t ) = Vc cos ⎜ ωct + sin(ωmt ) + α ⎟ (7.2)
⎝ ωm ⎠

Δω
where α  is  a  constant  of  integration  and  can  be  set  equal  to  zero  without  loss  of  generality.  Here ω m  is  a  
dimensionless  ratio  and  is  a  measure  of  how  much  frequency  modulation  is  present.  Because  of  its  importance  

it  is  given  the  name  modulation  index  and  the  symbol β .  Now  we  have  
vFM (t ) = Vc cos[ωct + β sin(ωmt )]  where
Δω Δf
β= = .
ωm fm

  Frequency  modulation  is  produced  by  causing  the  instantaneous  frequency  of  the  RF  carrier  to  vary  
systematically  by  an  amount  proportional  to  the  modulating  signal.  Thus,  the  rate  of  the  variation  relates  to  the  
frequency  of  the  modulating  source,  and  the  maximum  extent  of  variation  to  the  amplitude  of  the  modulating  
signal.  

82
  One  of  the  problems  involved  in  describing  FM  is  that  there  are  several  different  frequencies  to  keep  
track  of.    So  far,  we  have  the  instantaneous  frequency, f inst ,  the  carrier  frequency,   f c ,  the  modulating  frequency,  
f m ,  and  the  frequency  deviation   Δf .  Moreover,  all  of  these  must  be  distinguished  from  just  plain  frequency,  
which  is  the  independent  variable  in  the  frequency  domain.  If  the  modulation  were  turned  down  to  zero  the  
frequency  of  the  sinusoid  would  be f c ,  the  carrier  frequency.  As  modulation  is  increased,  the  frequency  of  the  
wave  changes  back  and  forth,  from  high  pitch  to  low  pitch  to  high,  etc.  at  a  rate  equal  to  the  frequency  of  the  
modulating  signal.  The  maximum  amount  by  which  the  pitch  changes,  from  the  starting  point  of  no  modulation,  
is  the  frequency  deviation, Δ f ,  and  depends  on  the  amplitude  of  the  modulating  signal.  

7.3  FM  Spectrum  


  The  equation  for  an  FM  signal,  with  pure  tone  modulation,  looks  formidable.  The  function  involves  the  
cosine  of  a  sine.  Advanced  math  and  Fourier  analysis  can  be  used  to  find  the  spectrum  of  an  FM  signal.  The  
details  of  the  mathematics  are  beyond  the  scope  of  this  course  or  these  notes.  When  an  FM  wave  is  broken  
down  to  determine  its  spectral  content,  it  turns  out  that  there  are  theoretically  an  infinite  number  of  sideband  
Δf
frequencies  and  the  amplitudes  at  these  frequencies  depend  on β = .  This  sounds  a  bit  complicated  but  the  
fm
results  can  be  summarized  easily.  The  frequency  components  in  an  FM  signal  formed  by  the  frequency  
modulation  of  a  carrier  by  a  pure  tone  (single  frequency)  are  summarized  below:  

fc carrier

fc ± fm first sideband (upper and lower)

fc ± 2 fm second sideband (upper and lower)

fc ± 3 fm third sideband (upper and lower)

f c ± nf m nth sideband (upper and lower)

… and so on to infinity

  An  FM  wave  has  infinitely  many  sidebands  with  only  one  frequency  present  in  the  information  signal.  
For  a  more  realistic  multiple  component  modulating  signal  (music  for  example)  the  situation  is  even  more  
complicated.  Because  of  its  many  sidebands,  we  can  anticipate  that  an  FM  wave  will  require  a  much  larger  
transmission  bandwidth  than  a  comparable  amplitude  modulated  wave.  

  Figure  7-­‐2  illustrates  several  FM  spectra  and  the  dependence  of  the  FM  spectrum  on β .  By  careful  
consideration  of  the  relative  amplitudes  of  the  components  as  a  function  of   β  a  useful  rule  has  been  developed  
which  can  estimate  the  total  bandwidth  of  an  FM  signal.  For  a  given  modulation  index,  there  is  some  frequency  
beyond  which  the  sidebands  can  be  neglected.  For  example,  for β = 1,  everything  beyond  the  third  sideband  is  
insignificant.  This  means  that  although  the  FM  signal  theoretically  contains  infinitely  many  frequencies,  there  is  

83
a  band  of  frequencies  in  which  most  of  the  power  is  concentrated.  This  band  of  frequencies  is  called  the  
bandwidth  of  the  signal,  BW.  The  rule  that  estimates  this  BW  is  called  Carson’s  rule  and  is  given  by  the  equation:  

BW = 2(β + 1) f m = 2(Δf + f m ) (7.3)

  If  the  modulating  signal  contains  many  frequency  components  then  the  maximum  frequency  component  
should  be  used  in  place  of f m in  Carson’s  rule.  However,  the  situation  is  actually  more  complicated  because  the  
amplitude  of  the  highest  frequency  component  in  the  modulating  signal  might  be  quite  low.  

Figure 0-2:The Spectra of Sinusoidally Modulated FM Signals for Various Values of


β.

  It  is  worth  making  several  more  points  about  FM  and  the  FM  spectrum.  The  amplitude  of  any  given  
Δf
frequency  component  in  the  FM  spectrum  varies  as   β = is  changed.  It  is  possible  that  the  amplitude  at  the  
fm
frequency  of  the  carrier  is  zero.  Does  this  mean  zero  power  in  the  FM  wave?  

No,  the  power  in  the  FM  signal  is  not  a  function  of β .  The  power  is  determined  by  Vc,  the  amplitude  of  the  un-­‐
modulated  carrier  (the  amplitude  of  the  wave  is  unaffected  by  frequency  modulation  as  shown  in  Figure  5-­‐2).  
Δf
As  long  as  Vc  is  not  changed,  the  power  in  an  FM  signal  remains  the  same.  As   β =  is  varied,  the  amount  of  
fm
power  in  each  sideband  component  changes  but  the  total  power  remains  the  same.    Another  important  feature  

84
of  an  FM  spectrum  is  the  uniform  spacing  of  the  lines.  Each  line  in  Figure  7-­‐2  is  separated  from  its  neighbor  by
f m ,  the  modulating  frequency.  

7.4  Advantages  and  Disadvantages  of  FM  


To  conclude  this  section  several  advantages  and  disadvantages  of  FM  compared  to  AM  are  listed.  

1.  The  major  advantages  of  FM  are:  

  a)  Superior  performance  in  the  presence  of  noise  and  interference.  

  b)  Smaller  geographic  interference  areas.  That  is,  two  FM  stations  can  operate  substantially  
  closer     without  interference  compared  to  two  similar  AM  stations.  This  is  because  of  the  higher  
  carrier    frequencies  used  in  commercial  FM.  Commercial  AM  carriers  have  more  of  a  tendency  to  
  skip  off  layers     in  the  Earth’s  atmosphere.  

  c)  Less  power  need  be  transmitted  for  same  amount  of  power  received  compared  to  DSB-­‐LC  AM  
  transmission.  

2.  The  major  disadvantages  of  FM  are:  

  a)  Up  to  20  times  more  bandwidth  is  required.  

  b)  FM  systems  are  considerably  more  complicated.  

  c)  FM  is  a  very  non-­‐linear  process  making  analysis  of  FM  more  difficult  than  that  of  AM.  

  In  summary,  you  may  recall  from  our  discussion  of  AM  that  the  two  sidebands  (USB  and  LSB)  are  
redundant  since  the  same  information  is  contained  in  each.  In  fact,  as  we  saw,  single-­‐sideband  (SSB)  modulation  
is  based  on  the  fact  that  anyone  sideband  contains  all  the  necessary  information.  When  we  examine  the  
sideband  structure  of  an  FM  signal,  especially  a  wide  band  FM  signal,  we  find  that  the  information  is  replicated  
many  times  over.  FM  is  extremely  redundant!  But  it  is  this  redundancy  which  results  in  the  outstanding  noise  
rejection  associated  with  FM  systems.  The  magnitudes;  frequencies  and  phases  of  all  the  sidebands  are  related  
to  the  carrier  and  to  each  other  in  a  very  specific  manner.  Because  of  this  precise  relationship,  the  sidebands  are  
said  to  be  coherent  with  each  other  and  with  the  carrier.  Noise,  on  the  other  hand,  can  be  thought  of  as  
consisting  of  a  very  large  collection  of  sinusoids,  each  with  a  random  amplitude,  frequency  and  phase.  

  The  many  components  in  the  random  distribution  of  noise  cannot  have  the  same  special  relationship  to  
each  other  and  to  the  carrier  that  an  FM  signal  has.  For  this  reason,  an  FM  receiver  can  extract  the  intricate,  but  
coherent,  FM  signal  out  of  the  incoherent  noise  much  more  readily  than  an  AM  receiver  can  extract  its  signal,  
which  has  only  two  sidebands.  The  price  for  the  redundancy  of  the  FM  signal  is  increased  signal  bandwidth.  This  
is  in  general  true  for  all  signal  transfer  systems.  Redundancy  improves  transfer  accuracy  and  reduces  noise,  but  
we  pay  with  bandwidth.  In  other  words  we  can  trade  bandwidth  for  performance  or  vice  versa.  

7.5  FM  Receiver  


  In  order  to  demodulate  the  FM  signal  we  need  to  have  a  device  which  produces  an  output  whose  
amplitude  is  linearly  proportional  to  the  frequency  of  the  input  signal.  We  have  already  studied  such  a  device.  
An  RLC  resonant  circuit  which  has  the  amplitude  response  of  a  band  pass  filter  can  do  the  job  provided  the  
proper  operating  region  of  the  filter  is  used.  

85
  Over  the  narrow  operating  region  shown  in  Figure  7-­‐3,  the  slope  of  the  filter  approaches  a  straight  line.  
In  this  region  the  amplitude  of  the  signal  exiting  from  our  detector  will  be  linearly  proportional  to  the  frequency  
of  the  input  signal.  If  we  restrict  our  system  to  operate  in  the  narrow  linear  region,  we  can  effectively  
demodulate  an  FM  signal  using  an  envelope-­‐detection  system.  The  result  of  passing  an  FM  signal  through  a  
slope  detector  is  shown  in  Figure  7-­‐4.  The  output  of  the  FM  detector  is  now  amplitude  modulated  as  well  as  
retaining  its  frequency  modulation.  The  information  to  be  recovered  is  now  contained  in  both  the  amplitude  
and  the  frequency  of  the  signal.  The  signal  can  be  viewed  as  an  amplitude  modulated  FM  signal.  The  information  
can  now  be  retrieved  using  a  peak  detector.

Figure 0-3: Transfer Function of a Slope Detector.

Figure 0-4: Input and Output for a Slope Detector.

Figure 0-5: Block Diagram of a FM Demodulator.


 
86
  An  FM  demodulation  scheme  is  shown  in  Figure  6-­‐5.  It  consists  of  an  FM  detector  followed  by  the  
familiar  peak  detector  to  extract  the  information  signal  from  the  variations  in  the  amplitude  resulting  from  the  
FM  slope  detector.  The  FM  demodulator  is  only  one  part  of  an  FM  receiver.  Most  FM  receivers  are  superhets.  
Like  their  AM  counterparts,  the  tuning  control  tunes  both  the  RF  filter  and  the  local  oscillator.  In  fact  there  is  no  
real  difference  between  the  two  receivers  except  that  the  FM  receiver  adds  a  limiter  to  eliminate  noise  and  a  
slope  discriminator  to  demodulate  the  FM  signal.  Many  modern  FM  receivers  use  something  called  a  phase  
locked  loop  instead  of  a  slope  detector  to  demodulate  the  FM  but  the  idea  remains  much  the  same.  A  typical  FM  
receiver  is  shown  in  Figure  7-­‐6.  

  The  IF  of  a  commercial  FM  receiver  is  10.7  MHz  contrasted  to  the  455  KHz  of  commercial  AM  receivers.  
The  IF  bandwidth  is  also  larger,  with  a  200  KHz  compared  with  10  KHz  for  AM.  The  wider  bandwidth  permits  
FM  signals  to  carry  a  wider  baseband  than  AM  and  therefore  afford  better  signal  fidelity.  Unlike  AM,  music  from  
an  FM  station  will  contain  frequency  components  well  above  5  KHz  .  The  limiter  is  also  an  important  distinction  
between  FM  and  AM.  Much  of  the  noise  that  adversely  affects  AM  is  picked  up  in  the  transmission  medium  and  
is  amplitude  in  nature.  For  example,  lightning  causes  amplitude  spikes  which  become  audible  for  AM  broadcast.  
This  amplitude  noise  will  also  be  picked  up  in  the  transmission  medium  by  FM  but  it  can  be  eliminated  within  
the  receiver  by  a  limiter  type  circuit  before  the  signal  is  FM  demodulated.  Since  output  signal  amplitude  
information  comes  from  input  frequency  variations,  the  limiter  removes  all  amplitude  noise  spikes  before  a  
slope  detector  creates  the  amplitude  variations  for  final  output  envelope  detection.  

Figure 0-6: Block Diagram of a FM Superheterodyne Receiver.

Example 7.1

The  output  signal  of  an  FM  transmitter  is  shown  below.  It  is  applied  to  an  antenna  which  can  be  represented  by  
a  75Ω  resistor  (radiation  resistance).  Assume  that  the  modulating  signal  is  sinusoidal.  Estimate:    

(a)  The  frequency  of  the  modulating  signal,    


(b)  The  frequency  of  the  un-­‐modulated  carrier,    
(c)  The  peak  frequency  deviation,    
(d)  The  modulation  index,    
(e)  The  approximate  RF  bandwidth,    
(f)  The  characteristics  of  the  RF  spectrum,  
(g)  The  total  power  radiated  by  the  antenna.  
87
Solution

(a)  The  period  of  the  modulating  signal  is  the  same  as  the  period  of  the  changes  in  the  frequency.  

Starting  at  10  µs  and  ending  at  30  µs  the  frequency  goes  from  a  minimum  through  a  maximum  and  back  
to  the  same  minimum.  Thus  the  period  of  the  modulating  signal  is  20  µs  and  the  modulating  frequency  is  
the  reciprocal  of  the  period  which  is  5  KHz.

(b)  Starting  at  10  µs  and  ending  at  30  µs  there  are  20  cycles  of  the  carrier  in  a  20  µs  interval  which  gives  
an  average  of  1  µs  per  cycle  or  1  MHz.  Because  the  modulating  signal  is  sinusoidal,  the  frequency  of  the  
carrier  should  also  be  equal  to  the  average  of  the  maximum  and  minimum  frequencies.  

(c)  By  carefully  using  a  ruler  with  a  mm  scale,  the  minimum  period  can  be  estimated  to  be  about  2/3  ms  
which  gives  a  maximum  frequency  of  1.5  MHz.  Similarly,  the  maximum  period  can  be  estimated  to  be  2  
ms  which  gives  a  minimum  frequency  of  0.5  MHz.  The  average  of  these  two  values  is  indeed  1  MHz  as  
was  found  in  part  (b)  above.  The  absolute  value  of  the  difference  between  the  carrier  and  either  the  
maximum  or  minimum  frequency  is  0.5  MHz  which  is Δf max .  

(d)  The  modulation  index  is  the  ratio  of  the  frequency  deviation  to  the  modulating  frequency  so  

Δf 500
β= = = 10 .
fm 50

(e)  Using  Carson’s  rule  we  get   BW = 2 f m (1 + β ) = 2 × 50(1 + 10) = 1.1 MHz.  

(f)  The  spectrum  consists  of  lines  symmetrically  placed  on  either  side  of  the  1  MHz  carrier  and  
separated  by  50  KHz.  If  we  have  10  lines  above  and  10  lines  below  the  carrier  this  would  give  a  total  
span  of  1000  KHz  and  if  we  have  11  above  and  11  below  1200  KHz.  The  estimated  BW  is  in  between  
these  two  values  but  we  can’t  have  half  lines.  This  simply  points  out  that  Carson’s  rule  is  an  
approximation  and  is  not  exact.  For  this  case  we  would  probably  assume  a  BW  of    1200  KHz  and  a  total  
of  23  significant  lines  (counting  the  carrier)  in  the  spectrum.  

(g)  The  power  depends  only  upon  the  amplitude  of  the  signal  (10  V)  and  the  radiation  resistance  of  the  
antenna,  75Ω.    Thus,    

88
102 2
P = 0.5 = W.  
75 3
Example 7.2

A  100  MHz  sinusoidal  carrier  is  frequency  modulated  with  a  3-­‐KHz,  3-­‐V  peak  sinusoid.  If  the  modulator  has  a  
sensitivity  of  2  KHz/V,  determine:    

(a)  The  amplitude  of  the  frequency  deviation  of  the  carrier,    
(b)  The  modulation  index,    
(c)  The  approximate  signal  bandwidth  using  Carson’s  rule,    
(d)  The  expression  as  a  function  of  time  for  the  FM  signal  for  a  cosine  carrier  of  5  V  peak.  Assume  the  
modulating  signal  is  a  cosine.  

Solution

(a)  The  sensitivity  of  2  KHz/V  is  the  constant  k  which  represents  the  conversion  from  modulating  signal  
amplitude  to  carrier  frequency  deviation.  To  get  the  frequency  deviation,  we  simply  multiply  this  
sensitivity  by  the  amplitude  of  the  modulating  signal  to  get Δf = (2 KHz / V )(3V ) = 6 KHz .  

(b)  The  modulation  index  is  the  ratio  of  the  frequency  deviation  to  the  modulating  frequency  so  
Δf 6
β= = = 2  
fm 3

(c)  Using  Carson’s  rule  we  get     BW = 2 f m (1 + β ) = 2 × 3(1 + 2) = 18  KHz.  

(d)  Filling  in  the  values  found  above  into  the  equation  for  an  FM  signal  we  get  
vFM (t ) = 5cos[2π ×108 t + 2sin(2π × 3000t )] V.  

Example 7.3

Measurements  on  an  FM  signal  indicate  a  maximum  period  of   1.001×10−8 s  and  a  minimum  period  of  
0.999 ×10−8 s.  The  modulating  signal  is  a  20  KHz  pure  tone.    

(a)  Determine  the  value  of  the  carrier  frequency,    

(b)  Find  the  value  of  the  frequency  deviation.  

Solution

1
(a)  The  maximum  frequency  is  the  inverse  of  the  minimum  period  which  is   f max = = 100.1  
0.999 ×108

1
MHz  and  the  minimum  frequency  is  the  inverse  of  the  maximum  period  which   f min = = 99.9
1.001×10−8
MHz.  The  carrier  is  the  average  of  these  two  frequencies  which  gives   f c = 100 MHz.  

89
(b)  The  frequency  deviation  is   f max − f c = f c − f min = 0.1  MHz,  which  means β = 5 .  

90
7.6  Homework  Problems  

Problem 7.1

If  the  instantaneous  carrier  frequency  is  varied  sinusoidally  from  a  minimum  value  of  99.8  MHz  to  a  maximum  
of  100.2  MHz  by  FM  modulating  the  carrier  with  a  40  KHz,  sinusoidal  modulating  signal,  determine:  

a. The  modulation  index  


b. The  approximate  bandwidth  by  use  of  Carson’s  rule  
c. Sketch  and  label  the  amplitude  spectrum  

Problem 7.2

A  100  MHz  carrier  is  modulated  such  that  the  value  of  the  instantaneous  frequency  varies  sinusoidally  from  
99.1  MHz  to  100.1  MHz  and  the  period  of  this  variation  is  0.05  ms.    

a. What  type  of  modulation  is  being  used?  


b. Determine  the  frequency  of  the  modulating  signal  being  used.  
c. Determine  the  approximate  RF  bandwidth  of  the  signal.  

Problem 7.3

When  a  93.4  MHz  carrier  is  frequency  modulated  by  a  4  KHz  sine  wave  the  resultant  frequency  deviation  is  40  
KHz.  

a. Determine  the  highest  and  lowest  frequencies  attained  by  the  modulated  signal.    
b. Determine  the  modulation  index.  
c. Determine  the  approximate  bandwidth  and  sketch  the  frequency  spectrum.  

Problem 7.4

The  FCC  has  allocated  the  range  from  88  to  108  MHz  for  commercial  FM  broadcasting.  The  RF  bandwidth  
allotted  each  station  is  200  KHz.  

a. How  many  stations  can  be  assigned  different  carriers  over  the  full  FM  band?  
b. Over  what  range  must  the  local  oscillator  of  an  FM  superheterodyne  receiver  be  tunable  if  the  LO  is  
tuned  to  a  higher  frequency  than  the  carrier?  
c. Determine  the  maximum  image  frequency  that  could  appear  in  the  IF  stage  of  an  FM  superheterodyne  
receiver  and  the  maximum  BW  of  the  RF  amplifier  such  that  this  image  is  not  a  problem.  
d. Determine  the  minimum  BW  of  the  audio  amplifier  of  a  good  FM  superheterodyne  receiver  so  that  none  
of  the  source  audio  content  is  lost.  

Problem 7.5

When  a  carrier  is  frequency  modulated  by  a  4  KHz  sine  wave  the  resulting  FM  signal  has  a  maximum  frequency  
of  106.218  MHz  and  a  minimum  frequency  of  106.196  MHz.  

a. Determine  the  carrier  frequency.  


b. Determine  the  frequency  deviation.  
c. Determine  the  approximate  RF  band  width  and  sketch  the  spectrum.    
91
Chapter  8:    Noise  in  Communication  

8.1  Introduction  
  Under  ideal  conditions,  the  signal  generated  at  the  source  is  identical  to  the  signal  reproduced  at  the  
destination  after  passing  through  the  transmission  and  reception  processes  and  the  intervening  channel.    This  
distortionless  transmission  is  the  goal  as  illustrated  in  Figure  8-­‐1.  

Channel

Source Transmitter Receiver Destination


Input signal at source is identical to output signal at destination

Figure 0-1: Ideal Communication System

  However,  as  expected,  the  ideal  is  rarely  real.    At  each  stage  in  the  communication  system,  deviations  are  
introduced  which  cause  the  final  output  signal  to  vary  from  the  initial  input  signal  as  depicted  in  Figure  8-­‐2.      

Internal Noises Internal Noises


Thermal/Johnson External Noises Thermal/Johnson
Shot/Semiconductor Equipment/Industrial Shot/Semiconductor
Excess Atmospheric Excess
Transit-Time Space/Extraterrestrial Transit-Time

Channel

Source Transmitter Receiver Destination

Figure 0-2: Realistic Communication System

  In  analog  communications  systems,  the  difference  between  the  initial  input  signal  and  the  final  output  
signal  is  generically  called  “noise”.    (In  digital  systems,  the  variation  is  called  “bit  error”.)    Since  noise  is  defined  
as  the  difference  between  the  output  signal  and  the  input  signal,  by  direct  extension,  the  output  signal  is  
modeled  as  the  sum  of  the  input  signal  plus  noise  as  illustrated  in  Figure  8-­‐3.  

In  this  chapter  we  will  explore  how  we  measure  noise,  where  it  comes  from  and  some  of  the  methods  used  to  
overcome  its  degrading  effects.  

93
Noise

Input Output
Signal Signal

Figure 0-3: Noise Modeling

8.2  Expressing  Noise  –  SNR  and  Noise  Ratio/Figure  


  Since  both  signal  strength  and  noise  level  can  vary,  a  means  of  comparing  the  two  power  levels  was  
created  called  Signal  to  Noise  Ratio  (SNR)  defined  as:  

SNR=PS/PN (8.1)

Where:  
SNR=  Signal  to  Noise  Power  Ratio  
PS=Signal  Power  
PN=Noise  Power  

If  both  the  signal  voltage  and  noise  voltage  are  applied  across  a  resistor,  using  P=V²/R  and  some  simple  algebra:  

SNR=(VS/VN)² (8.2)

Where:  
SNR=  Signal  to  Noise  Power  Ratio  
VS=Signal  Voltage  
VN=Noise  Voltage  
 

Commonly,  SNR  is  expressed  in  dB  which  results  in  the  following  expressions:  

SNRdB=10×log(PS/PN)=20×log(VS/VN) (8.3)

  The  performance  of  most  communication  systems  is  generally  much  more  dependent  on  this  ratio  of  
signal  to  noise  power  rather  than  the  absolute  value  of  either  power  independently.    In  order  to  compensate  for  
spreading  and  attenuation  losses  in  the  channel,  most  receivers  have  sufficient  gain  to  make  a  very  weak  signal  
power  audible.    However,  they  can  experience  significant  troubles  recognizing  that  same  signal  in  the  presence  

94
of  significant  noise  power.    Some  high  quality  systems  are  designed  with  special  signal  formats  and  extensive  
processing  to  enable  reproduction  of  the  original  signal  even  when  the  noise  power  exceeds  the  signal  power  
received.  

Example 8.1: Across  a  1Ω  resistor,  a  signal  voltage  is  measured  at  1µVRMS  while  the  noise  voltage  is  measured  
at  63.24nVRMS.    Calculate  the  signal  power,  the  noise  power,  the  SNR  in  rational  form  and  the  SNR  in  dB  via  two  
methods.

Solution

PS=VS²/R=(1µV)²/1Ω=1W  

PN=VN²/R=(63.24nV)²/1Ω=4×10-­‐15W  

SNR=1pW/(4×10-­‐15W)=250  

SNRdB=10×log[1pW/(4×10-­‐15W)]=23.98dB  

SNRdB=20×log(1µV/63.24nV)=23.98dB  

Since  all  amplifiers  contribute  their  own  noise  to  the  signal,  the  SNR  at  the  output  of  an  amplifier  is  always  
smaller  than  at  the  input.    This  degradation  is  quantified  in  a  term  known  as  Noise  Ratio  (NR):  

NR=SNRinput/SNRoutput (8.4)

When  the  noise  ratio  for  an  amplifier  is  expressed  in  dB,  it  is  called  Noise  Figure  (NF):  

NF=10×log(NR)=10×log(SNRinput/SNRoutput) (8.5)

  Low  noise  amplifiers  have  low  noise  ratios  (just  larger  than  one)  and  low  noise  figures  (just  larger  than  
zero).    Amplifiers  on  their  own,  without  filters  or  other  processing,  cannot  achieve  an  output  SNR  greater  than  
the  input  SNR  to  attain  noise  ratios  or  figures  lower  than  these  ideals.  

PNinput
NR1 NR2 NR3
PSinput A1 A2 A3 (PSoutput + PNoutput)

Figure 0-4: Noise Ratio for Cascaded Amplifiers

95
  If  we  “cascade”  or  use  several  amplifiers  in  series  of  gains  A1,  A2,  A3,  etc.  and  corresponding  noise  ratios  
of  NR1,  NR2,  NR3,  etc.  as  illustrated  in  Figure  8-­‐4,  some  algebra  will  derive  Friis’  Formula  which  relates  the  
composite  noise  ratio  (NRT)  of  the  set  of  amplifiers  to  the  gains  and  noise  ratios  of  the  individual  stages:  

NRT=(PSinput×PNoutput)/(PNinput×PSoutput) (8.6)

NRT=NR1 + (NR2-1)/A1 + (NR3-1)/(A1×A2) + … (NRn-1)/(A1×A2×…An-1) (8.7)

Example 8.2: The  SNR  at  the  input  to  a  two  stage  amplifier  is  250  and  125  at  the  output  of  the  first  stage  
amplifier.    The  gain  of  the  first  amplifier  stage  is  10,  the  gain  of  the  second  amplifier  stage  is  100  and  the  noise  
ratio  of  the  second  amplifier  stage  is  5.    Calculate  the  noise  ratio  of  the  first  stage  in  rational  form  and  the  noise  
figure  in  dB  form.    Calculate  the  gain  and  noise  ratio  of  the  composite  two  stage  amplifier  and  the  output  SNR.  

Solution

NR=SNRinput/SNRoutput  =250/125=2  

NF=10×log(SNRinput/SNRoutput)=10×log(250/125)=3dB  

AT=  A1×A2=10×100=1000  

NRT=  NR1+(NR2-­‐1)/A1  =  2+  (5-­‐1)/10  =  2.4  =  SNRinput/SNRoutput  =  250/  SNRoutput  

SNRoutput  =250/2.4=104.2  

Note  that  the  noise  ratio  of  the  first  stage  dominates  the  composite  noise  ratio.    This  makes  sense  since  any  
noise  introduced  at  this  stage  will  be  amplified  by  all  subsequent  stages.    Thus,  special  attention  is  generally  
devoted  to  ensuring  that  the  first  stages  of  multiple  stage  amplifiers  have  the  lowest  possible  noise  ratios.  

8.3  Sources  of  Noise  -­‐  External  Noise  


  Since  distortion  generating  noise  is  undesirable,  the  first  logical  question  leading  to  reducing  or  
eliminating  it  is  “From  where  does  it  come?”    Depending  on  the  stage  of  the  communication  system  at  which  the  
noise  is  introduced,  the  noise  is  defined  as  either  externally  generated  or  internally  generated  noise.    
Undesirable  signal  added  to  the  communication  system  in  the  communication  channel  is  created  outside  the  
system  and  hence  is  referred  to  as  external  noise.    External  noise  comes  from  two  sources:  man-­‐made  and  
natural.  

  Random  man-­‐made  noise  is  referred  to  as  equipment  or  industrial  noise.    It  results  from  large,  rapidly  
changing  currents  and/or  any  operation  which  results  in  the  creation  of  a  spark/plasma.    Typical  sources  
include  unshielded  transformers,  switches,  automobile  engines,  brushed  electric  motors  and  fluorescent  lights.    
Minimization  of  this  source  of  noise  typically  consists  of  shielding  the  source  from  the  channel,  maximizing  the  

96
distance  from  the  source  to  the  communication  systems  elements  and/or  minimizing  the  time  that  the  noise  
source  and  the  communication  system  are  operating  simultaneously.  

  Non-­‐random  man-­‐made  noise  is  referred  to  as  interference.    It  can  be  unintentional  such  as  a  radio  
station  from  a  neighboring  city  experiencing  just  the  right  atmospheric  conditions  that  its  signal  overwhelms  
the  desired  local  station  signal  in  a  given  location.    It  can  be  caused  by  capacitive,  magnetic,  radiative,  or  ground  
loop  coupling  which  provides  alternate  signal  paths  into  the  receiver  for  signals  other  than  the  desired  one.    
Man-­‐made  interference  can  also  be  intentional,  in  which  case  it  is  called  “jamming”.    Defeating  interference  is  
commonly  done  by  removing  undesired  coupling  paths,  exploiting  some  unique  characteristic  of  the  desired  
signal  to  differentiate  it  from  the  interference  or  shifting  the  desired  signal  frequency  away  from  the  interfering  
signal  frequency  band.  

  Natural  external  noise  originates  from  two  primary  sources:  atmospheric  and  extraterrestrial.    
Atmospheric  noise  principally  results  from  lightning,  a  very  large  spark  or  plasma.    It  creates  a  very  large  but  
short-­‐lived  noise  spike  over  long  distances  at  frequencies  up  to  ~30MHz.    It  can  be  minimized  by  “noise  
blanking”  or  disabling  the  receiver  until  the  large  amplitude  spike  passes.    Unfortunately,  any  signal  sent  during  
the  spike  is  still  lost.  

  Extraterrestrial  noise  comes  from  the  sun  and  stars  as  the  solar  wind,  solar  flares,  sun  spots  and  cosmic  
radiation.    It  produces  random  voltages  primarily  in  the  10MHz  to  1.5GHz  range.    Extraterrestrial  noise  in  the  
channel  has  to  be  filtered  out  in  the  manner  to  be  described  later  in  this  chapter.  

8.4  Internal  Noise  


  While  external  noise  is  introduced  to  the  communication  channel  from  sources  outside  the  
communication  system,  internal  noise  is  contributed  by  the  system  components  themselves.    Internal  noise  is  
minimized  by  prudent  engineering  trade-­‐offs  in  the  design  and  operation  of  the  communication  system.    There  
are  four  common  sources  of  internal  noise  in  most  communication  systems:  thermal,  shot,  excess  and  transit-­‐
time.  

  Thermal  noise,  also  known  as  Johnson  or  resistance  noise,  is  induced  by  the  random  motion  of  electrons  
in  resistors  due  to  heat.    Thermal  noise  is  considered  “white  noise”  in  that  its  magnitude  is  the  approximately  
same  across  the  measurable  spectrum.    Consequently,  the  thermal  noise  effects  observed  are  directly  related  to  
the  frequency  span  over  which  the  signal  is  studied.    The  open  circuit  noise  voltage  (VN)  induced  by  this  motion  
is  a  function  of  the  prevalent  temperature  of  the  selected  resistor,  the  bandwidth  over  which  the  noise  is  
measured  and  the  value  of  the  resistor  as  given  by  Johnson’s  Formula:    

VN=(4kTBR)½ (8.8)

Where:  
VN=Noise  Voltage  
k=Boltzman’s  Constant=1.381×10-­‐23J/ºK  
T=Temperature  in  ºK;  (T[ºK]=T[ºC]+273.15;  T0=290ºK~  room  temperature)  
B=Bandwidth  over  which  the  noise  is  observed,  in  Hertz  
R=Resistance  across  which  voltage  is  measured  in  Ω    
 

97
  To  determine  the  maximum  noise  power,  which  can  be  transferred  to  a  load,  we  evaluate  the  Thevenin  
equivalent  circuit  with  the  load  resistor  (RL)  selected  of  the  same  value  as  the  noise  generating  resistor  (R)  as  
shown  in  Figure  8-­‐5.  

IL

R
+
RL VL
+ -
VN
-

Figure 0-5: Thevenin Equivalent of Noise Generating Circuit

Since  RL=R,  VL=VN/2  and  IL=VN/2R.    So,  PN=PL=VL×IL=VN²/4R    And,  after  some  minor  algebra:  

PN=kTB (8.9)

Where  :    
PN=Noise  Power  transferred  
k=Boltzman’s  Constant=1.381×10-­‐23J/ºK  
T=Temperature  in  ºK;  (T[ºK]=T[ºC]+273.15;  T0=290ºK~  room  temperature)  
B=Bandwidth  over  which  the  noise  is  observed,  in  Hertz  
 
  Thus,  for  a  given  bandwidth,  any  noise  power  level  could  be  expressed  as  a  noise  temperature  and  any  
noise  added  to  a  signal  could  correspond  to  an  equivalent  noise  temperature  added  to  the  initial  noise  
temperature.    For  an  amplifier  with  an  assumed  input  noise  power  equivalent  to  T=T0=290ºK  and  a  Noise  
Ratio=NR,  a  little  algebra  determines  the  equivalent  noise  temperature  of  the  amplifier:  

Teq=290°K(NR-1) (8.10)

This  provides  an  alternative  method  to  noise  ratios  for  calculating  SNRs  at  the  outlet  of  an  amplifier  using  
equivalent  noise  temperatures:  

SNRoutput=PSoutput/PNoutput=PSinput/[k(T+Teq)B] (8.11)

Where:  
T=Effective  Noise  Temperature  of  input  in  ºK;  (T0=290ºK~  room  temperature)  
Teq=Effective  Noise  Temperature  of  amplifier  in  ºK;  (T0=290ºK~  room  temperature)  

98
Example 8.3:  The  input  to  an  amplifier  has  a  signal  power  of  1pW  and  a  noise  temperature  of  290ºK  for  a  
bandwidth  of  1MHz.  The  noise  ratio  of  the  amplifier  is  3.    Calculate  the  SNR  at  the  input,  the  effective  noise  
temperature  of  the  amplifier  and  the  SNR  at  the  output  of  the  amplifier  using  both  noise  ratio  and  effective  
temperature  methods.  

Solution

PN=kTB=1.381×10-­‐23J/°K×290°K×1MHz=4×10-­‐15W  

SNRinput=1pW/4×10-­‐15W=250  

Teq=290(NR-­‐1)=290°K(3-­‐1)=580°K  

SNRoutput  =SNRinput/NR=  250/3=83.3  

SNRoutput=PSinput/[k(T+Teq)B]=1pW/[1.381×10-­‐23J/°K×(290°K+580°K)×1MHz]  SNRoutput  =83.3  

  As  is  apparent  from  the  equations  above,  the  best  ways  to  minimize  thermal  noise  voltages  are  to  
minimize  the  temperatures  and  resistances  of  the  components  and  operate  over  the  minimum  bandwidth  
possible.  

  The  second  source  of  internal  noise  is  referred  to  as  “shot”  or  “semiconductor”  noise.    Current  flow  is  
really  not  continuous  but  the  average  movement  of  a  large  number  of  discrete  charges  (electrons  or  holes).    
These  charges  cross  the  junctions  in  semiconductors  at  random  times  and  by  random  paths,  creating  a  random  
variation  in  the  average  current  flow  (IN).    In  devices  where  current  flows  are  uniting  or  separating,  such  as  
bipolar  junction  transistors,  a  related  effect  causes  variations  in  the  current  split  between  the  flows  and  is  
referred  to  as  “partition  noise”.    Shot  noise  is  another  “white  noise”  whose  effect  is  directly  related  to  both  the  
bandwidth  observed  and  the  DC  bias/average  current  across  the  PN  junction  as  described  in  the  following  
equation:  

IN=(2qI0B)½ (8.12)

Where:  
IN=RMS  noise  current  in  Amps  
q=electron  charge=1.6×10-­‐19  Coulomb  
I0=DC  bias  current  in  the  device  in  Amps  
B=Bandwidth  over  which  the  noise  is  observed,  in  Hertz  
 

  From  this  equation,  the  best  ways  to  minimize  shot  noise  currents  are  to  minimize  the  bias  currents  of  
the  components  and  operate  over  the  minimum  bandwidth  possible.  

  Not  all  internal  noise  is  “white  noise”.    Excess  noise,  sometimes  called  flicker  noise,  pink  noise  or  1/f  
noise,  affects  semiconductors,  resistors  and  conductors.    It  appears  to  be  caused  by  spatial  variations  in  charge  
carrier  density  hence  effective  resistance.    It  is  most  severe  at  low  frequencies  and  drops  as  1/f  at  higher  

99
frequencies.    It  is  minimized  through  judicious  choice  of  construction  materials  for  resistors  or  by  operating  at  
frequencies  above  that  where  excess  noise  becomes  less  important  than  other  noise  contributors.  

  While  excess  noise  dominates  the  internal  noise  terms  at  low  frequencies,  “transit-­‐time  noise”  is  most  
important  at  high  frequencies,  just  below  the  cutoff  frequency  of  the  device.    When  the  period  of  the  signal  gets  
close  to  the  time  required  for  carriers  to  cross  a  junction,  the  carriers  may  not  make  it  all  the  way  across  the  
junction  before  being  pulled  or  drifting  back.    This  creates  a  variation  in  current  flow  directly  proportional  to  
operating  frequency.    Its  impact  is  minimized  by  operating  at  frequencies  well  below  cutoff  frequency.      

  The  typical  frequency  dependence  of  these  noise  sources  is  illustrated  in  Figure  8-­‐6.    Note  that  internal  
noise  is  generally  minimized  by  choice  of  operating  frequency  in  the  “bowl”  of  the  total  noise  curve  or  choice  of  
components  to  put  the  minimum  noise  band  in  the  vicinity  of  the  operating  frequency.  

8.5  Overcoming  Noise:  Filtering  


  In  defeating  the  various  sources  of  noise  described  so  far,  each  type  had  its  own  countermeasures.    
However,  there  is  one  technique  which  works  with  most  varieties  of  noise:  signal  matched  frequency  “filtering”.    
Since  most  noise  types  introduce  many  frequency  components  beyond  those  already  present  in  the  desired  
signal,  eliminating  frequencies  not  present  in  the  desired  signal  will  reduce  noise  power  levels  and  improve  
SNR.    To  illustrate,  let’s  examine  the  sample  signal  in  Figure  8-­‐7.  

Noise vs Frequency

1.00E-07
Noise Voltage (V)

1.00E-08 1/f (V)


1.00E-09 Therm (V)
Shot (V)
1.00E-10 Transit (V)
1.00E-11 Total (V)

1.00E-12
10000

100000
100

1000

1E+06

1E+07

1E+08
10

Frequency (Hz)

Figure 0-6: Typical Noise vs. Frequency Curves

100
Impact of Filtering on Noise
Input Signal
1.5 Output Signal

0.5
Voltage

100
1
4
7
10
13
16
19
22
25
28
31
34
37
40
43
46
49
52
55
58
61
64
67
70
73
76
79
82
85
88
91
94
97
-0.5

-1

-1.5
Time

Figure 0-7: “Clean” Input Signal and “Noisy” Output Signal


 

  Examination  of  the  two  waveforms  shows  that  they  are  related  but  the  noisy  signal  is  obviously  not  very  
“pretty”.    Next  we’ll  compare  the  frequency  spectral  of  the  two  signals  in  Figures  8-­‐8  and  8-­‐9.  

Frequency Spectrum of Input Signal

50

45

40

35

30
Voltage

25

20

15

10

0
1

11

13

15

17

19

21

23

25

27

29

31

33

35

37

39

41

43

45

47

49

51

53

55

57

59

61

63

65

Frequency

Figure 0-8: Frequency Spectrum of “Clean” Input Signal

101
Frequency Spectrum of Output Signal

50

45

40

35

30
Voltage

25

20

15

10

0
1

11

13

15

17

19

21

23

25

27

29

31

33

35

37

39

41

43

45

47

49

51

53

55

57

59

61

63

65
Frequency

Figure 0-9: Frequency Spectrum of “Noisy” Output Signal


 

  The  clean  signal  has  a  “well  behaved”  spectrum  with  an  apparent  peak  at  the  fundamental  frequency  of  
the  waveform.    The  noisy  signal  still  has  that  basic  peak,  but  it  is  rising  out  of  a  “grass”  of  random  noise  
frequency  components  which  mask  the  smaller  magnitude  components  of  the  clean  signal.    If  we  “zero”  out  the  
frequency  components  of  the  noisy  signal  which  are  outside  the  frequency  peak  from  the  original  signal  and  
reconstruct  the  resulting  waveform,  we  get  the  result  of  Figure  8-­‐10.  

Impact of Filtering on Noise

1.5
Input Signal
Filtered Output

0.5
Voltage

0
1

10

13

16

19

22

25

28

31

34

37

40

43

46

49

52

55

58

61

64

67

70

73

76

79

82

85

88

91

94

97

100

-0.5

-1

-1.5
Time

Figure 0-10: “Clean” Input Signal and Filtered “Noisy” Output Signal

  This  process  of  removing  frequency  components  outside  the  desired  range  is  called  “filtering”  and,  as  
can  be  seen  above,  is  a  very  effective  way  to  restore  the  output  signal  to  very  close  to  the  input  signal.    There  is  
still  some  variation  in  the  filtered  signal  due  to  noise  magnitude  and  phase  components  at  frequencies  close  to  
102
the  desired  frequency;  however,  the  deviation  of  the  filtered  waveform  from  the  input  signal  is  significantly  less  
than  that  of  the  unfiltered  waveform.    SNR  has  been  significantly  improved.  

  So,  what  does  all  this  mean  in  real  life?    You  are  in  the  process  of  building  your  Elenco  AM  radio  kit.    Part  
of  the  test  procedure  associated  with  the  construction  involves  checking  the  bandwidth  of  various  filters  in  the  
radio.    If  your  bandwidths  fall  directly  on  the  specified  values,  your  filters  are  matched  to  the  signal  at  those  
points  in  the  circuit.    They  are  not  too  narrow  such  that  part  of  the  signal  is  lost  along  with  the  eliminated  noise.    
They  are  not  too  wide  such  that  excessive  noise  is  brought  in  along  with  the  signal.    Your  radio  will  achieve  the  
maximum  possible  SNR.    This  means  your  radio  will  be  able  to  pick  out  more  and  weaker  stations  than  radios  
with  poorer  SNR  performance.    Received  radio  stations  will  suffer  less  static,  hiss  and  interference  than  less  well  
matched  receivers.  

  Noise  commonly  becomes  a  major  limiter  in  the  performance  of  communication  systems.    It  is  measured  
in  several  ways  including  ratios,  logarithmic  scales  and  equivalent  temperatures.    It  comes  from  sources  both  
internal  to  the  transmission  and  reception  equipment  and  externally  from  the  channel  itself.    The  primary  way  
it  is  eliminated  is  through  frequency  filters  matched  to  the  desired  signal  spectrum.    When  the  signal  to  noise  
ratio  of  a  system  is  improved,  its  ability  to  detect,  receive  and  demodulate  the  desired  signal  is  vastly  improved.  

References

[1] Blake, Roy, Electronic Communication Systems, 2nd Edition, Delmar, 2002

[2] Frenzel, Louis E., Principles of Electronic Communication Systems, 2nd Edition, McGraw Hill, 2003  

103
Chapter  9:    Digital  Communications  

9.1  Introduction  
  The  storage,  processing,  manipulation  and  transmission  of  information  represented  in  digital  form  by  l’s  
and  0’s  has  become  more  and  more  important.  Music  is  stored  in  digital  form  on  CD’s,  digital  computers  are  
essential  in  business,  engineering,  entertainment  and  science.  A  few  TV  stations  in  the  United  States  are  already  
broadcasting  high  definition  television,  HDTV.  The  format  for  HDTV  in  the  U.S.  is  digital.  Much  of  the  
information  traveling  around  the  United  States  and  the  world  is  first  converted  into  a  digital  form  before  
transmission.  Digital  communications  offer  a  number  of  advantages:  

1.  For  long  distance  communications,  the  digital  1’s  and  0’s  can  be  reconstituted  by  intermediate  
repeater  stations  with  essentially  zero  error.  The  digital  format  is  more  tolerant  of  noise  and  noise  does  
not  build  up  with  increasing  distance  as  it  does  with  analog  communications.  

2.  Much  of  the  circuitry  used  for  modulation  and  demodulation  is  digital  which  means  that  it  is  highly  
reliable  and  stable  and  can  be  easily  fabricated  on  integrated  circuits.  

3.  Information  can  easily  be  stored  in  digital  form  for  later  retrieval.  For  example,  packets  of  information  
relayed  by  satellites  can  temporarily  be  stored  until  the  satellite  is  over  the  intended  recipient  of  the  
information.  

4.  Computers  can  easily  manipulate  and  encrypt  information  in  digital  form.  Secure  communications  is  
very  important  in  the  military  and  in  business  and  industry.  

5.  Very  efficient  algorithms  exist  for  the  compression  of  digital  information,  for  example,  the  jpeg  and  gif  
formats  for  pictures  which  are  used  on  the  Internet.  

6.  Digital  codes  exist  for  reducing  the  effects  of  noise  and  for  detecting  and  correcting  errors  of  
transmission.  For  example,  the  use  of  the  parity  bit  allows  the  receiver  to  detect  certain  errors  and  
request  a  retransmission  if  desired.  

  Offsetting  these  advantages  to  some  degree  is  the  added  complexity  and  comparatively  larger  
bandwidth  requirements  of  digital  communications  systems.  However,  modern  integrated  circuitry  and  modern  
digital  computers  make  complexity  much  less  an  issue.  We  will  discuss  some  of  the  concepts  and  subsystems  of  
digital  communications.  Years  of  study  are  required  for  anything  approaching  complete  mastery  of  the  whole  
area.  Many  of  the  concepts  and  techniques  involved  in  digital  communications,  such  as  analog-­‐to-­‐digital  and  
digital-­‐to-­‐analog  conversion  and  binary  numbers  and  logic,  are  included  elsewhere  in  this  course.  First  we  will  
look  at  the  conversion  of  analog  information  into  digital  form,  called  pulse  code  modulation  (PCM).    

9.2  Pulse  Code  Modulation  


  The  process  of  pulse  code  modulation  is  summarized  in  Figure  9.1  as  a  block  diagram.  The  elements  of  
this  process  are  illustrated  in  Figure  9.2  discussed  below.  It  should  be  carefully  studied  for  full  understanding.  

105
Figure 0-1: Pulse Code Modulation Block Diagram.

Figure 0-2: Pulse Code Modulation Process Illustrated.

9.2.1  Sampling  
  Sampling  is  the  first  process  involved  in  the  conversion  of  an  analog  into  a  digital  signal.  Sampling  is  the  
measurement  of  a  signal  at  discrete  and  regular  times.  Hourly  sampling  of  the  temperature  outside  would  result  
in  a  sequence  of  numbers,  one  for  each  hour.  The  processes  of  sampling,  quantization  and  encoding  are  
illustrated  in  Figure  7-­‐2.  First,  the  continuous  analog  signal  is  processed  by  a  sampling  circuit  which  measures  
the  value  of  the  signal  at  discrete  times  indicated  by  the  arrows  at  the  bottom  of  Figure  7-­‐2.  Usually  the  sample  
times  are  uniformly  spaced.  The  output  of  the  sampling  process  is  a  sequence  of  numbers  representing  the  
input.  To  avoid  losing  any  information  the  samples  have  to  be  spaced  closely  enough  together  so  that  the  shape  

106
of  the  analog  input  signal  is  not  distorted  or  lost.  The  samples  must  be  taken  frequently  enough  to  avoid  loss  of  
information.  Music  stored  in  a  CD  would  not  sound  very  good  if  the  sampling  rate  were  1  KHz.  

  How  fast  is  fast  enough?  The  Sampling  Theorem  states  that  to  avoid  loss  of  information,  a  band  limited  
signal  must  be  sampled  at  a  rate  equal  to  or  greater  than  twice  the  bandwidth  of  the  signal.  If  an  analog  signal  is  
sampled  fast  enough,  the  information  can  be  retrieved  by  low  pass  filtering  the  sequence  of  samples.  If  we  are  
dealing  with  a  baseband  signal  containing  frequency  components  from  about  zero  on  up  to  some  maximum  
frequency,  then  the  sampling  rate  must  be  equal  to  or  greater  than  twice  that  maximum  frequency.    

  An  audio  signal  with  frequency  components  from  about  40  Hz  up  to  about  20  KHz  is  an  example  of  a  
baseband  signal.  High  quality  digital  audio  requires  a  minimum  sampling  rate  of   40 ×103 samples/sec.  If  the  
signal  of  interest  is  a  band-­‐limited  communications  signal  modulated  onto  a  high  frequency  carrier,  the  
minimum  sampling  rate,  required  to  preserve  the  information  content,  is  equal  to  twice  the  range  of  frequencies  
around  the  carrier.  Thus,  if  the  bandwidth  of  the  information  signal  is  10  KHz  and  it  is  centered  around  a  100  
MHz  carrier,  the  minimum  sampling  rate  is  20  KHz,  not  2(100  MHz  +  5  KHz).  In  equation  form  the  Sampling  
Theorem  translates  to f sampling ≥ 2 f Bandwidth .  This  minimum  rate  is  called  the  Nyquist  rate,  named  after  the  
engineer  who  investigated  the  mathematics  of  the  sampling  process.  The  theoretical  limit  is  never  really  fast  
enough.  For  example,  to  make  music  CD  recordings,  the  input  signal,  which  has  a  maximum  frequency  of  20  
KHz,  is  sampled  at  about  44  KHz.  

  Many  signals  have  high  frequency  components  that  do  not  contain  essential  information  but  that  can  
cause  problems  when  sampling  is  done.  The  problem  of  aliasing  occurs  when  the  sampling  rate  is  lower  than  
twice  the  highest  frequency  of  the  signal.  It  results  in  high  frequency  components  masquerading  as  lower  
frequency  values  and  causing  distortion.  Musical  instruments  can  create  frequencies  higher  than  20  KHz  which  
are  not  audible.  To  avoid  aliasing  problems,  a  music  signal  is  first  low  pass  filtered  to  remove  any  components  
greater  than  20  KHz.  This  is  what  is  meant  by  band  limiting  a  signal.  Filtering  is  used  to  remove  all  but  a  limited  
range  of  frequencies  from  a  signal  while  preserving  the  essential  information  content.  If  all  frequencies  above  
about  3  KHz  are  removed  from  a  person’s  voice  before  telephone  transmission,  the  voice  remains  both  
intelligible  and  recognizable  although  they  may  not  sound  exactly  the  same  as  in  person.  

9.2.2  Pulse  Amplitude  Modulation  


  Rectangular  pulses,  shown  in  gray,  are  located  at  each  sample  point  in  Figure  7-­‐2.  These  gray  pulses  
illustrate  the  idea  of  pulse  amplitude  modulation  (PAM).  Instead  of  transmitting  the  full  analog  signal,  pulses  of  
amplitudes  proportional  to  the  sample  values  are  transmitted.  One  advantage  of  this  technique  is  that  the  pulses  
from  several  different  analog  signals  can  be  interleaved  together  and  transmitted  over  the  same  line.  In  this  
way,  each  signal  gets  to  time  share  the  common  resource  of  a  transmission  line.  Time  sharing  in  this  way  is  
called  time  division  multiplexing  (TDM).  It  can  also  be  used  to  multiplex  many  digital  channels  together  and  will  
be  discussed  again  below.  The  tops  of  the  pulses  in  Figure  9-­‐2  are  shown  as  being  flat.  Ideally,  the  top  of  each  
pulse  would  follow  the  contour  of  the  analog  signal  during  the  pulse  interval.  However,  it  is  much  easier  to  
create  flat  top  pulses  using  sample  and  hold  circuitry.  The  distortion  created  by  the  flat  tops  is  minimal  at  the  
receiver,  especially  if  the  pulses  are  of  short  duration.  

  Another  issue  with  PAM  is  the  amount  of  bandwidth  required  in  the  baseband  to  faithfully  represent  the  
pulses.  If  the  band  width  is  gradually  reduced  by  low  pass  filtering,  the  pulses  will  first  become  more  and  more  
rounded  until  they  are  distorted  to  the  point  that  information  is  lost.  Band  width  becomes  important  with  
respect  to  the  medium  over  which  the  pulses  are  transmitted.  This  medium  might,  for  example,  be  wires  that  

107
would  have  a  low  pass  filter  response.  If  the  bandwidth  of  the  transmission  medium  is  not  enough,  the  pulses  
will  become  too  distorted  to  retain  complete  information.  The  theoretical  minimum  bandwidth  is  given  by
BW > 0.5 f sampling .  This  bandwidth  would  not  preserve  the  rectangular  pulse  shape  but  would  preserve  the  
amplitude  information  at  the  sample  times.  If  the  minimum  bandwidth  is  used,  the  pulses  will  no  longer  look  
rectangular  but  instead  become  rounded.  A  better  and  more  conservative  rule  is  to  take BW > f sampling .  The  
issue  of  baseband  bandwidth  will  come  up  again  for  the  digital  representation  of  the  signal,  because  pulses  are  
also  used  to  represent  1’s  and  0’s.    

9.2.3  Other  Analog  Pulse  Modulation  Schemes  


  A  few  time  division  multiplexed  analog  systems  don’t  have  the  dynamic  range  or  sensitivity  to  
successfully  pass  a  PAM  signal.    For  such  systems  there  are  two  more  pulse  modulation  schemes  which  can  
work.    The  first  option  is  called  Pulse  Duration  or  Pulse  Width  Modulation  (PDM/PWM).    It  can  be  thought  of  as  
turning  a  pulse  amplitude  signal  “on  its  side”.    All  pulses  are  of  uniform  magnitude,  but  their  duration  during  the  
sample  interval  is  directly  proportional  to  the  magnitude  of  the  signal  to  be  passed  in  the  same  way  that  the  
amplitude  of  a  PAM  signal  is  proportional  to  the  magnitude.    This  technique  is  somewhat  less  sensitive  to  noise  
or  signal  strength  variations  than  PAM  since  these  degradations  act  principally  on  the  magnitude  and  not  the  
timing  of  the  signal.    Additionally,  since  the  transmitted  power  is  proportional  to  the  square  of  the  pulse  
amplitude  but  only  the  first  power  of  pulse  duration,  variations  in  transmitted  power  are  reduced  going  from  
PAM  to  PDM.    However,  since  the  PDM  pulses  do  not  occupy  the  entire  sample  interval,  higher  frequencies  and  
greater  bandwidth  are  required  than  for  the  equivalent  PAM  signal.      

  The  other  analog  pulse  modulation  scheme  which  can  be  used  as  an  alternative  to  PAM  is  called  Pulse  
Position  Modulation  (PPM).    In  this  technique,  the  position  of  a  narrow  pulse  of  uniform  amplitude  and  duration  
in  the  sample  interval  is  proportional  to  the  amplitude  of  the  signal.    This  approach  has  all  the  advantages  and  
drawbacks  of  PDM  with  the  additional  bonus  of  steady  transmitter  power  because  pulses  are  now  not  only  of  
uniform  amplitude  but  also  uniform  duration  as  well.    A  comparison  of  the  same  signal  sent  by  each  of  the  
modulation  schemes  is  illustrated  in  Figure  9-­‐3  below.  

PAM

10
9
8
7
6
5
4
3
2
1
0
1

11

21

31

41

51

61

71

81

91

108
PDM

1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
1

11

21

31

41

51

61

71

81

91
PPM

1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
1

11

21

31

41

51

61

71

81

91

Figure 0-3: Comparison of PAM, PDM & PPM Signals

If  the  signal  is  to  be  transmitted  in  digital  vice  analog  form,  the  signal  is  usually  left  in  PAM  format  for  the  next  
step  in  the  PCM  process:  quantization.  

9.2.4  Quantization  
  Each  PAM  level  must  be  rounded  off  to  the  nearest  discrete  quantization  level  to  continue  the  
transformation  of  an  analog  into  a  digital  signal.  This  is  because  the  amplitude  of  the  PAM  pulses  varies  
continuously  but  a  digital  representation  allows  for  only  a  finite  number  of  levels.  For  example,  if  3  bits  (1’s  and  
0’s)  are  used  in  the  digital  code,  then  only  8  different  levels  can  be  represented.  For  music  representation  on  
high  quality  CD’s,  the  number  of  levels  is 216 = 65536 .  

  The  exponent  16  is  the  number  of  bits  used  in  the  binary  code.  Another  application  might  not  require  
anywhere  near  that  many  bits  and  levels.  In  Figure  9-­‐2  eight  quantization  levels  are  included  and  shown  as  
dotted  horizontal  lines.  The  number  of  levels  used  is  equal  to  2  raised  to  an  integer  power.  This  exponent,  or  
power,  is  the  number  of  bits  of  the  corresponding  digital  code.  The  eight  quantization  levels  of  Figure  9-­‐2  can  be  
represented  by  the  eight  different  combinations  of  three  binary  bits.  At  each  sample  time  both  the  sample  value  
109
and  its  corresponding  quantization  level  are  shown  at  the  bottom  of  the  graph  in  Figure  9-­‐2.  Usually  the  levels  
are  spaced  uniformly.  The  Step  size  is  the  difference  between  adjacent  levels  and  is  given  by  the  Range  divided  
by  the  number  of  levels,  where  the  range  is  the  difference  between  the  maximum  and  minimum  analog  signal  
Range
values.    In  equation  form  this  becomes: stepsize =  where  n  is  the  number  of  bits.  
2n − 1

  Every  time  the  signal  is  quantized,  some  error  is  made,  though  it  may  be  small.  This  error  is  called  
quantization  error  and  can  be  made  smaller  by  increasing  the  number  of  bits  and  quantization  levels.  For  the  
scheme  illustrated  in  Figure  9-­‐2,  the  maximum  quantization  error  is  one  half  the  step-­‐size  since  the  nearest  
level  is  used  to  represent  the  sample  value.  In  other  schemes  the  nearest  larger  or  the  nearest  smaller  level  
might  always  be  used  which  would  imply  a  maximum  quantization  error  of  one  whole  step  size.  For  our  
example,  the  total  range  of  the  signal  is  from  -­‐4  V  to  +4  V  which  gives  a  Range  =  8  V.  There  are   23 = 8 levels  and  
steps.  This  gives  a  step  size  of  8V/7steps  =  1.14  V/step  and  a  maximum  Quantization  error  of  0.57  V.  By  
choosing  more  bits,  and  therefore  more  steps  and  levels,  this  error  can  be  made  as  small  as  we  like  but  at  the  
expense  of  more  circuit  complexity.  

9.2.5  Digital  Encoding  


  To  complete  the  conversion  process,  a  digital  code  is  generated  for  each  quantum  level.  In  our  example  
of  Figure  9-­‐2,  the  binary  number  000  is  generated  if  the  sample  value  is  rounded  off  to  3.5  V  and  111  is  
generated  if  the  roundoff  value  is  +3.5  V.  Each  intermediate  value  also  has  its  own  binary  code.  Pulses  are  
typically  then  used  to  represent  binary  1’s  and  0’s  electronically.  Because  there  are  8  levels  and  3  bits  for  the  
example  of  Figure  9-­‐2,  3  binary  pulses  are  generated  for  each  PAM  pulse.  This  is  illustrated  in  Figure  9-­‐4  where  
it  has  been  assumed  that  the  code  for  the  first  of  the  two  PAM  pulses  is  101.  For  three-­‐bit  conversion,  the  pulses  
representing  the  PCM  bit  stream  will  be  1/3  the  duration  of  the  PAM  pulses.  

Figure 0-4: PAM to PCM Time Change.

110
  This  means  the  PCM  bit  stream  will  require  about  3  times  the  BW  of  the  PAM  signal  or  in  the  more  
general  case  of  n-­‐bit  conversion,  n  times  the  PAM  bandwidth.  If  a  more  precise  conversion  using  more  bits  is  
required,  then  the  number  of  pulses  is  n  for  each  PAM  pulse,  where  n  is  the  number  of  bits.  For  16-­‐bit,  CD  music  
this  is  16  pulses  for  every  sample.  More  bits  implies  a  greater  baseband  bandwidth  requirement  for  the  digital  
PCM  signal.  The  pulses  will  be  transmitted  in  some  manner,  such  as  directly  over  wires,  or  by  modulation  onto  
an  RF  carrier  for  transmission  through  the  air,  perhaps  to  a  satellite.  In  either  case,  the  required  BW  must  be  
carefully  considered.  If  direct  transmission  over  wires  is  used,  then  these  wires  must  have  sufficient  bandwidth  
to  preserve  enough  of  the  pulse  shape  to  maintain  information,  and  if  modulation  onto  an  RF  carrier  is  used,  
then  the  RF  bandwidth  required  will  depend  on  the  baseband  BW  of  the  PCM  bit  stream.  Since  multiple  pulses  
are  being  generated  for  each  PAM  pulse,  their  duration  in  time  must  be  correspondingly  shorter.  PCM  requires  n  
times  more  BW  than  PAM.  In  equation  form,  BW  >  (number  of  bits)(fsamp1ing)  for  PCM.  The  increase  in  
required  bandwidth  is  the  price  to  be  paid  for  digital  transmission  of  information.  The  benefits  are  those  listed  
in  the  introduction  such  as  better  noise  immunity.  

Example 9.1

Determine  the  minimum  baseband  bandwidth  required  for  16  bit  CD  quality  music.  

Solution

Assuming  a  maximum  frequency  component  of  20  KHz  gives  a  sampling  rate  of  about  44  KHz  (slightly  greater  
than  the  theoretical  minimum).  With  16  bits  we  get    

BW  >  16  x  44  KHz  =  704  KHz,  a  very  large  baseband  bandwidth  compared  to  20  KHz  for  the  analog  signal  and  
this  is  for  only  one  channel.  If  we  want  stereo  and  we  time  multiplex  the  two  channels  together,  the  BW  
requirement  doubles.  

Example 9.2

Determine  the  maximum  quantization  error  for  conversion  of  music  into  16-­‐bit  PCM  form  if  the  input  analog  
signal  varies  over  the  range  of  -­‐1  V  to  +  1  V  and  the  conversion  takes  place  over  that  same  range  using  the  same  
scheme  as  illustrated  in  Figure  9-­‐2.  

Solution

The  Range  is  1  -­‐  (-­‐1)  =  2  V  and  the  number  of  steps  is   216 = 65536  giving  a  step-­‐size  of    2V/65535  =  30.51  mV.  
The  maximum  quantization  error  is  half  the  step-­‐size,  or  15.25  mV.  

Example 9.3

Determine  the  dynamic  range  expressed  in  dB  for  CD  music  recordings.  Refer  to  Figure  9-­‐2.  

Solution

The  dynamic  range  is  defined  as  the  ratio  of  the  greatest  possible  change  in  amplitude  to  the  smallest.  The  total  
range  (8  V  in  the  case  of  Figure  9-­‐2)  is  the  largest  possible  change  in  the  signal  amplitude.  The  smallest  possible  
change  is  the  difference  between  adjacent  levels.  The  number  of  steps  is   2n  and  the  size  of  one  step  is  

111
Range
stepsize = n
 (in  the  case  of  Figure  9-­‐2,  the  step-­‐size  is  1.14  V).  Taking  the  ratio  we  get 2n − 1,  and  
2 −1
n
20 × log(2 − 1)  is  96.3  dB  for  16  bits.  Not  accidentally,  this  is  very  close  to  the  dynamic  range  from  a  whisper  to  
the  threshold  of  pain  for  the  human  ear.  Adding  more  than  16  bits  for  CD  recordings  would  not  improve  the  
quality.  

  The  PCM  baseband  signal  is  modulated  onto  an  RF  carrier  in  some  applications.  One  application  for  
which  modulation  onto  an  RF  carrier  is  required  is  the  digital  up  link  or  down  link  to  a  communications  
satellite.  A  variety  of  modulation  techniques  can  be  used.  Three  are  amplitude,  frequency  and  phase  modulation.  
Because  there  are  only  two  types  of  symbols  to  be  transmitted,  l’s  and  0’s,  only  two  values  of  amplitude  are  
required  for  AM,  only  two  different  carrier  frequencies  are  required  for  FM  or  two  phases  for  PM.  The  AM  
technique  is  called  ASK  for  amplitude  shift  keying,  the  FM  technique  is  called  FSK  frequency  shift  keying  and  the  
PM  technique  is  called  PSK  for  Phase  Shift  Keying.  An  example  of  each  is  shown  in  Figure  9-­‐5.  

Amplitude Shift Keying (ASK)


Digital Baseband
1.5
1.2
1
1

0.8 0.5
Voltage

0.6 Voltage
0
1 11 21 31 41 51 61 71 81 91 101
0.4
-0.5
0.2
-1
0
1 11 21 31 41 51 61 71 81 91 101
-1.5
Time
Time

Frequency Shift Keying (FSK) Phase Shift Keying (PSK)

1.5 1.5

1 1

0.5 0.5
Voltage

Voltage

0 0
1 11 21 31 41 51 61 71 81 91 101 1 11 21 31 41 51 61 71 81 91 101
-0.5 -0.5

-1 -1

-1.5 -1.5
Time Time

Figure 0-5: Digital Modulation of RF Carriers.

Example 9.4

A  baseband  analog  signal  comprising  frequency  components  from  0  to  10  KHz  is  to  be  sampled  and  then  
converted  to  an  8  bit  PCM  signal.  Determine  the  minimum  sampling  rate  and  the  minimum  bandwidth  of  the  
resulting  baseband  PCM  bit  stream.  What  is  the  minimum  RF  bandwidth  if  this  bit  stream  is  modulated  onto  an  
RF  carrier?  

Solution

112
The  theoretical  minimum  sampling  rate  is  twice  the  maximum  frequency  content  of  the  analog  signal.  In  this  
case  twice  10  KHz  or f sampling = 20  KHz.  A  more  practical  and  achievable  rate  would  be  about  25  KHz.  The  
minimum  required  bandwidth  for  the  resulting  PCM  bit  stream  is  equal  to  the  number  of  bits  times  the  
sampling  rate  which  gives  BW>  8(20  KHz)  =  160  KHz,  or  8(25  KHz)  =  200  KHz  for  the  more  practical  rate.  The  
RF  bandwidth  will  be  a  minimum  if  SSB  AM  is  used.  In  this  case  the  RF  bandwidth  is  the  same  as  the  PCM  
baseband  bandwidth.  An  FM  technique  would  require  much  more  bandwidth.  Because  the  baseband  BW  of  the  
PCM  bit  stream  is  often  very  large  compared  to  the  analog  BW,  the  RF  carriers  are  often  chosen  to  be  relatively  
high  to  give  enough  room  in  the  spectrum  for  frequency  division  multiplexing.  Satellite  communications  are  in  
the  gigahertz  range  ( 109 Hz).  

  We  have  covered  the  processes  involved  in  converting  an  analog  signal  into  its  PCM  counterpart.  All  of  
these  processes  are  combined  together  in  an  analog-­‐to-­‐digital  converter  (ADC).  ADC’s  are  fabricated  in  
integrated  circuit  form,  often  on  board  a  computer,  and  are  fast,  reliable  and  inexpensive,  although,  very  high  
conversion  speed  can  cost  a  lot.  

9.3  Digital  Receivers  


  If  the  digital  baseband  is  modulated  onto  an  RF  carrier  by  an  AM  or  FM  technique,  then  a  receiver  must  
first  capture  the  signal  with  an  antenna,  amplify  it,  and  then  demodulate  the  digital  bit  stream  back  off  the  
carrier.  The  specific  type  of  demodulator  depends  upon  the  type  of  modulation  used  at  the  transmitter,  but  the  
RF  receiver  will  likely  look  very  much  like  an  AM  or  FM  superheterodyne  receiver.  If  the  digital  bit  stream  is  
transmitted  directly  without  any  RF  modulation,  then  no  RF  demodulation  is  necessary.  In  some  systems  PCM  
pulses  are  sent  directly  over  wires,  coaxial  cable,  or  an  optical  fiber  using  light.  Once  the  digital  bit  stream  has  
been  retrieved,  it  can  be  stored  or  converted  back  to  the  original  analog  signal.  Storage  media  include  magnetic  
tape  and  computer  floppy  and  hard  drives.  Conversion  of  the  PCM  signal  back  to  its  analog  counterpart  can  be  
done  with  a  digital-­‐to-­‐analog  conversion  circuit  (DAC).  Discussion  of  this  type  of  circuit  is  included  elsewhere  in  
this  course.  DAC’s  are  typically  fabricated  in  integrated  circuit  form  and  are  fast,  reliable  and  relatively  
inexpensive.  Many  computers  come  with  one  or  more  DAC’s  built  in.  The  MIDI  sound  interface  in  PCs  is  an  
example.  The  output  of  a  DAC  will  not  be  perfectly  smooth.  A  sample  DAC  output  is  shown  below  in  Figure  9-­‐6.  

Figure 0-6: Digital-to-Analog Converter Output.

  It  will  be  a  staircase  approximation  to  the  original  analog  signal  with  a  step  size  equal  to  that  used  in  the  
PCM  conversion  at  the  transmitter.  If  the  number  of  bits  is  high,  then  the  step  size  will  be  very  small  and  the  
113
approximation  to  the  original  analog  signal  very  good.  The  jagged  edge  on  the  DAC  output  can  be  low  pass  
filtered  for  smoothing.    

  A  major  advantage  of  digital  communications  is  the  ability  of  a  digital  receiver  to  reject  noise.  Digital  
receivers  can  process  digital  data  to  remove  the  effects  of  noise.  At  the  time  of  arrival  of  each  bit,  the  receiver  
has  to  decide  if  the  bit  is  a  1  or  a  0.  This  can  be  done  accurately  in  the  presence  of  a  moderate  amount  of  noise  as  
long  as  the  noise  is  not  so  great  as  to  make  a  1  look  like  a  0  or  vice  versa.  A  similar  amount  of  noise  would  be  a  
big  problem  for  an  analog  signal  because  the  noise  adds  directly  to  the  analog  value.  An  illustration  of  a  0  and  a  
l,  first  without  and  then  with  added  noise,  is  shown  below  in  Figure  9-­‐7..  

  One  way  for  a  receiver  to  determine  the  presence  of  a  1  or  a  0  is  to  sample  at  some  point  in  the  bit  
interval  and  compare  to  a  threshold.  The  center  of  each  bit  interval  is  a  convenient  choice  of  sample  time  and  
half  way  between  the  voltage  levels  for  a  1  and  a  0  is  a  good  choice  for  the  threshold.  In  this  way,  the  correct  
decision  will  always  be  made  unless  the  noise  exceeds  half  the  difference  between  a  1  and  a  0  at  the  midpoint  of  
a  bit.  The  1  and  the  0  are  represented  by  A  Volts  and  0  Volts,  respectively  in  Figure  9-­‐7.  Other  choices  of  voltage  
assignment  are  possible,  but  the  idea  remains  the  same.  The  average  value  of  many  types  of  noise  is  zero.  
Because  of  this  fact  a  further  improvement  in  noise  rejection  is  possible.  If  the  receiver  averages  the  signal  over  
each  bit  interval  the  chances  of  correctly  identifying  l’s  and  0’s  is  further  increased.  The  noise  tends  to  average  
out  and  leave  only  the  signal  due  to  the  bit  value.  

Figure 0-7: Two Bits of a Signal, with and without Noise.

9.4  Error  Detection  and  Correction  


  Even  with  the  relative  noise  immunity  which  a  digital  format  offers,  sometimes  mistakes  in  identifying  
1’s  and  0’s  do  occur.    These  bit  errors  are  the  digital  equivalent  of  noise  and  the  number  of  errors  per  bits  
transmitted  or  “bit  error  rate”  is  analogous  to  the  inverse  of  analog  SNR.    Just  as  analog  circuits  use  filtering  to  
reduce  noise  effects,  digital  signals  add  redundancy  to  the  message  signal  which  creates  a  conflict  when  an  error  

114
is  present  in  either  the  original  message  or  the  redundancy.    This  redundancy  can  be  used  for  one  of  two  general  
approaches:  simple  error  detection  or  forward  error  correction.  

1.  Simple  error  detection:    For  simple  error  detection,  the  basic  approach  is  to  send  an  automatic  request  for  
retransmission  (ARQ)  of  the  block  of  data  in  which  an  error  is  detected.    Since  the  error  detection  code  is  
providing  a  relatively  simple  function  it  can  be  relatively  short  resulting  in  few  additional  bits  being  transmitted  
unless  an  error  is  found.    Disadvantages  of  this  approach  are  that  the  entire  block  containing  the  error  must  be  
retransmitted  and  the  receiver  must  transmit  back  to  the  originator  of  the  message  to  get  the  corrected  data  
block.    If  the  receiver  cannot  transmit  and  errors  cannot  be  tolerated,  forward  error  correction  is  required.  
There  are  several  techniques  used  to  implement  error  detection:  

 
  a)  Encoding  methods:  The  format  of  the  transmitted  waveshape  can  help  to  identify  bit  errors,  such  as  a  
  “bipolar  return  to  zero  alternate  mark  inversion:  encoding  where  successive  “ones”  in  the  sequence  
  should  have  alternate  polarities  (+/-­‐5V)  and  “zeros”  are  at  0V.    If  two  successive  “ones”  are  received  
  with  the  same  polarity,  one  is  not  a  “one”  or  a  “one”  was  missed  between  the  two  received.  

  b)  Redundant  transmission:  This  is  conceptually  the  most  simple  means  of  error  detection.    Each  block  
  of  data  is  transmitted  twice.    At  the  receiver,  the  blocks  are  compared  and  if  found  different,  
  retransmission  is  requested.    The  primary  disadvantage  of  this  technique  is  that  is  sends,  at  best,  50%  
  new  information  in  each  transmission,  and  less,  if  errors  are  actually  detected.    There  are  more  efficient  
  methods.  
 
  c)  Parity  check:  This  method  of  error  detection  is  both  simple  and  efficient.    It  works  by  simply  totaling  
  the  number  of  “1”s  in  a  data  block  and  adding  a  bit  which  makes  the  total,  including  the  parity  bit,  the  
  designated  parity.    For  example,  in  an  even  parity  system,  if  the  seven  bit  sequence,  “1011101”,  is  to  be  
  sent,  an  eighth  parity  bit  of  “1”  is  appended  to  the  sequence  to  make  the  total  number  of  “1”s  in  the  
  sequence  even.    At  the  receiver,  the  eight  bits  are  summed.    If  the  result  is  an  odd  total,  an  error  has  been  
  detected  and  the  block  is  requested  to  be  resent.    At  low  bit  error  rates  (number  of  incorrect  bits  
  received  per  total  number  of  bits  received),  efficiency  can  be  improved  with  longer  sequences  before  
  appending  the  parity  bit;  however,  then  a  longer  block  must  be  resent  if  an  error  is  detected.    One  
  disadvantage  of  this  relatively  simple  approach  is  that  is  only  detects  an  odd  number  of  errors  in  a  
  sequence.    To  detect  any  number  of  errors  in  a  sequence,  a  more  complex  method  is  required.  

 
  d)  Cyclic  redundancy  check/checksum:  One  of  the  most  effective  and  efficient  means  to  detect  multiple  
  errors  is  called  cyclic  redundancy  check,  which  treats  the  entire  data  block  as  one  long  binary  number,  
  divides  it  by  a  pre-­‐selected  fixed  constant  and  transmits  the  remainder  after  the  division  along  with  the  
  message.    The  receiver  re-­‐performs  the  division  on  the  data  block  at  the  destination  using  the  same  fixed  
  constant  and  compares  the  received  and  calculated  remainders.    In  checksum,  several  blocks  or  sub-­‐
  blocks  of  data  are  added  together  then  the  cyclic  redundancy  check  is  performed  on  the  result.    The  
  likelihood  of  one  or  more  errors  creating  the  same  remainder  is  extremely  remote,  particularly  with  a  
  judicious  selection  of  the  pre-­‐selected  constant.    The  primary  disadvantage  of  this  method  is  that  it  is  
  somewhat  calculation  intensive.  
 

115
2.  Forward  error  correction  (FEC):    This  approach  not  only  identifies  that  there  is  an  error,  but  specifies  the  
location  of  the  error  to  the  bit,  which  can  simply  be  complemented  to  the  correct  value  at  the  receiver  without  
requesting  a  retransmission  of  any  data.    Several  methods  of  FEC  are  simply  extensions  of  simple  error  
detection  methods.    However,  they  require  additional  redundant  bits  to  be  transmitted  which  make  them  less  
efficient  from  an  information  rate  standpoint.  Like  error  detection,  error  correction  can  be  accomplished  by  one  
of  several  techniques:  

  a)  Redundant  transmission:  By  transmitting  each  data  block  three  times,  not  only  are  incorrect  bits  
  identified,  but  by  a  2/3rds  vote,  the  correct  bit  state  is  determined.    Weaknesses  of  this  approach  
  include  at  best  one  third  new  information  in  each  data  block  sent  (i.e.  2/3rds  of  the  bandwidth  is  
  “wasted”  on  error  correction)  and  the  nonzero  probability  that  the  same  bit  could  be  in  error  in  two  of  
  the  three  transmissions.    However,  this  technique  detects  and  corrects  multiple  errors  more  simply  than  
  most  other  methods.    

 
  b)  Block  check  character/longitudinal  redundancy  check/horizontal  redundancy  check:  This  technique  
  is  a  direct  extension  of  the  parity  check  best  illustrated  by  an  example.      
  Given  an  odd  parity  system  and  seven  7  bit  sequences,  arrange  them  in  a  7×7  block.    Establish  an  eighth  
  column  populated  with  the  parity  bits  for  each  row  and  an  eighth  row  populated  with  the  parity  bits  for  
  each  column.    (The  eighth  row  and  column  position  can  be  a  parity  bit  for  the  row  parity  bits,  the  
  column  parity  bits  or  their  sum.)  
 

D a t a Row

Parity Bits

D 1 0 0 1 1 1 0 1

a 0 0 1 0 0 0 1 1

t 0 1 1 1 1 0 1 0

a 1 1 0 1 0 0 0 0

0 0 1 0 1 1 1 1

1 0 0 1 0 1 0 0

1 1 0 0 1 1 0 1

Column 1 0 0 1 1 1 0 1

Parity Bits

 
  As  should  be  relatively  apparent,  any  single  bit  error  will  create  a  parity  discrepancy  in  both  its  row  and  
column,  positively  identifying  the  incorrect  bit  for  correction.    Any  multiple  bit  errors  will  generate  ambiguity  as  
to  the  location  of  the  errors  and  will  require  retransmission  of  the  entire  block.    If  two  errors  occur  in  the  same  
row,  the  appropriate  columns  will  show  the  discrepancies,  but  all  the  row  parity  bits  will  appear  correct.    If  the  
116
two  errors  happen  to  be  in  different  rows  and  columns,  then  the  ambiguity  consists  of  inability  to  determine  
which  of  the  discrepant  row  and  column  pairs  correspond  to  the  incorrect  bits  since  two  rows  and  two  columns  
identify  four,  not  two,  possible  error  locations.  
  c)  Hamming/Reed-­‐Solomon  Codes:  Hamming  codes  are  cleverly  designed  to  not  only  identify  that  a  bit  
  was  transmitted  in  error  but  also  specify  its  location.  Operation  of  the  Hamming  Code  is  best  illustrated  
  by  an  example:      

  The  number  of  “Hamming  bits”  required  to  be  added  to  the  information  bits  is  determined  by:  

2n≥m+n+1 (7.1)

Where:  
n=  #  of  Hamming  bits  
m=#  of  information  bits  
 
  For  this  example,  10  information  bits  (1011001001)  are  to  be  transmitted  and  the  Hamming  bits  are  to  
be  in  the  LSB  positions  evenly  divisible  by  three.  To  determine  how  many  Hamming  bits  are  required,  
2n≥m+n+1=10+n+1  

 
  If  n=4,  the  inequality  works,  so  4  Hamming  bits  are  required  in  positions  3,  6,  9,  and  12.    (Hamming  bits  
are  typically  distributed  among  the  information  bits  to  reduce  the  likelihood  that  all  the  Hamming  bits  could  be  
garbled  at  once.)    The  sequence  will  take  the  following  form:  
 

Bit Position 14 13 12 11 10 9 8 7 6 5 4 3 2 1

Contents 1 0 H 1 1 H 0 0 H 1 0 H 0 1

 
The  Hamming  bits  are  determined  by  XORing  the  position  numbers  corresponding  to  the  ones  in  the  
information  bits:  
 
Position     Binary  Code  
1       0001  
5       0101  
10       1010  
11       1011  
14       1110  
Hamming  bits:     1011  
 
The  final  transmitted  sequence  is:  
 

Bit Position 14 13 12 11 10 9 8 7 6 5 4 3 2 1

Contents 1 0 1 1 1 0 0 0 1 1 0 1 0 1

117
If  the  bits  are  transmitted  without  error,  the  receiver  extracts  the  bits  from  the  designated  Hamming  bits  
positions  to  XOR  them  with  the  position  numbers  corresponding  to  the  ones  in  the  information  bits  and  confirm  
that  there  are  no  errors:  
 
Position     Binary  Code  
Hamming  bits:     1011  
1       0001  
5       0101  
10       1010  
11       1011  
14       1110  
Error  Position:     0000   (No  errors)  
 
If,  on  the  other  hand,  one  bit  is  in  error,  say  in  position  4,  the  received  sequence  would  be:  
 

Bit Position 14 13 12 11 10 9 8 7 6 5 4 3 2 1

Contents 1 0 1 1 1 0 0 0 1 1 1 1 0 1

 
Repeating  the  extraction  and  XORing  process  produces:  
 

Position     Binary  Code  


Hamming  bits:     1011  
1       0001  
4       0100  
5       0101  
10       1010  
11       1011  
14       1110  
Error  Position:     0100   (Error  in  position  4)  
 
The  bit  in  position  4  is  complemented  and  the  error  free  sequence  is  then  processed.      
 
More  complex  Hamming  and  Reed-­‐Solomon  codes  can  detect  and  correct  multiple  errors,  including  in  the  coded  
bits  themselves,  but  those  codes  are  beyond  the  scope  of  this  course.  

9.5  Channel  Capacity  


  The  entire  purpose  for  any  communication  system  is  to  get  original  information  from  one  point  to  
another.    There  are  a  number  of  factors  which  limit  how  fast  this  information  can  be  transferred.    Similar  to  
analog  systems,  noise  and  bandwidth  play  their  parts.    Unique  to  digital  systems,  “overhead”  and  encoding  also  
serve  to  limit  information  rate.    We  will  discuss  each  of  these  factors  in  turn.  

118
  First,  let’s  start  off  with  some  definitions.    Information  is  the  “intelligence”  which  is  desired  to  be  
transferred  from  one  location  to  another  and  is  the  reason  for  the  transmission  in  the  first  place.    Ideally,  it  is  
completely  unpredictable  such  that  one  part  of  the  information  message  tells  nothing  about  any  other  part.    In  
binary  digital  systems,  the  unit  of  information  is  called  a  bit,  a  “1”  or  a  “0”.      

  Data  is  different  from  information  in  that  it  includes  “overhead”  in  addition  to  the  baseline  information.    
Such  overhead  can  include  “start”  and  “stop”  bits  to  delineate  data  blocks/frames,  error  detection/correction  
codes,  encryption  codes,  routing  instructions,  frame  reassembly  directions,  frames  re-­‐transmitted  due  to  errors  
or  omissions,  etc.    The  units  for  data  are  exactly  the  same  as  for  information;  however,  no  real  digital  
communication  system  can  achieve  a  data  rate  equal  to  its  information  rate,  because  all  real  systems  have  to  
include  some  overhead.    In  other  words,  data  equals  information  plus  overhead.      

  For  the  remainder  of  this  chapter,  we  will  discuss  data  and  define  Channel  Capacity  in  terms  of  data  rate.    
Some  texts  will  use  the  terms  information  rate  and  data  rate  interchangeably,  but  you  should  recognize  that  the  
information  rate  you  can  push  through  a  digital  communication  system  will  never  reach,  even  under  ideal  
conditions,  the  advertised  Channel  Capacity,  a  data  rate  which  includes  overhead  added  and  needed  by  the  
system.  The  data  rate  which  a  digital  channel  can  carry  is  determined  by  channel  bandwidth  in  a  manner  similar  
to  an  analog  system.    Per  Hartley’s  Law,  given  sufficient  time  and/or  channel  bandwidth,  any  quantity  of  data  
can  eventually  be  transferred:  

I=ktB (9.2)

Where:  
I=amount  of  data  to  be  sent  [bits]  
k=a  constant  [bits/cycle]  
t=transmission  time  [sec]  
B=baseband  channel  bandwidth  [Hz]  
 

Since  infinite  time  or  bandwidth  is  rarely  available,  a  typically  more  useful  concept  is  the  rate  at  which  
information  can  be  sent  over  a  channel  also  known  as  “channel  data  rate”:

R=I/t (9.3)
 
Where:  
R=data  rate  [bits/sec]  
I=amount  of  information  to  be  sent  [bits]  
t=transmission  time  [sec]  
 

Rewriting  Hartley’s  Law  with  this  new  variable:  

R=kB (9.4)

Where:  
R=data  rate  [bits/sec]  
k=a  constant  [bits/cycle]  
B=baseband  channel  bandwidth  [Hz]  

119
  Since  channel  bandwidth  is  defined  in  the  same  way  as  for  analog  signals,  the  remaining  mystery  
variable  in  this  equation  is  “k”.    Through  the  Shannon  Limit,  in  a  derivation  beyond  the  scope  of  this  course,  the  
maximum  practical  value  of  “k”  was  determined  to  be  a  function  of  the  channel  signal  to  noise  ratio  (in  rational,  
not  dB  form):

Rmax=C=B×log2(1+S/N) (9.5)

Where:  
R=data  rate  [bits/sec]  
C=Channel  data  rate  Capacity  [bits/sec]  
B=baseband  channel  bandwidth  [Hz]  
S/N=signal-­‐to-­‐noise  power  ratio  (not  in  dB)  
  (S/N=10SNRdB/10)  
 

  Thus,  the  channel  capacity  is  determined  by  the  bandwidth  available  and  the  SNR  experienced  in  the  
same  way  the  analog  channels  require  sufficient  bandwidth  and  adequate  noise  margin  to  reproduce  the  
original  signal  at  the  destination  with  the  specified  fidelity.    While  the  bandwidth  is  typically  fixed  at  system  
design,  the  SNR  can  varying  greatly  during  operation  based  on  environmental  factors.    Therefore,  during  initial  
design,  a  minimum  expected  SNR  is  set  and  used  to  determine  the  channel  capacity.    If  the  SNR  experienced  in  
operation  falls  below  this  assumed  value,  the  channel  capacity  falls,  the  error  rate  in  the  channel  climbs  quickly  
and  the  system  performance  degrades  rapidly.  

Since  SNRs  in  digital  channels  can  be  quite  high,  how  do  we  design  our  data  stream  to  take  advantage  of  as  
much  of  the  channel  capacity  as  possible?    Let’s  go  back  to  our  variation  of  Hartley’s  Law:

R=kB (9.4)

Where:  
R=data  rate  [bits/sec]  
k=a  constant  [bits/cycle]  
B=baseband  channel  bandwidth  [Hz]  
 

We  need  to  redefine  “k”  in  terms  of  the  transmitted  data  stream.    We’ll  start  with  the  simplest  case:  a  straight  
binary  signal  transmitted  in  a  sequence  requiring  the  greatest  possible  channel  bandwidth.    The  sequence  
requiring  the  greatest  bandwidth  (having  the  highest  frequency  components)  would  be  one  which  simply  
alternated  high  and  low  (since  any  consecutive  matching  bits  would  change  less  frequently  requiring  less  
bandwidth).    Figure  9-­‐8  illustrates  this  limiting  condition.  

120
Bit Transfer at Channel Bandwidth Limit
Binary Signal at
Bandwidth Limit
1 Original Binary Signal
0.8
0.6
0.4
0.2
Voltage

0
0

10

20

30

40
-0.2
-0.4
-0.6
-0.8
-1
Time

Figure 0-8: Bit Transfer at Channel Bandwidth Limit

  Note  that  one  high  bit  and  one  low  bit,  for  two  bits  total,  can  be  transmitted  per  cycle  of  the  bandwidth  
limited  signal.    Thus,  for  this  situation,  k=2  bits/cycle.      

  This  simple  example  can  be  generalized  by  acknowledging  that,  under  certain  conditions,  the  signal  is  
not  limited  to  only  two  distinct  symbols  per  a  half  cycle,  but  “N”  different  states  per  symbol.    For  the  generalized  
case,  “k”  [bits/cycle]  is  broken  into  two  factors:  bits/symbol  and  symbols/cycle.    Since  N  different  symbols  can  
be  represented  by  n  bits  (N=2n),  then:

n=bits/symbol=log2N (9.6)

This  results  in  two  different,  but  equivalent  forms  of  the  Shannon-­‐Hartley  Theorem:  

R=S×log2N =2B×log2N (9.7)

Where:  
R=data  rate  [bits/sec]  
S=baud  rate  [symbols/sec]  
N=number  of  possible  states  per  symbol  (If  N=2  {binary},  R=2B)  
2  symbols/cycle    
B=baseband  channel  bandwidth  [Hz]  
 

121
  Note  that  the  baud  rate  in  symbols  per  second  is  different  than  the  channel  data  rate  in  bits  per  second  
because  each  transmitted  symbol  can  represent  more  than  one  bit  of  information.    Also  note  that  these  
parameters  are  set  at  system  design  and  do  not  change  during  operation.    Once  the  transmitter  and  receiver  are  
built  to  exchange  a  four  level  signal,  they  do  not  reconfigure  to  an  eight  level  signal  just  because  the  noise  level  
drops  to  permit  an  increased  theoretical  channel  capacity.  

  There  are  some  assumptions  inherent  in  this  form  of  the  Shannon-­‐Hartley  Theorem  that  impact  its  
application.    First,  the  baseband  bandwidth  is  assumed  to  be  perfectly  rectangular.    Since  real  filters  have  “roll-­‐
off”,  additional  bandwidth  is  required  to  send  a  given  data  rate,  or  conversely,  real  filters  of  a  given  bandwidth  
result  in  lower  than  ideal  (Shannon-­‐Hartley  Theorem)  data  rates.    Additionally,  if  the  baseband  signal  
modulates  a  carrier  using  other  than  AM  Single  Side  Band  (SSB),  the  required  passband  is  twice  (or  greater  for  
FM,  PM)  the  width  of  the  baseband  signal  for  a  given  data  rate.    Correspondingly,  the  data  rate  for  a  given  
modulated  passband  bandwidth  is  one  half  or  less  than  that  indicated  by  using  this  bandwidth  for  “B”  in  the  
Shannon-­‐Hartley  Theorem.  

  So,  how  do  we  generate  “N”  different  states  per  symbol  in  a  digital  transmission  line?    The  secret  is  in  
recognizing  that  “digital”  is  not  limited  to  “binary”.    Instead  of  being  limited  to  just  two  distinct  amplitudes,  
frequencies  or  phases  as  illustrated  in  Figure  9-­‐5,  multiple  amplitudes,  frequencies  or  phases  are  permitted  as  
long  as  they  can  be  recognized  as  different  symbols  at  the  receiver.    There  is  also  no  prohibition  regarding  
varying  both  amplitude  and  either  frequency  or  phase  to  create  even  more  individual  symbols.    (Because  both  
frequency  and  phase  are  angle  modulations,  they  cannot  generally  be  varied  independently.)    In  fact,  a  very  
popular  digital  modulation  format  is  called  Quadrature  Amplitude  Modulation  (QAM)  which  varies  both  
amplitude  and  phase  to  send  multiple  bits  of  information  per  symbol.    An  8-­‐QAM  signal  is  plotted  in  Figure  9-­‐9.  

Quadrature Amplitude Modulation (QAM)

1.5

0.5
Voltage

0
0

-0.5

-1

-1.5

-2
Tim e

Figure 0-9:A Sample QAM Signal

Let’s  see  how  all  this  would  work  together  to  determine  and  best  use  the  capacity  of  a  channel.    First,  determine  
the  minimum  baseband  bandwidth  and  signal  to  noise  ratio  expected  on  the  channel  over  which  the  digital  data  
122
will  be  transferred.    These  establish  the  maximum  channel  data  rate  capacity  using  the  Shannon  Limit.    Next,  
use  this  channel  data  rate,  baseband  bandwidth  and  any  correction  for  band  “roll-­‐off”  in  the  Shannon-­‐Hartley  
Theorem  to  determine  the  whole  number  of  different  symbols  which  are  needed  to  achieve  this  data  rate.    
(Round  down.)    This  is  the  number  of  symbols  designed  into  the  system  modulation  scheme  and,  when  worked  
back  through  the  Shannon-­‐Hartley  Theorem,  gives  the  design  channel  data  rate.    To  illustrate,  let’s  look  at  an  
example:      

Example 9.5

A  digital  channel  has  a  baseband  bandwidth  of  20KHz  and  a  minimum  expected  SNR  of  50dB.    The  digital  
encoding  scheme  is  expected  to  be  QAM  with  bits/symbol  equal  to  an  even  power  of  2.    Assuming  no  filter  roll-­‐
off  corrections  are  required,  determine  the  maximum  channel  capacity  based  on  bandwidth  and  noise,  the  
design  form  of  QAM  to  be  used  (8-­‐QAM  [3  bits/symbol],  64-­‐QAM  [6  bits/symbol],  etc.)  and  the  expected  
maximum  channel  data  rate.  

Solution

SNR  (rational  form)=1050dB/10dB=105  

From  equation  (7.5):  C=B×log2(1+S/N)=20KHz×log2(100,001)bits/cycle=332Kbps  

From  equation  (7.7):  R=2B×log2N=332Kbps=2×20KHz×log2N;  

Nmax=28.3=315.2  states/symbol;    

Ndesign=28=256  states/symbol  =>  256-­‐QAM  [8  bits/symbol]  

From  equation  (7.7):  Rmax=2B×log2N=2×20KHz×log2256=320Kbps  

9.6  Time  Division  Multiplexing  


  AM  and  FM  radio  stations  are  frequency  division  multiplexed  so  that  multiple  stations  can  use  the  same  
transmission  medium.  This  is  done  by  assigning  each  station  a  different  carrier  and  limiting  the  side  band  
content  of  each  station  so  that  it  does  not  overlap  with  its  neighbors.  Because  radio  stations  are  frequency  
division  multiplexed,  receivers  can  tune  to  individual  stations  by  using  filters.  PCM  digital  signal  format  allows  a  
different  kind  of  multiplexing  called  time  division  multiplexing  (TDM).  Time  division  multiplexing  is  important  
because  it  allows  the  simultaneous  use  of  a  common  resource,  such  as  a  telephone  line  or  a  satellite  
communications  link,  by  multiple  digital  signals.  The  total  number  of  channels  multiplexed  together  can  number  
in  the  100’s  or  even  1000’s.  A  diagram  of  a  TDM  system  is  shown  in  Figure  9-­‐10.  Two  PCM  signals  are  shown  
time  multiplexed  together.  This  is  done  by  first  sending  a  bit  from  Signal  A  then  a  bit  from  Signal  B  and  then  
back  to  A  etc.  

  In  this  way  bits  from  each  signal  become  interleaved  as  they  flow  from  the  output  of  the  multiplexer  and  
into  the  transmission  line.  The  bit  format  chosen  for  Figure  9-­‐10  is  called  return  to  zero  with  a  digital  zero  
represented  by  zero  volts.  There  are  many  other  formats  possible,  some  of  which  are  more  efficient  in  one  way  
or  another,  but  the  choice  for  this  figure  is  convenient  for  illustration.  

123
Figure 0-10:Time Division Multiplexing of Two PCM Signals.

  The  heart  of  the  multiplexing  process  is  a  commutator  which  is  shown  in  the  figure  as  a  rotating  
mechanical  switch.  As  this  switch  rotates  around,  it  first  contacts  Signal  A  then  signal  B  and  so  on.  The  
commutator  must  complete  one  revolution  during  one  bit  interval  of  each  input  so  that  no  information  is  lost  
from  either  signal.  This  means  that  the  commutator  frequency  is  the  same  as  the  bit  rate  of  each  of  the  signals,  
which  are  assumed  to  be  the  same.  To  be  practical,  the  commutator  switch  must  be  electronic  rather  than  
mechanical  to  be  fast  enough  to  keep  up.  This  type  of  switching  is  no  problem  using  modern  electronics.  

  During  one  bit  interval  of  either  input,  two  bits  must  flow  from  the  output  of  the  multiplexer,  one  for  
signal  A  and  one  for  B.  Thus,  each  bit  at  the  output  of  the  multiplexer  is  only  half  as  long  as  the  bits  at  the  inputs.  
Another  way  of  stating  this  is  that  at  the  output,  transmission  bit  rate  is  double  the  bit  rate  of  either  of  the  
inputs.  This  means  that  the  resulting  bandwidth  of  the  multiplexer  output  is  also  double  the  bandwidth  of  either  
input  bit  stream.  Typically,  many  more  than  two  PCM  signals  are  multiplexed  together.  For  N  signals  
multiplexed  together,  there  would  be  N  commutator  segments  and  the  bit  rate  at  the  output  of  the  multiplexer  
would  be  n  times  the  rate  of  any  of  the  inputs.  The  required  transmission  bandwidth  would  then  be  N  times  that  
of  anyone  of  the  input  signals.  

  At  the  receiver  there  is  a  similar  commutator  to  unshuffle  the  bits  interleaved  together  at  the  
transmitter.  To  do  this  correctly,  the  two  commutators  must  switch  at  the  same  frequency  and  be  perfectly  
synchronized.  If  not,  the  information  will  become  garbled  or  sent  to  the  wrong  destinations.  At  the  output  of  the  
receiver  commutator,  the  bit  streams  are  routed  to  their  intended  destinations  where  they  can  be  stored  or  
converted  back  into  analog  form  by  DAC’s.  

  The  inputs  to  the  time  multiplexer  discussed  above  are  digital  signals.  The  inputs  to  a  time  multiplexer  
can  also  be  analog,  in  which  case  the  multiplexer  both  samples  and  multiplexes.  In  this  case,  the  output  will  be  
time  multiplexed  PAM  pulses.  The  multiplexed  PAM  pulses  can  then  be  sent  over  a  transmission  line  with  no  
further  processing  or  they  can  be  converted  to  PCM  form  and  transmitted  as  a  bit  stream.  

Example 9.6

Ten  analog  signals  are  to  be  converted  to  10-­‐bit  digital  PCM  form  and  then  time  division  multiplexed  for  
transmission  over  a  common  transmission  line.  Each  analog  signal  is  band-­‐limited  from  0  to  8  KHz.  Determine  
124
the  bit  rate  for  each  input  channel,  the  frequency  of  rotation  of  the  commutator  switch  and  the  minimum  
bandwidth  requirement  for  the  transmission  line.  

Solution

From  the  sampling  theorem,  each  signal  must  first  be  sampled  at  a  minimum  of  2  x  8  =  16  KHz  and  so  we  will  
pick  a  sampling  rate  of  20  KHz  to  be  conservative.  10  bits  will  be  generated  for  each  signal,  which  gives  a  bit  
rate  of  l0x20  =  200  Kbits/s  for  each  input.  The  required  frequency  of  rotation  of  the  commutator  is  the  same  as  
the  bit  rate  of  anyone  input  channel,  so  this  becomes   f commutator = 200 KHz.  The  required  baseband  bandwidth  
for  a  single  channel  is  the  same  as  the  PCM  bit  rate  which  is  200  KHz.  For  10  channels,  the  requirement  will  be  
10  times  as  much,  or   BWoutput = 10 × 200 = 2000 KHz  =  2  MHz.  

125
9.7  Homework  Problems  

Problem  9.l  

The  sampling  rate  used  in  the  conversion  of  an  analog  signal  to  PCM  is  20  KHz.  The  analog  signal  contains  
components  of  high  enough  frequency  to  cause  aliasing  problems.  To  avoid  aliasing,  the  analog  signal  is  low  
pass  filtered  before  being  sampled.  What  value  should  the  cutoff  frequency  of  the  low  pass  filter  have?  Assume  
ideal  filtering.  

Problem 9.2

The  quantization  error  for  an  analog  signal,  which  varies  between  0  and  5  V,  is  to  be  less  than  1  mV  when  that  
signal  is  converted  to  PCM  form.  Find  the  minimum  number  of  bits  for  the  digital  conversion.  What  is  the  value  
of  the  actual  quantization  error  for  that  minimum  number  of  bits?  Assume  use  of  the  same  conversion  scheme  
as  illustrated  in  Figure  9-­‐1.  How  would  your  answers  change  if  the  scheme  were  modified  such  that  the  sample  
values  were  always  rounded  down  to  the  next  nearest  quantization  level?  

Problem 9.3

Two  analog  signals  each  having  frequency  content  from  0  to  4  KHz  are  frequency  division  multiplexed  onto  
carriers  at  10  KHz  and  20  KHz  by  a  DSB-­‐SC  AM  process.  The  composite  signal  is  then  sampled  and  converted  to  
a  10-­‐bit  digital  signal.  Determine  the  minimum  sampling  rate  for  the  composite  analog  signal  and  the  minimum  
transmission  bandwidth  for  the  PCM  bit  stream.  

Problem 9.4

An  analog  signal  which  varies  over  the  range  -­‐1  V  to  1  V  is  to  be  converted  to  a  4-­‐bit  PCM  signal.  The  input  
analog  signal  is  band  limited  to  the  range  100  Hz  to  1000  Hz.  

a. Determine  the  minimum  sampling  rate  for  the  conversion.  


b. Assuming  the  same  conversion  scheme  as  illustrated  in  Figure  9-­‐1,  determine  the  number  of  
quantization  levels,  the  step  size  and  the  values  for  the  minimum  and  maximum  quantization  levels.  
c. Find  the  maximum  quantization  error.  
d. Find  the  minimum  transmission  bandwidth  for  the  resulting  PCM  signal.  

Problem 9.5

Each  of  20  analog  audio  signals  of  frequency  content  up  to  10  KHz  is  first  transformed  into  11-­‐bit  digital  form.  
The  20  PCM  signals  are  then  time  division  multiplexed  together  before  transmission  over  a  light  fiber.  

a. Determine  the  frequency  of  the  transmitter  commutator.  


b. Determine  the  minimum  transmission  bandwidth  for  the  light  fiber.  

Problem 9.6

A  binary  channel  with  a  bit  rate  of  28.8  Kbits/s  is  available  for  PCM  voice  transmission.  Find  appropriate  values  
for  the  sampling  rate  and  the  number  of  bits  for  quantization  if  the  maximum  frequency  content  of  the  audio  
signal  is  4  KHz.  

126
Problem 9.7

An  analog  signal  is  quantized  and  transmitted  using  a  PCM  code.  If  each  sample  at  the  receiving  end  of  the  
system  must  be  known  to  within  4%  of  the  Range  of  the  input  analog  signal  at  the  source,  how  many  bits  are  
required  for  each  sample?    

127
Chapter  10:  Networking  Overview  

10.1    Introduction  
  Networking  of  systems  has  always  been  driven  by  the  desire  to  share  resources.  Over  time,  these  
resources  have  changed  but  the  basic  premise  has  not.  The  early  days  of  personal  computers  allowed  people  to  
do  much  of  their  own  word  processing  or  spreadsheets,  yet  high  speed  laser  printers  were  still  very  expensive.  
Storage,  in  the  form  of  disk  drives  was  also  fairly  expensive;  PCs  had  5¼  inch  floppy  drives  that  held  a  
whopping  360K  of  data.  The  early  local  area  networks  were  focused  on  sharing  printers  and  file  servers  and  
were  developed  by  many  different  vendors.  Due  to  the  many  vendors  or  manufacturers,  many  different  
networking  “standards”  were  developed  that  did  not  interoperate.  This  lack  of  interoperability  severely  limited  
the  ability  to  share  files  and  resources  throughout  a  company  or  organization  unless  the  same  manufacturer  
was  used  for  all  computer  resources.  The  answer  to  this  lack  of  interoperability  was  to  have  one  standard  or  
protocol  that  all  vendors  could  agree  upon.  The  computer  network  we  know  today  is  simply  a  communication  
system  that  accomplishes  one  of  three  broad  uses:  

  1.  Share  Resources  (Printers,  Scanner,  Server,  etc.)  

  2.  Share  Files  or  Data  through  common  storage  space  or  “Shared”  drive  

  3.  Communicate  (Email,  IM,  etc.)  

10.2    Basic  Networking  Components    


  Before  reading  any  articles  describing  the  technical  inner-­‐workings  of  networks,  a  basic  education  of  
terminology  is  required.  Networks  are  composed  of  three  basic  parts:  Nodes,  Links  and  communications  
Protocols.    Often,  nodes  and  links  are  lumped  into  something  known  as  a  network  cloud.    Nodes  are  usually  a  
device  used  to  access  the  network,  such  as  a  PC.    However,  it  can  be  a  device  which  is  primarily  accessed,  such  
as  a  networked  printer  or  a  file  server.  Links  are  the  transmission  lines  between  nodes.  As  we  shall  see,  we  do  
not  require  the  links  to  be  hard  wired  or  shared,  they  can  be  wireless  or  point-­‐to-­‐point.  A  Local  Area  Network  
(LAN)  can  be  made  up  of  as  few  as  two  nodes  and  one  link.  Finally,  the  network  cloud  is  used  to  describe  a  
method  to  connect  nodes  (which  can  be  LANs)  inside  of  which  is  a  myriad  of  different  hardware  devices  and  
links  needed  to  accomplish  this  task.  Networks  also  require  a  set  of  rules  or  protocol,  which  they  use  to  
communicate.  This  common  protocol  enables  many  different  vendor  products  to  communicate  by  agreeing  
upon  a  common  set  of  rules  for  the  network.  By  enforcing  that  communicating  parties  adhere  to  a  common  
protocol,  communication  is  made  possible.    A  protocol  architecture  is  a  structured  set  of  protocols  that  
implements  the  exchange  of  information  between  computers.  

10.3    Networking  Entities  


  Networks  are  usually  classified  as  either  centralized  or  distributed  based  upon  the  functions  the  network  
provides  and  the  role  of  the  attached  nodes.  If  everyone  on  the  network  is  equal  and  the  network  is  used  to  
share  information  equally  between  peers,  the  network  is  classified  as  a  distributed  network.  If  there  is  some  
hierarchy  amongst  the  nodes,  where  one  acts  as  a  server  for  many  other  nodes,  known  as  clients,  then  it  is  
described  as  being  a  centralized  network,  centered  around  the  server  which  in  the  past  was  a  mainframe  
computer.  This  client-­‐server  paradigm  is  still  used  today  though  the  server  is  usually  just  another  node  with  a  
large  centralized  repository  of  data.  Two  good  examples  are  web  servers  and  email  servers.  Today,  many  

129
computers  share  processing  power,  data,  resources  and  services  in  client-­‐server  and  peer  to  peer  networks  and  
this  is  called  collaborative  or  cooperative  computing.  Networks  can  also  be  classified  based  on  their  size  or  area  
coverage.  

Figure 10.1: Typical Network made up of Nodes and Links Connected by a Cloud.

  The  most  common  of  these  is  the  Local  Area  Network  or  LAN.  Typically  found  in  an  office  or  home,  it  is  
local  in  scope  and  small  in  size.  The  next  size  is  the  lesser  used  Metropolitan  Area  Network  (MAN)  which  was  
used  to  describe  a  city-­‐wide  coverage  area.  Another  popular  network  category  is  the  Wide  Area  Network  
(WAN).  These  can  cover  great  distances  (including  global)  and  imply  a  large  number  of  smaller  LANs  connected  
through  some  sort  of  backbone  (links  in  the  cloud).  The  Internet  is  the  best  example  of  a  WAN.  Another  term  
used  to  describe  internal  networks  which  may  or  may  not  be  connected  to  a  WAN  is  an  intranet.  These  are  
usually  protected  from  the  rest  of  the  world  by  software  known  as  a  firewall,  keeping  local  data  local  and  only  
allowing  inside  users  access  to  the  outside  world  and  not  the  other  way  around.  A  campus  network  is  a  good  
example  of  an  intranet.  One  other  type  of  network  worth  noting  is  the  Enterprise  WAN,  which  connects  widely  
separated  computer  resources  of  a  single  organization  across  any  distance.  Finally,  the  technology  of  wireless  
networks  is  creating  a  new  classification  known  as  the  Personal  Access  Network  (PAN),  using  Bluetooth  
wireless  networking  devices  to  connect  cell  phones  to  personal  digital  assistants  (PDAs).  

  Several  network  applications  are  used  in  either  a  LAN,  MAN  or  WAN.  Some  of  these  applications  my  be  
familiar  to  you  and  include,  Internet  Explorer,  Email,  Instant  Messenger,  Media  Player,  Blackboard  and  many  
others.  These  programs  are  either  controlled  from  a  central  point  (Like  email  for  example)  or  rely  on  the  shared  
communication  paths  that  a  computer  network  provides  in  order  to  retrieve  data  or  communicate.    

10.4    Hardware  and  Software  


  As  described  above,  two  key  components  required  to  complete  a  network  connection  are  the  physical  
media  (links)  and  a  communication  protocol  that  allows  the  two  end  systems  to  understand  each  other.  Often  
times  these  are  mixed  or  used  interchangeably.  For  example  a  standard  serial  communications  protocol  known  
as  RS232  is  actually  a  combination  of  physical  standards  (wires,  electrical  signals,  grounds,  etc)  and  protocols  
for  communicating  (transmit/receive,  flow  control,  etc).  It  is  precisely  this  mixing  of  low  level  standards  that  
causes  confusion  for  most  networking  novices.  Separating  these  two  parts,  while  at  the  same  time  recognizing  
the  need  for  both  to  be  able  to  make  the  connection,  is  illustrated  through  the  use  of  the  layered  model.  The  best  
way  to  eat  an  elephant  is  one  bite  at  a  time  and  the  best  way  to  understand  computer  networks  is  to  take  a  small  

130
piece  or  layer  of  the  network  and  look  at  the  function  that  it  performs.  In  order  to  accomplish  this  layering  
approach  we  use  a  model  of  the  layers  in  a  computer  network.  This  model  is  not  “real”  but  models  the  actual  
functions  a  successful  network  must  provide  in  order  to  share  resources  and  communicate.  While  other  models  
exist,  the  most  common  reference  model  in  use  is  the  Open  Systems  Interconnect  (OSI)  seven-­‐layer  model.  The  
primary  purpose  for  creating  a  layered  model  is  to  separate  the  functionality  of  software  and  hardware  and  
break  up  the  complexity  of  the  protocol  architecture  into  functional  groups  making  it  easier  to  understand  and  
implement.  The  model  provides  a  common  ground  when  discussing  the  various  functions  of  a  network.  The  
downside  with  implementing  the  OSI  model  in  practice  is  that  it  has  been  overcome  by  events.  Years  ago,  the  
market  determined  that  TCP/IP  network  protocols  (another  layered  model)  would  be  implemented  even  before  
the  OSI  model  was  developed.  Thus,  while  it  does  not  serve  as  a  good  model  for  implementation,  it  does  serve  a  
useful  function  as  a  reference  model  to  which  different  implementations  can  be  compared.    

10.5    The  OSI  Model  


  The  International  Organization  for  Standardization  (ISO)  began  developing  the  Open  Systems  
Interconnection  (OSI)  reference  model  in  1977.  It  was  created  to  standardize  the  rules  of  networking  in  order  
for  all  systems  to  be  able  to  communicate.  In  order  for  communication  to  occur  on  a  networking  using  different  
device  drivers  and  protocol  stacks,  the  rules  for  communication  must  be  explicitly  defined.    The  OSI  model  deals  
with  the  following  issues:  

• How  a  device  on  a  network  sends  it's  data,  and  how  it  knows  when  are  where  to  send  it    
• How  a  device  on  a  network  receives  its  data,  and  how  to  know  where  to  look  for  it.    
• How  devices  using  different  languages  communicate  with  each  other.    
• How  devices  on  a  network  are  physically  connected  to  each  other.    
• How  protocols  work  with  devices  on  a  network  to  arrange  data.    

  The  OSI  model  is  broken  down  into  7  layers.    Although  the  first  layer  is  #1,  it  is  always  shown  at  the  
bottom  of  the  model.    We'll  explain  why  later.    For  now,  remember  this  mnemonic:  Please  Do  Not  Throw  
Sausage  Pizza  Away,  to  help  remember  the  seven  layers.    Here  are  the  seven  layers.  

7. Application Layer Application

6. Presentation Layer Presentation

5. Session Layer Session

4. Transport Layer Transport

3. Network Layer Network

2. Data Link Layer Data Link

1. Physical Layer Physical

131
10.6    Protocol  Stacks  
  In  order  for  each  layer  of  the  model  to  communicate  with  the  levels  above  and  below  it,  certain  rules  
were  developed.    These  rules  are  called  Protocols,  and  each  protocol  provides  a  specific  layer  of  the  model  with  
a  specific  set  of  tasks  or  services.    Each  layer  of  the  model  has  its  own  set  of  protocols  associated  with  it.    When  
you  have  a  set  of  protocols  that  create  a  complete  OSI  model,  it  is  called  a  Protocol  Stack.    An  example  of  a  
protocol  stack  is  TCP/IP,  the  standard  for  communication  over  the  internet,  or  Appletalk  for  Macintosh  
computers.  

  As  stated  before,  protocols  define  how  layers  communicate  with  each  other.    Protocols  specifically  work  
with  only  the  layer  above  and  below  them.    They  receive  services  from  the  protocol  below,  and  provide  services  
for  the  protocol  above  them,  which  limits  the  complexity  of  each  layer  by  eliminating  the  need  for  each  layer  to  
understand  the  functions  of  all  other  layers.  This  order  maintains  a  standard  that  is  common  to  all  forms  of  
networking.    

  In  order  for  two  devices  on  a  network  to  communicate,  they  must  both  be  using  the  same  protocol  
stack.    Each  protocol  in  a  stack  on  one  device  must  communicate  with  its  equivalent  stack,  or  peer,  on  the  other  
device.    This  allows  computers  running  different  operating  systems  to  communicate  with  each  other  easily,  such  
as  having  Macintosh  computers  on  a  Windows  NT  network.  

10.6.1    Communication  Between  Stacks      


  When  a  message  is  sent  from  one  machine  to  another,  it  travels  down  the  protocol  stack  or  layers  of  the  
model,  and  then  up  the  layers  of  the  stack  on  the  other  machine.    As  the  data  travels  down  the  stack,  it  picks  up  
headers  from  each  layer.    This  is  called  encapsulation.    Headers  contain  information  that  is  read  by  the  peer  
layer  on  the  stack  of  the  other  computer.    As  the  data  travels  up  the  levels  of  the  peer  computer,  each  header  is  
removed  by  its  equivalent  protocol.    These  headers  contain  different  information  depending  on  the  layer  they  
receive  the  header  from,  but  tell  the  peer  layer  important  information,  including  packet  size,  frames,  and  
datagrams.      

10.6.2    Encapsulation  
Layering  a  protocol,  like  the  ISO  model,  also  implies  that  each  layer  is  adding  something  to  the  one  
above  or  below.  You  can  look  at  the  layered  model  two  ways,  from  the  bottom  up  or  the  top  down.  When  you  
view  the  model  from  the  top  down,  you  assume  that  the  layers  below  you  provide  you  with  some  type  of  service  
that  you  need  to  talk  to  another  entity  on  the  network.  For  example,  all  layers  above  the  physical  layer  assume  
that  some  kind  of  physical  connection  exists.  The  physical  layer  is  concerned  with  all  aspects  of  transmitting  
and  receiving  data  (bits)  on  the  network  media  and  several  key  characteristics  are  defined  but  the  layers  above  
could  care  less  about  the  physical  structure  of  the  network  or  the  mechanical  and  electrical  specifications  for  
using  the  medium.  This  is  what  is  meant  by  “breaking  up  a  complex  problem  into  bite-­‐size  chunks.”  Along  the  
same  lines,  from  the  vantage  point  of  the  lower  layers,  the  upper  layers  are  relied  upon  to  provide  sufficient  
data  in  their  headers  so  that  the  information  can  be  delivered  to  the  proper  destination.  You  can  think  of  the  7-­‐
layer  OSI  model  as  a  diagram  for  mail  delivery  from  the  Postal  Service.  This  diagram  is  created  with  extreme  
and  almost  ridiculous  detail.  
 
Example 10.1

  Count  the  layers  for  you  to  receive  a  letter  via  postal  mail.  The  letter  itself  is  the  Data  that  is  being  sent  
(layer  6).  The  letter  is  then  addressed  to  the  person  (layer  5)  then  the  street  and  address  is  listed  (layer  4),  then  

132
the  city  (layer  3),  zip  code  (layer  2),  country  (layer  1).  When  the  letter  arrives  to  the  correct  country  the  country  
layer  is  no  longer  needed.  The  letter  is  then  sent  to  the  sorting  area  that  will  get  it  shipped  to  the  correct  state,  
zip,  and  city.  When  the  post  office  of  that  city  receives  the  letter,  it  will  then  be  sorted  again  to  the  correct  postal  
employee  who  delivers  the  mail  to  the  correct  street.  When  your  house  receives  the  letter,  it  is  your  name  on  it  
that  communicates  that  the  letter  is  intended  for  you.  You  open  the  envelope  (stripping  away  all  the  lower  
layers)  because  you  really  only  care  about  the  data  (at  the  highest  layer).  This  is  very  similar  to  how  data  is  sent  
via  a  network.  

  The  end  result  of  the  layering  is  encapsulation,  where  each  layer  encapsulates  the  information  from  the  
layer  above  and  puts  its  own  header  or  trailer  on  the  data  to  be  read  and  acted  upon  by  their  peer  layer  at  the  
destination.  Figure  8.2  illustrates  layering  and  encapsulation.  

Figure 10.2: Encapsulation and Layering.

  The  encapsulation  process  allows  each  layer  of  the  OSI  model  to  implement  the  functions  of  a  single  
layer.  The  Physical  layer  does  not  need  to  be  concerned  with  what  the  application  layer  is  implementing  and  
likewise  for  the  remaining  layers.  This  reduces  the  complexity  that  each  layer  must  implement  as  it  is  only  
concerned  with  the  services  at  its  layer.  Encapsulation  breaks  down  a  complex  problem  into  manageable  pieces  
and  enables  each  layer  to  concentrate  on  the  implementation  details  on  its  own  level.  The  disadvantage  is  that  
each  layer  has  to  add  header  information  which  increases  the  size  of  the  packet  and  introduces  overhead.  
Overhead  is  the  extra  bits  that  are  added  to  the  data  as  the  data  moves  down  the  protocol  stack.  We  will  look  at  
an  example  of  Encapsulation  using  the  OSI  layers.    

Example 10.2:

An  OSI  segment  consisting  of  2200  bits  of  data  and  160  bits  of  header  is  sent  to  the  Data  Link  Layer,  which  
appends  another  160  bits  of  header.  This  is  then  transmitted  to  a  destination  network  that  uses  a  32-­‐bit  header  
for  the  Physical  Layer  and  has  a  maximum  packet  size  of  640  bits.  How  many  bits  including  headers  are  
delivered  to  the  destination  network?  

Solution

There  are  2200  bits  of  data  and  the  Network  layer  adds  160  bits  for  a  header,  and  the  Data  Link  Layer  adds  
another  160  bits  of  data  as  it  moves  down  the  stack.  When  the  Physical  Layer  receives  the  packet,  there  is  2520  

133
bits  of  data  from  the  Physical  Layer’s  perspective.  These  2520  bits  are  then  broken  into  packets  that  can  be  a  
maximum  of  640  bits,  with  32  bits  of  header  and  at  most  608  bits  of  data.  The  2520  is  split  into  packets  as  
follows:  

Notice  it  takes  5  packets  to  send  the  2200  bits  of  original  data  and  a  total  of  2680  bits  are  delivered  to  the  
destination  network  due  to  the  headers  added  at  each  level  of  the  stack.  The  encapsulation  process  added  480  
bits  of  overhead.  

10.7    The  Physical  Layer  


The  lowest  layer  in  the  OSI  model,  and  probably  the  easiest  to  understand,  is  the  physical  layer.    This  layer  deals  
with  the  physical,  electrical,  and  cable  issues  involved  with  making  a  network  connection.    It  associates  with  any  
part  of  the  network  structure  that  doesn't  process  information  in  any  way.  The  physical  layer  is  the  physical  
connections  including  the  cables,  Network  Interface  Cards  (NIC),  and  devices  that  make  up  the  network.  

The  physical  layer  is  responsible  for  sending  the  bits  across  the  network  media.    It  does  not  define  what  a  bit  is  
or  how  it  is  used,  merely  how  it's  sent.    The  physical  layer  is  responsible  for  transmitting  and  receiving  the  
data.    It  defines  pin  assignments  for  serial  connections,  determines  data  synchronization,  and  defines  the  entire  
network's  timing  base.  Items  defined  by  the  physical  layer  include  hubs,  cables  and  cabling,  connectors,  
repeaters,  multiplexers,  transmitters,  receivers,  and  transceivers.    Any  item  that  does  not  process  information  
but  is  required  for  the  sending  and  receiving  of  data  is  defined  by  this  layer.  There  are  several  items  addresses  
by  this  layer.  They  are:  

• Network  connections  types,  including  multi-­‐point  and  point-­‐to-­‐point  networks.    


• Network  Topologies,  including  ring,  star,  bus,  and  tree  networks.    
• Analog  or  Digital  signaling.    
• Bit  Synchronization  (When  to  send  data  and  when  to  listen  for  it).    
• Baseband  versus  broadband  transmissions.    
• Multiplexing  (Combining  multiple  streams  of  data  into  one  channel).    

10.7.1    The  Data  Link  Layer  


  The  Data  Link  Layer  is  responsible  for  the  flow  of  data  over  the  network  from  one  device  to  
another.    This  layer  is  where  the  network  packets  are  translated  into  raw  bits  (00110101)  to  be  transmitted  on  
the  physical  layer.  It  accepts  data  from  the  Network  Layer,  packages  that  data  into  frames,  and  sends  them  to  
the  Physical  Layer  for  distribution.    In  the  same  way,  it  receives  frames  from  the  physical  layer  of  a  receiving  
computer,  and  changes  them  into  packets  before  sending  them  to  the  Network  Layer.  This  is  also  a  layer  that  
uses  the  most  basic  addressing  scheme,  Media  Access  Control  (MAC)  Addresses,  which  uniquely  identifies  each  
Network  Interface  Card  (NIC).    Bridges,  Intelligent  Hubs,  and  NICs  are  all  associated  with  the  Data  Link  Layer.  

  Media  Access  Control  gives  a  unique  12  digit  hexadecimal  address.    These  addresses  are  used  to  set  up  
connections  between  devices.    Every  MAC  address  must  be  unique  or  they  will  cause  identity  crashes  on  the  
network.    The  MAC  address  is  normally  set  at  the  factory,  and  conflicts  are  rare.    The  first  half  of  the  address  is  
assigned  to  the  manufacturer.  If  a  manufacturer  uses  all  the  available  addresses,  they  must  apply  for  another  
134
address  assignment.  The  rest  of  the  address  is  determined  by  the  manufacturer.  Some  may  format  parts  of  the  
address  to  match  each  different  product.    In  the  case  of  a  conflict,  the  MAC  address  is  user  set-­‐able.    

Since  the  main  purpose  of  a  MAC  address  is  to  provide  a  unique  identifier  for  each  host  this  does  not  provide  
any  means  for  routing  or  organizing  the  hosts  that  participate  on  a  network.  If  we  only  had  MAC  addresses  and  
no  logical  addresses  (found  in  the  Network  Layer)  all  routers  and  switches  would  have  to  memorize  all  
addresses  available  and  the  routes  needed  to  get  to  the  destination.  This  would  make  the  Internet  extremely  
slow  and  all  network  devices  unbearably  expensive  because  of  the  massive  amounts  of  memory  needed  in  
creating  routing  tables.  Not  to  mention  when  you  would  add  a  new  PC  to  the  internet,  it  would  take  a  
considerable  amount  of  time  for  your  MAC  address  and  the  path  to  your  PC  to  propagate  throughout  the  
Internet.  This  means  that  there  is  a  need  for  another  layer  of  addressing  to  group  machines  together.  The  third  
layer  is  the  Network  Layer.  

10.7.2    The  Network  Layer  


  The  third  layer  of  the  OSI  model  is  the  Network  layer.    The  network  layer  is  responsible  for  logical  
addressing  (e.g.  Internet  Protocol,  IP).  It  allows  for  grouping  computers  together  unlike  the  MAC  address  where  
there  may  be  no  similarity  from  one  MAC  address  to  another.    In  a  logical/IP  address,  the  leading  3  Bytes  
identify  the  network  and  the  remaining  Bytes  identify  the  host  on  that  network.    The  IP  address  is  “mapped”  to  
a  specific  Network  Interface  Card  (NIC)  with  a  unique  MAC  address.  Think  of  a  MAC  address  as  a  person's  
driver's  license  number,  it  is  just  a  number  that  is  unique  and  different  from  anyone  else's.  Now  think  of  an  IP  
address  as  a  person's  mailing  address.  The  mailing  address  groups  people  into  zones  by  using  the  zip  code,  city,  
state,  and  street  identifiers.  

  Thus,  this  layer  is  responsible  for  making  routing  decisions  and  forwards  packets  that  are  farther  then  
one  link  away.    By  making  the  network  layer  responsible  for  this  function,  every  other  layer  of  the  OSI  model  
can  send  packets  without  dealing  with  where  exactly  the  system  happens  to  be  on  the  network,  whether  it  is  1  
hop  or  10  hops  away.  A  hop  is  and  intermediate  connection  in  a  string  of  connections  that  allow  two  nodes  or  
devices  to  communicate.    

  In  order  to  provide  its  services  to  the  data  link  layer,  it  must  convert  the  logical  network  address  into  
physical  machine  addresses,  and  vice  versa  on  the  receiving  computer.    This  is  done  so  that  no  relaying,  routing,  
or  networking  information  must  be  processed  by  a  level  higher  in  the  model  then  the  Network  
level.    Essentially,  any  function  that  doesn't  provide  an  environment  for  executing  user  programs  falls  under  
this  layer  or  lower.      

  Because  of  this  restriction,  all  systems  that  have  packets  routed  through  their  systems  must  provide  the  
bottom  three  layers'  services  to  all  packets  traveling  through  their  systems.    Thus,  any  routed  packet  must  
travel  up  the  first  three  layers  and  then  down  those  same  three  layers  before  being  sent  farther  down  the  
network.    Routers  and  gateways  are  the  principal  users  of  this  layer,  and  must  fully  comply  with  the  network  
layer  in  order  to  complete  routing  duties.  

  When  a  network  card  receives  a  stream  of  bits  over  the  network,  it  receives  the  data  from  the  wires  (the  
first  layer),  then  the  second  layer  is  responsible  for  making  sense  of  these  random  1s  and  0s.  The  second  layer  
first  checks  the  destination  MAC  address  in  the  packet  to  make  sure  the  data  was  intended  for  this  computer.  If  
the  destination  MAC  address  matches  the  MAC  address  of  the  network  card,  the  packet  is  then  sent  to  the  
computer's  operating  system,  the  rest  of  the  layers  (3  -­‐  7).  
135
10.7.3    Transport  Layer  
  The  transport  layer's  main  duty  is  to  ensure  that  packets  are  send  error-­‐free  to  the  receiving  computer  
in  proper  sequence  with  no  loss  of  data  or  duplication.    This  is  accomplished  by  the  protocol  stack  sending  
acknowledgements  of  data  being  sent  and  received,  and  proper  parity/synchronization  of  data  being  
maintained.  The  transport  layer  is  also  responsible  for  breaking  large  messages  into  smaller  packets  for  the  
network  layer,  and  for  re-­‐assembling  the  packets  when  they  are  received  from  the  network  layer  for  processing  
by  the  session  layer.

10.7.4    Session  Layer  


  The  session  layer  is  the  section  of  the  OSI  model  that  performs  the  setup  functions  to  create  the  
communication  sessions  between  computers.    It  is  responsible  for  much  of  the  security  and  name  look-­‐up  
features  of  the  protocol  stack,  and  maintains  the  communications  between  the  sending  and  receiving  computers  
through  the  entire  transfer  process.  The  session  layer  also  determines  who  can  send  data  and  who  can  receive  
data  at  every  point  in  the  communication.    Without  the  dialogue  between  the  two  session  layers,  neither  
computer  would  know  when  to  start  sending  data  and  when  to  look  for  it  in  the  network  traffic.  

10.7.5    The  Presentation  Layer  


  The  presentation  layer  is  responsible  for  protocol  conversation,  data  translation,  compression,  
encryption,  character  set  conversion,  and  graphical  command  interpretation  between  the  computer  and  the  
network.  

10.7.6    The  Application  Layer  


The  application  layer  provides  services  that  support  user  applications,  such  as  database  access,  e-­‐mail  services,  
and  file  transfers.      

  This  completes  the  overview  of  the  OSI  Model.  Remember  the  OSI  Model  is  not  implemented  by  
manufacturers  in  a  layer  by  layer  fashion.  The  OSI  Model  is  a  reference  so  that  everyone  has  a  common  
framework  to  discuss  the  functions  of  a  network.  The  layers  are  summarized  in  the  following  table.  

7. Applications Its primary function is to format and transfer files between


the communication message and the user’s applications
software.

6. Presentation Deals with the form and syntax of the message including
any code translations required.
5. Session Handles such things as management and synchronization of the
data transmission including network log on and log off
procedures.
4. Transport Layer includes multiplexing; error recovery; addressing and
flow control operations; and partitioning of data into smaller
units.

3. Network Determines the network configuration and the route the


transmission can take.

136
2. Data link Defines the framing information for the block of data and
identifies any error detection and correction methods, as
well as any synchronizing and control codes needed for
communication.

1. Physical Layer Defines the physical connections and the electrical standards for
the communication system.

10.8    Physical  Connection  of  a  Network    


  There  are  two  main  physical  connection  components  of  a  network,  consisting  of  the  network  media  and  
the  network  interface  card  (NIC).There  are  many  forms  of  network  media,  but  they  fall  into  two  distinct  
categories:    Physical  and  Wireless.  There  are  three  major  types  of  physical  cabling.  They  are  Coaxial,  Twisted  
Pair,  and  Fiber  Optics.  They  all  share  certain  attributes,  but  differ  in  their  uses.      

Coaxial  cabling  is  much  like  the  cable  used  on  cable  television  wiring,  but  has  certain  shielding  and  impedance  
properties  that  make  it  different  from  the  coax  used  for  TV.  It  is  also  sub-­‐divided  into  two  different  categories;  
RG-­‐8  and  RG-­‐58.    They  differ  in  their  shielding,  and  therefore  their  methods  of  use.  

Twisted  Pair  consists  of  pairs  of  wires  that  looks  much  like  telephone  cabling,  but  with  a  much  different  
connection  end.    Again,  there  are  two  forms  of  Twisted  Pair;  UTP  (Unshielded  Twisted  Pair)  and  STP  (Shielded  
Twisted  Pair).    They  also  can  differ  on  the  number  of  pairs  of  wires  used  to  connect,  usually  using  either  2  or  4  
pairs  of  wires.  

Fiber  Optic  Cable  is  different  from  the  other  two  forms  of  wiring.    Instead  of  using  electricity  to  send  signals  
across  the  cable,  it  uses  light.    Depending  on  the  Spectrum  used,  Fiber  Optics  is  generally  the  fastest  form  of  
network  cabling.  

Wireless  media  consist  of  infra-­‐red  (IR),  radio  frequency  (RF),  microwave,  and  satellite  systems.    All  these  
media  forms  share  one  common  element;  Instead  of  using  a  physical  form  of  transfer,  they  use  wave  forms  
designed  to  flow  through  the  air  to  send  their  signals.  

137
Chapter  11:  Network  Hardware  

11.1    The  Physical  Layer    


  The  Physical  Layer  (layer  1)  of  the  OSI  model  defines  the  electrical  and  mechanical  characteristics  of  the  
network.    The  two  types  of  mediums  we  will  discuss  are  cables  and  cards.  Cable  is  the  medium  through  which  
information  usually  moves  from  one  network  device  to  another.  There  are  several  types  of  cable  which  are  
commonly  used  with  LANs.  In  some  cases,  a  network  will  utilize  only  one  type  of  cable,  other  networks  will  use  
a  variety  of  cable  types.  The  type  of  cable  chosen  for  a  network  is  related  to  the  network's  topology,  protocol,  
and  size.  Understanding  the  characteristics  of  different  types  of  cable  and  how  they  relate  to  other  aspects  of  a  
network  is  necessary  for  the  development  of  a  successful  network.  The  following  types  of  cables  are  used  in  
networks:  

• Unshielded  Twisted  Pair  (UTP)  Cable    


• Shielded  Twisted  Pair  (STP)  Cable    
• Coaxial  Cable    
• Fiber  Optic  Cable    
• Wireless  LANs    

Unshielded  Twisted  Pair  (UTP)  Cable  

Twisted  pair  cabling  comes  in  two  varieties:  shielded  and  unshielded.  Unshielded  twisted  pair  (UTP)  is  the  most  
popular  and  is  generally  the  best  option  (See  Figure  9.1).    

Figure 11.1: Unshielded Twisted Pair


 

  The  quality  of  UTP  may  vary  from  telephone-­‐grade  wire  to  extremely  high-­‐speed  cable.  The  cable  has  
four  pairs  of  wires  inside  the  jacket.  Each  wire  is  separately  insulated  and  each  pair  is  twisted  with  a  different  
number  of  twists  per  inch  to  help  eliminate  interference  from  adjacent  pairs  and  other  electrical  devices.  The  
tighter  the  twisting,  the  higher  the  supported  transmission  rate,  and  the  greater  the  cost  per  foot.  The  EIA/TIA  
(Electronic  Industry  Association/  Telecommunication  Industry  Association)  has  established  standards  of  UTP  
and  rated  five  categories  of  wire.    

  The  most  common  is  Category  3  or  Category  5.  If  you  are  designing  a  10  Mbps  Ethernet  network  (a  
moderate  speed  internet  connection)  and  are  considering  the  cost  savings  of  using  a  Category  3  wire  instead  of  
Category  5,  remember  that  the  Category  5  cable  will  provide  more  “room  to  grow”  as  transmission  technologies  
increase.  Category  6  is  relatively  new  and  is  used  for  gigabit  connections.  

139
Type Use

Category 1 Voice Only (Telephone Wire)

Category 2 Data to 4 Mbps (LocalTalk)

Category 3 Data to 10 Mbps (Ethernet)

Category 4 Data to 20 Mbps (16 Mbps Token Ring)

Category 5 Data to 100 Mbps (Fast Ethernet)

Advantages:  

• Inexpensive  and  readily  available.  


• Flexible  and  light  weight.    
• Easy  to  work  with  and  install.  

Disadvantages:  

• Susceptibility  to  interference  and  noise.  


• Attenuation  problem.  
• For  analog,  repeaters  needed  every  5-­‐6  Km.  
• For  digital,  repeaters  needed  every  2-­‐3  Km.  
• Relatively  low  bandwidth  (3000  Hz).  

  Each  pair  of  twisted  wires  is  a  transmission  line.    One  pair  receives  data  signals  and  the  other  pair  
transmits  data  signals.    A  transmitter  is  at  one  end  of  one  of  these  lines  and  a  receiver  is  at  the  other  end.    A  
much  simplified  schematic  for  one  of  these  lines  and  its  transmitter  and  receiver  is  shown  in  Figure  9.2  below:  

Figure 11.2: Simplified Schematic of a Transmission Line


140
  Pulses  of  energy  travel  down  the  transmission  line  at  about  the  speed  of  light  (186,000  
miles/second).    The  principal  component  of  one  of  these  pulses  of  energy  is  the  voltage  potential  between  wires  
and  current  flowing  near  the  surface  of  the  wires.  This  energy  can  also  be  considered  as  residing  in  the  magnetic  
field  which  surrounds  the  wires  and  the  electric  field  between  the  wires.    In  other  words,  an  electromagnetic  
wave  which  is  guided  by,  and  travels  down  the  wires.  

  The  main  concern  is  the  transient  magnetic  fields  which  surround  the  wires  and  the  magnetic  fields  
generated  externally  by  the  other  transmission  lines  in  the  cable,  other  network  cables,  electric  motors,  
fluorescent  lights,  telephone  and  electric  lines,  lightning,  etc.  This  is  known  as  noise.    Magnetic  fields  induce  
their  own  pulses  in  a  transmission  line  which  may  literally  bury  the  Ethernet  pulses,  the  conveyor  of  the  
information  being  sent  down  the  line.  

  The  twisted-­‐pair  employs  two  principle  means  for  combating  noise.    The  first  is  the  use  of  balanced  
transmitters  and  receivers.  A  signal  pulse  actually  consists  of  two  simultaneous  pulses  relative  to  ground:  a  
negative  pulse  on  one  line  and  a  positive  pulse  on  the  other.  The  receiver  detects  the  total  difference  between  
these  two  pulses.    Since  a  pulse  of  noise  (shown  in  red  in  the  diagram)  usually  produces  pulses  of  the  same  
polarity  on  both  lines  one  pulse  is  essentially  canceled  by  out  the  other  at  the  receiver.    Also,  the  magnetic  field  
surrounding  one  wire  from  a  signal  pulse  is  a  mirror  of  the  one  on  the  other  wire.  At  a  very  short  distance  from  
the  two  wires  the  magnetic  fields  are  opposite  and  have  a  tendency  to  cancel  the  effect  of  each  other  out.    This  
reduces  the  line's  impact  on  the  other  pair  of  wires  and  the  rest  of  the  world.  

  The  second  and  the  primary  means  of  reducing  cross-­‐talk-­‐-­‐the  term  cross-­‐talk  came  from  the  ability  to  
(over)  hear  conversations  on  other  lines  on  your  phone-­‐-­‐between  the  pairs  in  the  cable,  is  the  double  helix  
configuration  produced  by  twisting  the  wires  together.    This  configuration  produces  symmetrical  (identical)  
noise  signals  in  each  wire.    Ideally,  their  difference,  as  detected  at  the  receiver,  is  zero.    In  actuality,  it  is  much  
reduced.  

Unshielded  Twisted  Pair  Connector  

  The  standard  connector  for  unshielded  twisted  pair  cabling  is  an  RJ-­‐45  connector.  This  is  a  plastic  
connector  that  looks  like  a  large  telephone-­‐style  connector  (See  Figure  9.3).  A  slot  allows  the  RJ-­‐45  to  be  
inserted  only  one  way.  RJ  stands  for  Registered  Jack,  implying  that  the  connector  follows  a  standard  borrowed  
from  the  telephone  industry.  This  standard  designates  which  wire  goes  with  each  pin  inside  the  connector.    

Figure 11.3: RJ-45 Connector

141
Shielded  Twisted  Pair  (STP)  Cable  

  A  disadvantage  of  UTP  is  that  it  may  be  susceptible  to  radio  and  electrical  frequency  interference.  
Shielded  twisted  pair  (STP)  is  suitable  for  environments  with  electrical  interference;  however,  the  extra  
shielding  can  make  the  cables  quite  bulky.  Shielded  twisted  pair  is  often  used  on  networks  using  Token  Ring  
topology.    

Coaxial  Cable  

  Coaxial  cabling  has  a  single  copper  conductor  at  its  center.  A  plastic  layer  provides  insulation  between  
the  center  conductor  and  a  braided  metal  shield  (see  Figure  9.  4).  The  metal  shield  helps  to  block  any  outside  
interference  from  fluorescent  lights,  motors,  and  other  computers.    

Figure 11.4: Coaxial Cable

  Although  coaxial  cabling  is  difficult  to  install,  it  is  highly  resistant  to  signal  interference.  In  addition,  it  
can  support  greater  cable  lengths  between  network  devices  than  twisted  pair  cable.  The  two  types  of  coaxial  
cabling  are  thick  coaxial  and  thin  coaxial.  Thin  coaxial  cable  is  also  referred  to  as  thinnet.  Thick  coaxial  cable  is  
also  referred  to  as  thicknet.  Thick  coaxial  cable  has  an  extra  protective  plastic  cover  that  helps  keep  moisture  
away  from  the  center  conductor.  This  makes  thick  coaxial  a  great  choice  when  running  longer  lengths  in  a  linear  
bus  network.  One  disadvantage  of  thick  coaxial  is  that  it  does  not  bend  easily  and  is  difficult  to  install.    

Advantages:  

• Higher  bandwidth  (400  to  600Mhz,  up  to  10,800  voice  conversations)  
• Can  be  tapped  easily  (pros  and  cons)  
• Much  less  susceptible  to  interference  than  twisted  pair  

Disadvantages:  

• High  attenuation  rate  makes  it  expensive  over  long  distance  


• Bulky    

Coaxial  Cable  Connectors  

The  most  common  type  of  connector  used  with  coaxial  cables  is  the  Bayone-­‐Neill-­‐Concelman  (BNC)  connector  
(see  Figure  9.5).  Different  types  of  adapters  are  available  for  BNC  connectors,  including  a  T-­‐connector,  barrel  
connector,  and  terminator.  Connectors  on  the  cable  are  the  weakest  points  in  any  network.  To  help  avoid  
problems  with  your  network,  always  use  the  BNC  connectors  that  crimp,  rather  than  screw,  onto  the  cable.  

142
Figure 11.5: BNC Connector
Fiber Optic Cable

  Fiber  optic  cabling  consists  of  a  center  glass  core  surrounded  by  several  layers  of  protective  materials  
(see  Figure  9.6).  It  transmits  light  rather  than  electronic  signals  eliminating  the  problem  of  electrical  
interference.  This  makes  it  ideal  for  certain  environments  that  contain  a  large  amount  of  electrical  interference.  
It  has  also  made  it  the  standard  for  connecting  networks  between  buildings,  due  to  its  immunity  to  the  effects  of  
moisture  and  lighting.    

  Fiber  optic  cable  has  the  ability  to  transmit  signals  over  much  longer  distances  than  coaxial  and  twisted  
pair.  It  also  has  the  capability  to  carry  information  at  vastly  greater  speeds.  This  capacity  broadens  
communication  possibilities  to  include  services  such  as  video  conferencing  and  interactive  services.  The  cost  of  
fiber  optic  cabling  is  comparable  to  copper  cabling;  however,  it  is  more  difficult  to  install  and  modify.  10BaseF  
refers  to  the  specifications  for  fiber  optic  cable  carrying  Ethernet  signals.    

Figure 11.6: Fiber Optic Cable

Facts  about  fiber  optic  cables    

• Outer  insulating  jacket  is  made  of  Teflon  or  PVC.    


• Kevlar  fiber  helps  to  strengthen  the  cable  and  prevent  breakage.    
• A  plastic  coating  is  used  to  cushion  the  fiber  center.    
• Center  (core)  is  made  of  glass  or  plastic  fibers.    

Advantages:  

• Greater  capacity  (bandwidth  of  up  to  2  Gbps).  


• Smaller  size  and  lighter  weight.  
• Lower  attenuation.  
• Immunity  to  environmental  interference.  
• Highly  secure  due  to  tap  difficulty  and  lack  of  signal  radiation.  

Disadvantages:  

• Expensive  over  short  distance.  


• Requires  highly  skilled  installers.  
• Adding  additional  nodes  is  difficult.  

143
Fiber Optic Connector

  The  most  common  connector  used  with  fiber  optic  cable  is  an  ST  connector.  It  is  barrel  shaped,  similar  
to  a  BNC  connector.  A  newer  connector,  the  SC,  is  becoming  more  popular.  It  has  a  squared  face  and  is  easier  to  
connect  in  a  confined  space.  The  ideal  interconnection  of  one  fiber  to  another  would  have  two  fibers  that  are  
optically  and  physically  identical,  held  by  a  connector  or  splice  that  squarely  aligns  them  on  their  center  axes.  
However,  in  the  real  world,  a  misalignment  due  to  poor  connections  is  a  factor.    Figure  9.7  shows  some  
possibilities:  

Figure 11.7: Examples of Misalignment of Optical Fibers.

Wireless

Figure 11.8: Wireless LAN

  Not  all  networks  are  connected  with  cabling;  some  networks  are  wireless  as  illustrated  in  Figure  9.8.  
Wireless  LANs  use  high  frequency  radio  signals,  infrared  light  beams,  or  lasers  to  communicate  between  the  

144
workstations  and  the  file  server  or  hubs.  Each  workstation  and  file  server  on  a  wireless  network  has  some  sort  
of  transceiver/antenna  to  send  and  receive  the  data.  Information  is  relayed  between  transceivers  as  if  they  were  
physically  connected.  For  longer  distance,  wireless  communications  can  also  take  place  through  cellular  
telephone  technology,  microwave  transmission,  or  by  satellite.  

  Wireless  networks  are  great  for  allowing  laptop  computers  or  remote  computers  to  connect  to  the  LAN.  
Wireless  networks  are  also  beneficial  in  older  buildings  where  it  may  be  difficult  or  impossible  to  install  cables.    

  The  two  most  common  types  of  infrared  communications  are  line-­‐of-­‐sight  and  scattered  broadcast.  
Line-­‐of-­‐sight  communication  means  that  there  must  be  an  unblocked  direct  line  between  the  workstation  and  
the  transceiver.  If  a  person  walks  within  the  line-­‐of-­‐sight  while  there  is  a  transmission,  the  information  would  
need  to  be  sent  again.  This  kind  of  obstruction  can  slow  down  the  wireless  network.    Scattered  infrared  
communication  is  a  broadcast  of  infrared  transmissions  sent  out  in  multiple  directions  that  bounces  off  walls  
and  ceilings  until  it  eventually  hits  the  receiver.    

  Wireless  LANs  have  several  disadvantages.  They  provide  poor  security,  and  are  susceptible  to  
interference  from  lights  and  electronic  devices.  They  are  also  slower  than  LANs  using  cabling.  

Network  Interface  Card  (NIC)  

  The  NIC,  or  network  interface  card,  plays  an  essential  role  in  computer  networking.  A  NIC  allows  a  
computer  to  have  a  dedicated  connection  to  the  LAN  to  transmit  data  back  and  forth  to  and  from  a  server  or  
other  workstations.    

  The  NIC  works  in  the  Physical  layer  of  the  OSI  model,  and  its  main  purpose  is  to  take  data  from  your  
computer  and  convert  it  into  data  frames  that  are  broadcast  onto  the  network  wire.  Each  of  the  above  forms  of  
networking  media  require  its  own  special  form  of  connection  to  a  computer  system.    A  Coaxial  connector  will  
not  work  with  a  Fiber  Optic  NIC,  and  a  UTP  connection  will  not  transmit  to  an  IR  NIC.    Therefore,  which  ever  
form  of  media  you  choose  to  connect  your  network,  you  must  choose  the  equivalent  form  of  Network  Interface  
Card.  

  Recall  that  the  MAC  address  provides  a  unique  identifier  for  each  computer.    It  is  in  the  NIC  where  this  
address  resides.  

Figure 11.9: Sample NIC with Coax Connector

The  NIC  can  also  provide  the  interface  for  a  wireless  network  and  examples  of  each  are  shown  in  Figure  9.10.    
145
Figure 11.10: NIC used as Interface

11.2    The  Data  Link  Layer    


  Above  the  Physical  Layer  in  the  ISO/OSI  seven-­‐layer  model  is  the  Data  Link  Layer  at  Layer  2.  The  Data  
Link  Layer  was  what  the  original  networking  pioneers  envisioned  we  would  need  when  they  came  up  with  the  
idea  of  connecting  computers  to  share  resources  through  “the  ether.”  As  far  back  as  1961,  researchers  had  
predicted  a  “galactic”  network  connecting  computing  resources  all  over  the  world.  More  practically,  in  1976,  
when  researchers  at  Xerox  Palo  Alto  Research  Center  (PARC)  were  tasked  with  finding  a  way  to  share  their  
expensive  printers  between  workstations,  Dr.  Robert  Metcalf  and  David  Boggs  worked  to  create  a  shared  
communications  mechanism.  Dr.  Metcalfe  wrote  a  paper  called  "Ethernet  Distributed  Packet  Switching  for  Local  
Computer  Networks"  [1]  in  which  he  describes  a  packet  switched  mechanism  which  could  be  used  for  a  local  
computer  network.  His  vision  is  described  in  Figure  9.11  reproduced  from  the  original  paper.  

  While  this  was  a  defining  moment  in  the  history  of  networking,  no  mention  of  a  worldwide  web  could  be  
found,  since  there  was  no  thought  at  the  time  of  reaching  beyond  a  few  limited  length  segments.  Thus,  what  is  
described  in  this  section  is  the  Data  Link  or  Layer  2  protocol,  layers  3  through  7  did  not  exist.  

Packet Switching vs. Circuit Switching

  Of  course,  Dr.  Metclaf’s  assumption  was  that  to  share  the  medium,  we  would  develop  a  packet  switched  
mechanism.  This  mechanism  takes  data  and  successfully  shares  a  common  channel  by  breaking  data  up  into  
packets.  This  had  been  proposed  in  1961  by  Dr.  Leonard  Kleinrock  in  his  landmark  paper  which  many  people  
credit  was  the  true  beginning  of  the  Internet.  [2]  The  reason  this  was  so  revolutionary  was  that  telephone  
system  from  its  inception  was  a  circuit  switched  network.  A  circuit  switched  network  creates  a  dedicated  line  
between  two  end  points  and  it  consists  of  three  phases:  

• Establishment  or  set-­‐up  


• Data  Transfer  
• Disconnect  

146
Figure 11.11: Two-segment Ethernet [1]

  In  addition,  the  network  must  have  the  capacity  to  handle  the  call  and  the  intelligence  to  route  it  
correctly.  Think  of  the  telephone  operator  in  the  early  days  of  phone  switching,  they  would  literally  “patch”  a  
call  together,  creating  a  dedicated  physical  link  between  the  two  parties.  Today,  all  of  these  people  are  replaced  
by  computers  and  automated  switches  but  the  principle  still  holds  true.  As  an  example,  in  Figure  9.12  below,  
you  are  calling  from  Annapolis  to  San  Diego,  your  phone  company  provides  the  dial  tone  when  you  lift  the  
receiver,  as  you  dial,  your  call  gets  “routed”  through  the  system,  effectively  establishing  a  connection  to  the  
other  end  where  the  phone  rings  until  answered.    

  The  problem  with  this  method  for  a  typical  data  transfer  application  is  that  there  is  a  tremendous  
amount  of  bandwidth  wasted  by  dedicating  a  line  to  two  users.  Think  of  when  you  browse  the  web,  you  send  a  
short  request  for  a  web  site  and  receive  the  data.  Then  you  may  spend  several  seconds  or  minutes  with  no  
transfers  while  you  view  the  web  page.  These  seconds  equate  to  wasted  bandwidth.  In  addition,  a  circuit  
switched  network  requires  overhead  to  establish  the  circuit  in  the  first  place  and  some  to  break  it  down  when  
finished.  We  shall  see  that  a  packet  switched  network  does  not  require  this  overhead.    

147
Central Long Distance
Central
Office Carrier (Sprint)
Office

Subscriber
Subscriber
Loop
Loop

Connecting
Connecting
Long Haul
Trunk
Trunk
Figure 11.12: Block Diagram of a Phone Communication System

  Packet  switched  networks  imply  that  the  information  you  are  sending  can  be  broken  up  into  small  
packets  and  sent  independently  over  a  network.  The  underlying  network  is  irrelevant  since  each  packet,  or  
datagram,  contains  enough  information  to  find  its  destination.  Thus,  packets  may  take  different  paths  and  may  
arrive  out  of  order.  This  gives  us  the  first  two  requirements  of  the  data  link  layer  in  a  packet  switched  network,  
an  addressing  scheme  and  sequence  numbers  in  the  data  packets.  More  importantly,  it  allows  us  to  share  higher  
bandwidth  trunk  lines  or  local  area  network  links  among  many  users  and  does  not  dedicate  any  hardware.  
Finally,  it  requires  no  set-­‐up  or  disconnect  phases  so  the  overhead  is  lower.  

  There  is  one  hybrid  version  that  is  used  quite  often  and  that  is  virtual  circuit  switching.  In  virtual  circuit  
switching,  a  path  is  pre-­‐planned  before  any  packets  are  sent,  however,  it  is  not  dedicated  to  a  sender/receiver  
pair.  Call  request  and  call  accept  packets  establish  the  connection  (handshake)  and  each  packet  contains  a  
virtual  circuit  identifier  instead  of  destination  address.  This  saves  time  in  the  network  since  routing  decisions  
are  not  required  for  each  packet.  It  also  requires  a  clear  request  to  drop  the  circuit  even  though  it  is  not  a  
dedicated  path.  

11.2.1    The  Data  Link  Layer    


  The  ISO  Model  describes  several  functions  the  data  link  layer  needs  to  provide  to  the  layers  above.  Some  
of  these,  because  of  the  lateness  of  the  reference  model  to  the  scene  have  histories  in  existing  protocols  that  
were  originally  intended  for  a  different  function  but  due  to  their  simplicity  and  the  fact  that  they  worked,  have  
lived  on,  in  part,  in  the  Institute  of  Electrical  and  Electronics  Engineers  (IEEE)  802  suite  of  standards.  It  is  worth  
pointing  out  the  different  reference  model  approaches  since  we  will  be  focusing  mainly  on  the  IEEE  versions  of  
the  protocols.  

148
Figure 11.13: IEEE 802 Reference Model versus OSI Model [3]

  It  is  important  to  note  that  the  IEEE  splits  the  Data  Link  Layer  into  two  sub-­‐layers,  the  logical  link  
control  layer  and  the  medium  access  control  layer,  each  with  their  own  standard.  In  turn,  these  medium  access  
control  (MAC)  standards  are  defined  for  a  variety  of  physical  media.  A  logical  link  control  (LLC)  standard,  a  
secure  data  exchange  standard,  and  medium  access  control  bridging  standards  are  intended  to  be  used  in  
conjunction  with  the  MAC  standards.  An  architecture  and  protocols  for  the  management  of  IEEE  802  LANs  are  
also  defined  by  the  IEEE  [3].  These  are  important  since  the  literature  often  uses  the  names  for  the  standards  
located  at  these  layers  without  referencing  the  layer.    

11.2.2    Functions  of  the  Data  Link  Layer  


  The  data  link  layer  provides  reliable  transit  of  data  across  a  physical  network  link.  Different  data  link  
layer  specifications  define  different  network  and  protocol  characteristics,  including  physical  addressing,  
network  topology,  error  notification,  sequencing  of  frames,  and  flow  control.  Physical  addressing  (as  opposed  to  
network  addressing)  defines  how  devices  are  addressed  at  the  data  link  layer.  Network  topology  consists  of  the  
data  link  layer  specifications  that  often  define  how  devices  are  to  be  physically  connected,  such  as  in  a  bus  or  a  
ring  topology.  Error  notification  alerts  upper-­‐layer  protocols  that  a  transmission  error  has  occurred,  and  the  
sequencing  of  data  frames  reorders  frames  that  are  transmitted  out  of  sequence.  Finally,  flow  control  moderates  
the  transmission  of  data  so  that  the  receiving  device  is  not  overwhelmed  with  more  traffic  than  it  can  handle  at  
one  time.  

  As  mention  above,  the  IEEE  has  subdivided  the  data  link  layer  into  two  sub-­‐layers:  Logical  Link  Control  
(LLC)  and  Media  Access  Control  (MAC).  Figure  9.14  illustrates  the  relationship  of  the  IEEE  sub-­‐layers  of  the  data  
link  layer.  

149
IEEE 802.2

Logical Link Control


Data
IEEE 802.1
Link
Bridging
Layer
IEEE 802.3 IEEE 802.5 IEEE 802.11a/b/g

MAC - Ethernet MAC - Token Ring MAC - Wireless


IEEE 802.3 IEEE 802.5 IEEE 802.11a/b/g Physical

Physical Physical Physical Layer

Figure 11.14: Some Popular Data Link Layer Sub-layers

  The  Logical  Link  Control  (LLC)  sub-­‐layer  of  the  data  link  layer  manages  communications  between  
devices  over  a  single  link  of  a  network.  LLC  is  defined  in  the  IEEE  802.2  specification  and  is  primarily  
responsible  for  the  error  and  flow  control  requirements  of  the  data  link  layer.  The  Media  Access  Control  (MAC)  
sub-­‐layer  of  the  data  link  layer  manages  protocol  access  to  the  physical  network  medium.  The  IEEE  MAC  
specification  defines  MAC  addresses,  which  enable  multiple  devices  to  uniquely  identify  one  another  at  the  data  
link  layer.  Finally,  in  Figure  9.14,  the  IEEE  802.1  Bridging  specification  details  how  we  can  connect  different  
types  of  physical  topologies  to  form  a  single  local  area  network.  

  The  line  between  hardware  and  software  blurs  in  the  data  link  later.  It  does  not  define  the  physical  
characteristics  of  the  links  between  computers  on  a  network,  yet  it  does  prescribe  how  those  computers  are  to  
be  connected.  The  protocols  which  support  the  data  link  layer  are  often  implemented  in  hardware,  for  
performance  reasons;  again,  blurring  the  distinction  between  where  the  physical  layer  stops  and  the  data  link  
layer  begins.    

  The  data  link  layer  can  also  be  confused  with  the  upper  layers  as  well.  The  most  important  distinction  
between  layer  2  and  the  layers  above,  aside  from  functions  performed,  is  the  addressing  scheme  used.  The  data  
link  layer’s  physical  addressing  is  unique  to  the  physical  device.  This  allows  any  computer  to  be  connected  to  
any  local  area  network  and  not  have  address  duplication  or  conflict.  In  a  sense,  the  data  link  layer  addressing  is  
“flat”  in  that  no  hierarchy  exists  in  the  addresses  and  all  are  unique.  The  layer  2  address  is  usually  found  on  a  
ROM  chip  as  part  of  a  Network  Interface  Card  or  NIC.  This  address  is  known  as  the  MAC  address  (Media  Access  
Control)  and  is  48  bits  long  consisting  of  two  equal  parts.  The  first  half  of  the  address  identifies  the  
manufacturer  of  the  NIC  and  the  manufacturer  assigns  the  rest.  The  format  used  to  specify  a  MAC  address  is  six  
groups  of  two  hex  digits  (00-­‐02-­‐B3-­‐BC-­‐10-­‐C5).  Given  that  we  have  48  bits  in  each  MAC  address  there  are  248  
possible  combinations,  giving  us  281.475x1012  addresses,  so  the  MAC  addresses  should  not  run  out  too  soon.    

11.2.3    Data  Link  Layer  Hardware  


  Much  of  what  layer  2  defines  is  a  protocol  architecture  and  the  protocols  to  support  the  layer  2  
functions.  Aside  from  the  network  interface  cards  that  are  used  in  computers  and  printers,  the  expansion  of  
local  area  networks  has  lead  to  a  family  of  devices  that  allow  us  to  extend  our  networks  to  greater  distances  and  
greater  numbers  of  users.  There  is  also  a  desire  to  make  wiring  local  area  networks  easier  and  simplify  adding  
and  removing  nodes  (users)  from  the  network.    
150
11.2.4    Layer  1  Hardware  Revisited  
  There  are  some  devices  which  are  actually  layer  1  devices  although  they  are  often  mistaken  for  a  layer  2  
device.  Repeaters  and  hubs  fall  into  this  category.  Both  of  these  devices  simply  take  a  signal  in  and  send  it  out,  
with  no  decision  making  in  the  process.  A  network  repeater  will  attempt  to  preserve  signal  integrity  by  
regenerating  an  incoming  electrical,  wireless  or  optical,  signals  to  improve  the  quality  of  the  signal  and  increase  
the  distance  a  network  can  span.  Repeaters  usually  contain  filters  and  amplifiers  to  accomplish  this  task.  A  
passive  hub  does  not  manipulate  the  signal  but  provides  a  convenient  mechanism  to  connect  devices.  A  hub  will  
take  an  incoming  signal  from  one  port  and  repeat  the  signal  on  all  other  ports.  Active  hubs  also  perform  the  
signal  preservation  function  of  repeaters.  Because  of  the  number  of  devices  that  can  be  connected  to  a  single  
hub,  an  active  hub  could  be  called  a  multi-­‐port  repeater  but  more  commonly  they  are  just  called  hubs.  Layer  1  
devices  are  shown  below  in  Figure  9.15  along  with  the  layered  representation  of  the  attached  devices.  

nodes

LAN
repeater

LAN
node node

LAN NIC repeater/hub NIC


nodes LLC LLC
MAC MAC
hub PHY PHY PHY

Figure 11.15: Layer 1 Hardware Diagrams with the Associated Layer Model

11.2.5    Layer  2  Hardware  


  Each  of  the  MAC  Layer  implementations  has  physical  limitations  with  respect  to  the  network  size.  The  
size  of  the  network  refers  to  both  the  number  of  nodes  on  the  network  as  well  as  the  distance  a  network  
segment  can  cover.  In  order  to  overcome  these  limitations,  a  layer  2  device  called  a  bridge  was  developed.  Early  
bridges  were  used  primarily  to  partition  larger  networks  into  smaller,  more  local  ones.  Another  valuable  use  for  
a  bridge  was  to  incorporate  the  functions  of  a  gateway,  or  network  translator,  and  connect  LANs  of  different  
MAC  layer  protocols,  i.e.  Ethernet  to  Token  Ring.  Probably  the  most  common  reason  for  using  a  bridge  is  
increasing  the  performance  of  the  overall  network.  This  is  the  result  of  segmenting  a  large  network  into  smaller,  
more  manageable  networks  allowing  for  communication  across  the  entire  network  while  keeping  local  traffic  
local.    

151
  Another  layer  2  device  not  designed  for  connecting  dissimilar  networks  is  the  layer  2  switch.  The  switch,  
often  confused  with  a  hub,  has  enough  intelligence  to  allow  for  multiple  independent  connections  to  be  active  
simultaneously  on  a  single  LAN.  Thus,  even  though  it  looks  like  a  hub,  it  does  not  echo  out  incoming  traffic  on  all  
other  links  but  rather  only  on  the  link  whose  MAC  address  is  associated  with  that  link.  This  implies  that  the  
switch  must  know  about  MAC  layer  addressing  and  implement  that  layer  of  the  data  link  protocol.  The  
difference  between  the  switch  and  the  bridge  are  shown  in  Figure  9.16  below.  Notice  that  the  functionality  
required  to  implement  either  of  these  devices  exists  in  the  data  link  layer.    

A
LAN B node node
nodes
NIC NIC
switch
switc LLC LLC
E
h MAC MAC MAC
PHY PHY PHY

C
D
node
s node node
bridge
LAN B
bridg NIC NIC
e LLC bridge LLC
LAN A MAC MAC MAC
PHY PHY PHY

LAN C

Figure 11.16: Layer 2 Hardware Diagrams with the Associated Layer Models

  One  of  the  results  of  the  growth  of  the  use  of  bridges  in  larger  networks  is  the  possibility  of  a  closed  loop  
being  created  by  having  multiple  bridges  on  a  network.  The  problem  that  this  creates  is  that  network  traffic  can  
be  forwarded  around  the  network  forever,  being  repeated  over  and  over  by  intermediate  destinations.  This  
would  eventually  cause  a  network  to  crash  after  performance  had  been  slowed  to  a  crawl.  To  solve  this  
problem,  the  bridges  execute  a  spanning  tree  algorithm  effectively  learning  the  topology  of  the  network  and  
creating  a  distribution  tree  with  no  loops.  

  Finally,  there  exists  some  confusion  about  the  introduction  of  the  layer  3  switch  which  is  often  confused  
with  a  layer  2  switch.  The  layer  3  switch  routes  packets  based  on  their  layer  3  addresses,  their  IP  address,  thus  
they  are  network  layer  devices.  Routers  also  fall  into  this  category.  

  We  have  seen  that  there  are  many  devices  used  to  improve  the  performance  and  extend  the  range  of  
local  area  networks.  Most  of  these  devices  operate  at  the  data  link  layer  which  the  IEEE  has  broken  into  2  
primary  sub-­‐layers.  These  devices  give  the  network  designers  a  great  deal  of  flexibility  in  the  implementation  of  
152
many  different  kinds  of  LAN  protocols.  The  LLC  sub-­‐layer  provides  a  common  interface  to  the  upper  layers  so  
that  the  network  layer  does  not  need  to  know  what  makes  up  the  underlying  topology  or  physical  layer.  This  is  a  
benefit  of  layering  in  networking,  hiding  the  details  of  the  lower  layers  from  the  upper  layers  while  providing  
the  same  expected  services  that  are  required  by  the  applications  above.  

11.3     The  Network  Layer    


  A  router  is  a  computer  networking  device  that  forwards  data  packets  toward  their  destinations  through  a  
process  known  as  routing.  Routing  occurs  at  layer  3  of  the  OSI  seven-­‐layer  model.  Routing  is  most  commonly  
associated  with  the  Internet  Protocol,  although  other  less-­‐popular  routed  protocols  remain  in  use.  In  the  
original  1960s-­‐era  of  routing,  general-­‐purpose  computers  served  as  routers.  Although  general-­‐purpose  
computers  can  perform  routing,  modern  high-­‐speed  routers  are  highly  specialised  computers,  generally  with  
extra  hardware  added  to  accelerate  both  common  routing  functions  such  as  packet  forwarding  and  specialised  
functions  such  as  IPsec  encryption.    

  Other  changes  also  improve  reliability,  such  as  using  battery  rather  than  mains  power,  and  using  solid-­‐
state  rather  than  magnetic  storage.  Modern  routers  have  thus  come  to  resemble  telephone  switches,  whose  
technology  they  are  currently  converging  with  and  may  eventually  replace.  The  first  modern  (dedicated,  
standalone)  routers  were  the  Fuzzball  routers.  A  router  must  be  connected  to  at  least  two  networks,  or  it  will  
have  nothing  to  route.  A  special  variety  of  router  is  the  one-­‐armed  router  used  to  route  packets  in  a  virtual  LAN  
environment.  In  the  case  of  a  one-­‐armed  router  the  multiple  attachments  to  different  networks  are  all  over  the  
same  physical  link.    

  A  router  which  connects  end-­‐users  to  the  Internet  is  called  Edge  router:  a  router  which  serves  to  
transmit  data  between  other  routers  is  called  Core  router.  A  router  creates  and/or  maintains  a  table,  called  a  
“routing  table”  that  stores  the  best  routes  to  certain  network  destinations  and  the  “routing  metrics”  associated  
with  those  routes.  Routing  is  a  core  concept  of  the  Internet  and  many  other  networks.  Routing  provides  the  
means  of  discovering  paths  along  which  information  (usually,  but  not  always,  packets)  can  be  sent.  Circuit-­‐
based  networks,  such  as  the  voice  telephone  network,  also  perform  routing,  to  find  paths  for  calls  through  the  
network  fabric.    

  Automatic  routing  makes  networks  autonomous.  Such  networks  can  use  their  routing  to  find  the  best  
route  to  deliver  data  to  a  destination;  choices  are  made  depending  upon  goals  such  as  finding  the  shortest  
distances  and  the  fastest  links  available  through  a  choice  of  network  connections.  This  allows  the  network  to  
route  around  network  failures  and  blockages,  and  can  make  many  aspects  of  the  day  to  day  running  of  such  
networks  automatic,  and  free  from  the  need  for  human  intervention    

  The  actual  process  of  passing  logically  addressed  packets  from  their  local  subnetwork  toward  their  
ultimate  destination  is  called  forwarding.  It  is  closely  related  to  routing,  in  that  routing  tells  the  forwarding  
where  to  send  packets,  but  they  are  logically  completely  separate.  In  large  networks,  packets  may  pass  through  
many  intermediary  destinations  before  reaching  their  destination.  Routing  and  forwarding  both  occur  at  layer  3  
of  the  OSI  seven-­‐layer  model.    

  Hubs  and  switches  move  data  on  what  appears  (to  the  connected  computers)  to  be  the  local  network,  
and  are  invisible  to  connected  computers,  while  the  router  is  explicitly  visible  to  them.  Knowing  where  to  send  
packets  requires  knowledge  of  the  structure  of  the  network.  In  small  networks,  routing  can  be  very  simple,  and  
is  often  configured  by  hand.  In  large  networks  the  topology  of  the  network  can  become  complex,  and  may  
change  constantly,  making  the  problem  of  constructing  the  routing  tables  very  complex.  As  routers  can  only  
recalculate  the  best  routes  very  slowly  relative  to  the  rate  of  arrival  of  packets,  routers  keep  a  routing table  that  
153
maintains  a  record  of  only  the  best  possible  routes  to  certain  network  destinations  and  the  routing  metrics  
associated  with  those  routes.    

11.3.1    Routed  versus  Routing  Protocol    


  There  is  often  confusion  between  routed  protocol  and  routing  protocol.  A  routed  protocol  is  any  
network  protocol  that  provides  enough  information  in  its  network  layer  address  to  allow  a  packet  to  be  
forwarded  from  one  host  to  another  host  based  on  the  addressing  scheme.  Routed  protocols  define  the  format  
and  use  of  the  fields  within  a  packet.  Packets  generally  are  conveyed  from  end  system  to  end  system.  IP  is  an  
example  of  a  routed  protocol.    

  Routing  protocols  facilitate  the  exchange  of  routing information  between  networks,  allowing  routers  to  
build  routing  tables  dynamically.  Traditional  IP  routing  stays  simple  because  it  uses  next-hop routing  where  the  
router  only  needs  to  consider  where  it  sends  the  packet,  and  does  not  need  to  consider  the  subsequent  path  of  
the  packet  on  the  remaining  hops.  Although  this  dynamic  routing  can  become  very  complex,  it  makes  the  
Internet  very  flexible,  and  has  allowed  it  to  grow  in  size  by  more  than  eight  orders  of  magnitude  over  the  years  
since  adopting  IP  in  1983.  Routing  algorithms  use  two  basic  technologies:    

1. Telling the world who your neighbors are:  link-­‐state  routing  protocols  such  as  OSPF.    
2. Telling your neighbors what the world looks like to you:  distance-­‐vector  routing  protocols  such  as  RIP.    

  There  is  also  a  third  method  called  hybrid.  Hybrid  protocols,  such  as  EIGRP,  are  a  combination  of  link-­‐
state  and  distance-­‐vector  routing  protocols.  Hybrid  protocols  have  rapid  convergence  (like  link-­‐state  protocols)  
but  use  much  less  memory  and  processor  power  than  link-­‐state  protocols.  Hybrid  protocols  use  distance-­‐
vectors  for  more  accurate  metrics  and  to  determine  the  best  path  to  destination.    

  A  routing metric  consists  of  any  value  used  by  routing  algorithms  to  determine  whether  one  route  is  
superior  to  another.  Metrics  can  cover  such  information  as  bandwidth,  delay,  hop  count,  path  cost,  load,  MTU,  
reliability,  and  communication  cost.  The  routing  table  stores  only  the  best  possible  routes,  while  link-­‐state  or  
topological  databases  may  store  all  other  information.  Depending  on  the  relationship  of  the  router  relative  to  
other  autonomous  systems,  various  classes  of  routing  protocols  exist:    

1. Ad hoc network routing protocols  appear  in  networks  with  no  or  little  infrastructure.    
2. Interior Gateway Protocols  (IGPs)  exchange  routing  information  within  a  single  autonomous  system.    
3. Exterior Gateway Protocols  (EGPs)  route  between  separate  autonomous  systems.    

11.3.2    TCP/IP  
  We  have  mentioned  that  several  protocols  or  sets  of  rules  are  used  to  communicate  over  a  computer  
network.  Each  of  these  protocols  must  accomplish  several  tasks  such  as  encapsulation,  fragmentation  and  
reassembly,  connection  control,  ordered  delivery,  flow  control,  error  control,  addressing,  multiplexing  and  
transmission  services.  Perhaps  the  most  common  of  protocol  used  to  accomplish  these  tasks  is  the  
Transmission  Control  Protocol  /  Internet  Protocol  or  TCP/IP.  TCP/IP  is  the  communications  protocol  that  hosts  
use  to  communicate  over  an  internet  and  it  establishes  a  virtual  connection  between  a  destination  and  source  
host.  TCP/IP  uses  two  protocols  to  accomplish  this  task,  TCP  and  IP  [4].  TCP  enables  two  hosts  to  establish  a  
connection  and  exchange  data.  TCP  will  guarantee  the  delivery  of  data  and  also  guarantees  that  the  packets  will  
be  delivered  in  the  same  order  in  which  they  were  sent.  Remember  packets  are  sent  through  a  network  
according  to  the  best  path.  This  “best  path”  choice  does  not  guarantee  that  all  packets  will  take  the  same  path,  

154
nor  will  they  arrive  in  the  same  order  they  were  sent.  TCP  has  the  job  of  ensuring  that  all  the  packets  are  
received  and  put  back  into  the  correct  order  before  they  are  passed  up  the  protocol  stack.    

  IP  determines  the  format  of  the  packets.  The  IP  packet  format  will  not  be  discussed  in  detail,  but  there  
are  a  total  of  20  octets  in  an  IP  Version  4  packets.  These  bits  are  used  to  select  the  type  of  service,  length  of  
datagram,  identification  number,  flags,  time  to  live,  next  higher  protocol,  header  checksum,  and  various  address.  
The  packet  IP  provides  a  function  similar  to  the  address  on  a  postal  letter.  You  write  the  address  on  the  letter  
and  put  it  in  the  mail  box.  You  and  the  receiver  know  where  it  is  sent  from  and  to  whom  it  is  being  sent,  but  the  
path  is  determined  by  someone  else.  That  someone  else  is  the  routers  in  the  network  between  the  destination  
and  the  source.  It  is  TCP/IP  that  establishes  the  connection  between  the  destination  and  the  source.  TCP  steps  in  
and  cuts  the  letter  up  into  smaller  pieces  or  packets  and  then  sends  them,  ensures  all  the  packets  are  received  
and  put  back  into  the  proper  order.      

11.3.3    UDP  
UDP  is  another  protocol  used  at  the  transport  level.  UDP  provides  a  connectionless  service  for  applications.  UDP  
provides  few  error  recovery  services,  unlike  TCP.  However,  like  TCP,  UDP  uses  IP  to  route  its  packets  
throughout  the  internet.  UDP  is  used  when  the  arrival  of  a  message  is  not  absolutely  critical.  You  may  recall  that  
from  time  to  time  your  receive  letters  for  current  resident  in  your  mailbox.  The  sender  of  this  “junk  mail”  is  not  
concerned  that  everyone  receives  the  package  they  send.  UDP  is  similar  to  the  “current  resident”  mail  and  is  
often  used  to  send  broadcast  messages  over  a  network.  A  broadcast  message  is  a  message  that  is  sent  
periodically  to  all  hosts  on  the  network  in  order  to  locate  users  and  collect  other  data  on  the  network.  UDP  
messages  are  also  used  to  request  responses  from  nodes  or  to  disseminate  information.  Another  application  of  
UDP  is  in  the  use  of  real-­‐time  applications.  With  real  time  applications,  retransmitting  and  waiting  for  arrival  of  
packets  is  not  possible  so  TCP  is  not  used  for  these  applications.  When  real  time  data  (voice  or  video)  is  routed  a  
connectionless  UDP  protocol  is  used.  If  packets  get  dropped  or  fail  to  arrive  the  overall  message  is  usually  not  
corrupted  beyond  recognition.      

References

[1] R Metcalf and D. Boggs, “Ethernet Distributed Packet Switching for Local Computer Networks,” Communications
of the ACM, Vol. 19, No. 5, July 1976, pp. 395 – 404

[2] L. Kleinrock, "Information Flow in Large Communication Nets", RLE Quarterly Progress Report, July 1961.

[3] IEEE Standards for Local and Metropolitan Area Networks: Overview and Architecture, IEEE Computer Society,
IEEE Standards Board Approved November 20, 1990

[4] www.webopedia.com).

155
Chapter  12:  Internet  and  Addressing  

12.1    Introduction  
  We  have  seen  the  components  that  make  up  a  network  and  how  information  travels  across  a  network,  
but  how  does  a  packet  find  its  intended  destination?    The  internet  is  organized  in  a  hierarchical  structure.  The  
entire  network  is  often  referred  to  as  the  “internet”  or  the  World  Wide  Web.  The  internet  is  subdivided  into  
several  smaller  networks  which  are  all  interconnected  by  routers  which  connect  one  network  to  another  within  
the  internet.      

  The  internet  connects  several  separate  segments  or  networks  together  using  routers.  Routers  need  
some  way  to  identify  the  destination  network  that  a  packet  is  bound  for.  Routers  accomplish  this  by  using  the  
Network  IP  address.  All  devices  on  that  network  share  the  same  network  address,  but  have  unique  host  
addresses.  Packets  get  routed  from  network  to  network  until  they  arrive  at  the  network  that  contains  the  host  
the  packet  has  been  sent  to.    

  A  good  example  of  the  hierarchal  structure  is  to  examine  the  structure  within  the  brigade.  The  brigade  is  
separated  into  2  Regiments  with  3  Battalions  per  Regiment.  Each  Battalion  has  5  companies.  Each  Company  
then  has  platoons  and  each  platoon  has  squads.  An  individual  midshipman  is  in  a  squad,  platoon,  company,  
battalion,  and  so  on.  If  I  want  to  contact  all  the  midshipmen  in  a  particular  company,  I  can  send  a  message  to  just  
that  company,  battalion,  or  platoon.  In  a  computer  network,  the  ability  to  send  messages  to  an  individual  host  
on  a  particular  network  is  also  important.  Each  network  is  then  connected  together  into  the  entire  internet  or  
the  “cloud”.    

  We  can  break  each  connection  to  the  cloud  into  its  own  network  and  each  network  would  be  connected  
to  the  cloud  using  a  router.  Every  computer  connected  off  the  router  is  considered  to  be  on  the  same  “network.”  
This  arrangement  is  similar  to  a  family.  The  router  would  represent  a  single  family,  like  the  Jones  family  and  all  
the  segments  represent  the  children  that  are  in  the  Jones  family.  We  can  easily  identify  who  is  in  the  Jones  
family  by  looking  at  the  last  name.  A  router  can  recognize  who  is  in  its  network  by  using  a  set  of  numbers  called  
an  IP  address.    

When a computer receives a packet from the router, the computer will first check the destination MAC address
of the packet at the Data Link Layer. If it matches, it's then passed on to the Network layer. At the Network layer, it
will check the packet to see if the destination IP address matches the computer's IP address. From there, the packet is
processed as required by the upper layers. On the other hand, the computer may be generating a packet to send to the
router. Then, as the packet travels down the OSI model and reaches the Network layer, the destination and source IP
address of this packet are added in the IP header.

12.2    IP  Addresses  


  The  format  of  an  IP  address  is  called  dotted  decimal,  and  consists  of  4  numbers  from  0  to  255  separated  
by  periods  or  dots.  Each  number  between  the  periods  is  considered  an  octet  because  it  represents  8  binary  bits.  

Example:  35.75.123.250  

The  dotted  decimal  format  is  convenient  for  people  to  use,  but  in  reality  the  router  will  convert  this  number  to  
binary,  and  it  sees  the  above  dotted  decimal  number  as  a  continuous  string  of  32  bits.  Each  bit  will  contain  a  one  

157
or  a  zero.  When  working  with  IP  addresses  we  write  them  in  dotted  decimal,  but  we  analyze  them  using  binary.  
Your  calculator  can  easily  convert  between  binary  and  decimal.    The  example  below  shows  an  IP  address  in  
decimal  notation,  which  we  understand  more  easily.  This  IP  address  (35.75.123.250)  is  then  converted  to  
Binary,  which  is  what  the  computer  understands.  You  can  see  how  big  the  number  gets.  Again,  it's  easier  for  us  
to  remember  four  different  numbers  than  32  zeros  or  ones.    

35        .75                    .123      .250  

00100011.01001011.01111011.11111010    

  An  IP  address  has  2  parts,  the  Network  ID  and  the  Host  ID.  Each  bit  will  contain  a  one  or  a  zero.  When  
working  with  IP  addresses  we  write  them  in  dotted  decimal,  but  we  analyze  them  using  binary.  Your  calculator  
can  easily  convert  between  binary  and  decimal.  The  above  IP  address  would  look  
00100011010010110111101111111010  like  to  the  computer.      

12.2.1    Classes  of  IP  Addresses  


  In  order  to  provide  flexibility,  the  early  designers  of  the  IP  Address  standard  sat  down  to  sort  out  the  
range  of  numbers  that  were  going  to  be  used  by  all  computers.  They  organized  the  IP  Address  into  5  classes,  and  
we  normally  use  3  of  these  classes.  When  someone  applies  for  IP  addresses  they  are  given  a  certain  range  
within  a  specific  class  depending  on  the  size  of  their  network.    

Class Range of IP Addresses

A 1.0.0.0 to 127.255.255.255

B 128.0.0.0 to 191.255.255.255

C 192.0.0.0 to 223.255.255.255

D 224.0.0.0 to 239.255.255.255

E 240.0.0.0 to 255.255.255.255

Table 12.1: Five Different Classes of IP Addresses

  In  the  above  table,  you  can  see  the  five  classes.  The  first  class  is  A,  and  our  last  is  E.  The  first  three  
classes  (A,  B  and  C)  are  used  to  identify  workstations,  routers,  switches  and  other  devices,  whereas  the  last  two  
classes  (D  and  E)  are  reserved  for  special  use.    The  IP  Addresses  listed  above  are  not  all  usable  by  hosts!      

  An  IP  address  consists  of  32  Bits,  which  means  its  four  Bytes  long.  The  first  octet  (first  eight  bits  or  first  
byte)  of  an  IP  address  is  enough  for  us  to  determine  the  class  to  which  it  belongs.  And,  depending  on  the  class  to  
which  the  IP  address  belongs,  we  can  determine  which  portion  of  the  IP  address  is  the  Network  ID  and  which  is  
the  Host  ID.    

  For  example,  if  you  were  told  that  the  first  octet  of  an  IP  address  is  “168,”  then,  using  the  above  table,  
you  would  notice  that  it  falls  within  the  128-­‐191  range,  which  makes  it  a  class  B  IP  address.    

158
  Earlier  you  read  that  companies  are  assigned  different  IP  ranges  within  these  classes,  depending  on  the  
size  of  their  network.  For  instance,  if  a  company  required  1000  IP  addresses,  it  would  probably  be  assigned  a  
range  that  falls  within  a  class  B  network  rather  than  a  class  A  or  C.  The  class  A  IP  addresses  were  designed  for  
large  networks,  class  B  for  medium  size  networks  and  class  C  for  smaller  networks.    

  In  order  to  get  the  information  to  the  correct  host,  the  IP  address  is  divided  into  2  parts,  the  Network  ID  
and  the  Host  ID.    These  two  parts  give  us  two  pieces  of  valuable  information:    

1. It  tells  us  which  network  the  device  is  part  of  (Network  ID).    
2. It  identifies  that  unique  device  within  the  network  (Host  ID).    

  Think  of  the  Network  ID  as  the  suburb  you  live  in  and  the  Host  ID  as  your  street  in  that  suburb.  You  can  
tell  exactly  where  someone  is  if  you  have  their  suburb  and  street  name.  In  the  same  way,  the  Network  ID  tells  us  
to  which  network  a  particular  computer  belongs  and  the  Host  ID  identifies  that  computer  from  all  the  rest  that  
reside  in  the  same  network.    The  picture  below  gives  you  a  small  example  to  help  you  understand  the  concept:  

  Routers  will  look  at  the  first  number  or  octet  to  determine  in  which  Class  is  the  IP  Address.  The  Class  
indicates  how  many  bits  are  used  to  represent  the  Network  ID  and  how  many  bits  are  used  to  represent  the  
Host  ID.    In  the  above  picture,  you  can  see  a  small  network.  We  have  assigned  a  Class  C  IP  range  for  this  
network.  Remember  that  Class  C  IP  addresses  are  for  small  networks.  Looking  now  at  Host  A,  you  will  see  that  
its  IP  address  is  192.168.0.2.  The  network  ID  portion  of  this  IP  address  is  in  blue,  while  the  host  ID  is  in  orange.    

  Table  10.2  contains  the  range  of  numbers  that  are  used  to  determine  the  class  of  the  network  and  the  
number  of  bits  that  are  available  to  assign  to  a  network  and  the  hosts  on  that  network.    

  For  example,  140.179.220.200  is  a  Class  B  address.    The  “140”  falls  within  the  128-­‐191  range,  which  
makes  it  a  class  B  IP  address.  So,  by  default  the  network  part  of  the  address  (also  known  as  the  Network  
Address)  is  defined  by  the  first  two  octets  (140.179.x.x)  and  the  node  part  is  defined  by  the  last  2  octets  
(x.x.220.200).    

159
Class Range of first Octet Number of Network ID bits Number of Host ID bits
A 1 – 126 8 Bits 24 Bits
B 128-191 16 Bits 16 Bits
C 192-223 24 Bits 8 Bits
Table 12.2: Identifying Network and Host ID

Now  we  can  see  how  the  class  determines,  by  default,  which  part  of  the  IP  address  belongs  to  the  network  (N)  
and  which  part  belongs  to  the  host  (h).    

Class A -- 0NNNNNNN.hhhhhhhh. hhhhhhhh. hhhhhhhh


Class B -- 10NNNNNN.NNNNNNNN. hhhhhhhh. hhhhhhhh
Class C -- 110NNNNN.NNNNNNNN.NNNNNNNN. hhhhhhhh

Consider Class A IP Address as an example to understand exactly what is happening. Any Class A network
has a total of 7 bits for the Network ID (bit 8 is always set to 0) and 24 bits for the Host ID. Now all we need to do is
calculate the number of networks and hosts: 27 = 128 networks, while 2 24 = 16,777,216 hosts in each network. Of
the 16,777,216 hosts in each network, two cannot be used. One is the Network Address and the other is the Network
Broadcast Address (see Table 10.3). Therefore when we calculate the valid hosts in a network we always subtract 2.

Therefore, if you are asked how many valid hosts you can have on a Class A network, you should answer
16,777,214 and NOT 16,777,216. The same thing applies for the other two classes we use, i.e., Class B and Class C,
the only difference is that the number of networks and hosts changes because the bits assigned to them are different for
each class. Again if you are asked how many valid hosts you can have on a Class B network, you should answer
65,534 and NOT 65,536. And if I you are asked how many valid hosts you can have on a Class C network, you should
answer 254 and NOT 256.

Now you’ve learned that even though we have three classes of IP addresses that we can use, there are some IP
addresses that have been reserved for special use. This doesn't mean you can't assign them to a workstation but in the
case that you did, it would create serious problems within your network. For this reason it's best to avoid using these IP
addresses. Table 10.3 shows the IP addresses that you should avoid using.

It is imperative that every network, regardless of Class and size, has a Network Address (first IP address e.g.
192.168.0.0 for Class C network) and a Broadcast Address (last IP address e.g. 192.168.0.255 for Class C network), as
mentioned in the table and explanation diagrams above, which cannot be used. So when calculating available IP
addresses in a network, always remember to subtract 2 from the number of IP addresses within that network.

160
IP address Function

Default - Network 0.0.0.0 Refers to the default route. This route is to simplify routing tables
used by IP.

Loopback - Network 127.0.0.1 Reserved for Loopback. The Address 127.0.0.1 is often used to refer
to the local host. Using this Address, applications can address a local
host as if it were a remote host.

Network Address - IP Address Refers to the actual network itself. For example, network
with all host bits set to "0" (e.g. 192.168.0.0 can be used to identify network 192.168. This type of
192.168.0.0) notation is often used within routing tables.

Subnet / Network Broadcast - IP IP Addresses with all node bits set to "1" are local network broadcast
Address with all node bits set to addresses and must NOT be used.
"1" (e.g. 192.168.255.255)
Some examples: 125.255.255.255 (Class A), 190.30.255.255 (Class
B), 203.31.218.255 (Class C).

Network Broadcast - IP Address The IP Address with all bits set to "1" is a broadcast address and
with all bits set to "1" (e.g. must NOT be used. These are destined for all nodes on a network, no
255.255.255.255) matter what IP address they might have.

Table 12.3: IP Addresses Unusable by Hosts

12.2.2    Reserved  Host  ID  Numbers  


  When  you  design  a  network,  you  will  be  given  a  Network  ID  from  a  controlling  authority.  The  Network  
ID  portion  of  your  internet  address  cannot  change,  but  the  Host  ID  portion  of  the  IP  Address  are  the  bits  you  
own,  and  you  can  assign  any  IP  within  the  Host  ID  portion  to  the  computers  on  your  network,  other  than  the  all  
zero  or  all  one  Host  ID  mentioned  above.  Typically  designers  will  assign  the  first  assignable  IP  to  the  first  host,  
but  there  is  no  restriction  and  any  number  can  be  assigned.    

  Just  as  the  name  Jones  identifies  the  family  members,  the  Network  ID  identifies  your  network,  but  how  
does  the  router  figure  out  that  the  Network  ID  is  a  match?  You  have  learned  that  information  travels  in  packets  
and  each  packet  has  a  header.  The  header  will  contain  the  IP  address  of  the  computer  that  the  packet  is  being  
sent  to.  The  router  will  use  a  special  sequence  of  bits  called  the  Network  Mask,  to  determine  if  the  packet  is  
being  sent  to  its  network.  The  Network  Mask  has  all  ones  in  the  Network  ID  and  all  zeroes  in  the  Host  ID.  This  
Mask  is,  then,  logically  ANDed  to  the  packet  and  the  Router  will  see  if  the  destination  host  is  on  its  network.  In  
our  35.0.0.0  network,  the  Network  Mask  would  be  255.0.0.0.  If  the  network  was  a  Class  B  the  Network  Mask  
would  be  255.255.0.0  and  for  a  Class  C  the  Network  Mask  would  be  255.255.255.0.    

161
12.3    Network  Mask  
  The  table  below  shows  our  three  network  classes  with  their  respective  Network  Masks:    
 

Class Type Network Range Network Masks

A 1.0.0.0 to 127.255.255.255 255.0.0.0

B 128.0.0.0 to 191.255.255.255 255.255.0.0

C 192.0.0.0 to 223.255.255.255 255.255.255.0

Table 12.4. Network Classes with Network Masks

  An  IP  address  consists  of  two  parts:  1)  The  Network  ID  and  2)  The  Host  ID.    We  can  see  this  once  again  
shown  below,  where  the  IP  address  is  analyzed  in  binary,  because  this  is  the  way  you  should  work  when  dealing  
with  Network  Masks:    

Class  C  IP  Address…  

IP Address: 192. 168. 0. 5

Network Mask: 255. 255. 255. 0

Conversion  to  binary…  

IP Address: 1100 0000. 1010 1000. 0000 0000. 0000 0101

Network Mask: 1111 1111. 1111 1111. 1111 1111. 0000 0000

Network ID Host ID

  The  Class  C  network  uses  21  bits  for  the  Network  ID  and  8  bits  for  the  Host  ID  (remember,  the  first  3  
bits  in  the  first  octet  are  set).    The  Network  Mask  is  what  splits  the  Network  ID  and  Host  ID.      
 
  We  are  looking  at  an  IP  address  with  its  Network  Mask  for  the  first  time.  What  we  have  done  is  take  the  
decimal  Network  Mask  and  converted  it  to  binary,  along  with  the  IP  address.  It  is  essential  to  work  in  binary  
because  it  makes  things  clearer  and  we  can  avoid  making  mistakes.  The  ones  (1)  in  the  Network  Mask  are  
ANDed  with  the  IP  Address  and  the  result  defines  the  Network  ID.  If  we  change  any  bit  within  the  Network  ID  of  
the  IP  address,  then  we  immediately  move  to  a  different  network.  So  in  this  example,  we  have  a  24  bit  Network  
Mask  (24  ones  counting  from  the  left  to  the  right).  
 
 Note:

• All class C IP addresses have a 24 bit Network Mask (255.255.255.0).


• All class B IP addresses have a 16 bit Network Mask (255.255.0.0).
• All class A IP addresses have an 8 bit Network Mask (255.0.0.0).
162
MAC Address versus IP Address

  Recall  the  MAC  address  that  was  discussed  earlier.  The  MAC  address  is  a  unique  address  that  is  assigned  
to  the  physical  device.  The  IP  Address  is  a  logical  address  that  is  used  to  determine  where  in  the  network  a  host  
is  located.  As  in  the  postal  example,  the  state  could  be  considered  the  Network,  with  the  city  the  subnet  and  the  
individual  person  is  the  host  in  the  network.  The  MAC  address  is  like  the  Social  Security  Number  that  each  
person  has.  The  SSN  gives  no  information  about  where  the  person  is  located,  but  does  uniquely  identify  that  
person.  The  MAC  identifies  the  manufacturer  and  is  has  a  unique  number  associated  with  it.  The  IP  Address  is  
used  to  find  where  the  MAC  is  so  that  packets  can  be  routed  to  the  host.    

163
Chapter  13:  Subnetting  

13.1    Introduction  
  As  the  network  grows,  it  becomes  increasingly  difficult  to  efficiently  route  all  the  traffic,  since  the  router  
needs  to  keep  track  of  all  the  hosts  on  its  network.  Let’s  say  that  we  have  a  simple  MAN  of  two  routers.  When  
ever  a  packet  is  sent  from  one  host  to  another  on  our  network,  the  router  will  route  the  packet  to  the  proper  
host.  When  the  a  packet  is  sent  to  a  host  connected  to  another  router,  the  Network  Mask  will  be  used  to  
determine  the  packet  is  not  for  our  network  and  the  router  will  send  the  packet  over  the  communication  link  to  
the  other  network.  If  each  router  only  has  a  few  hosts  connected  to  it,  then  the  amount  of  packets  that  the  
router  has  to  router  is  relatively  small.  As  the  network  grows,  this  number  of  hosts  gets  quit  large.  For  a  class  A  
network  there  could  be  224  Hosts  or  close  to  17  million.  As  a  means  to  help  the  routers  more  efficiently  route  
packets  and  manage  the  size  of  their  router  table,  a  technique  called  subnetting  is  used.    

13.2    Subnet  Mask  


When  we  subnet  a  network,  we  basically  split  it  into  smaller  networks.    By  subnetting  the  network,  we  can  
partition  it  into  as  many  smaller  networks  as  we  need.    By  default,  all  types  of  classes  (A,  B  and  C)  have  a  
Network  Mask.    However,  a  network  mask  other  than  the  default  can  be  used.    This  is  called  a  Subnet  Mask.    The  
use  of  an  IP  address  with  a  Subnet  Mask  other  than  the  Network  default  results  in  the  standard  host  bits  (the  
Bits  used  to  identify  the  HOST  ID)  being  divided  in  to  two  parts:  a  Subnet  ID  and  Host  ID.    Take  the  same  IP  
address  as  above,  and  divide  it  further  into  a  Subnet  ID  and  Host  ID,  thus  changing  the  default  network  mask:    

Class  C  IP  Address  

IP Address: 192. 168. 0. 37

Network Mask: 255. 255. 255. 224

Conversion  to  binary…  

IP Address: 1100 0000. 1010 1000. 0000 0000. 001 00101

Network Mask: 1111 1111. 1111 1111. 1111 1111. 111 00000

Subnet ID Host
Network ID ID

 
Looking  at  the  example  above  you  will  now  notice  that  we  have  a  Subnet  ID,  something  that  didn't  exist  before.    
As  the  example  shows,  we  have  borrowed  three  bits  from  the  Host  ID  and  used  them  to  create  a  Subnet  ID.    
Effectively  we  partitioned  our  Class  C  network  into  smaller  networks.    

165
  When  we  use  IP  addresses  with  their  Network  Masks,  e.g.  192.168.0.37  is  a  class  C  IP  address  so  the  
Network  Mask  would  be  255.255.255.0;  however,  if  IP  addresses  have  their  Subnet  Mask  modified  in  a  way  so  
that  there  is  a  “Subnet  ID.”  This  Subnet  ID  is  created  by  borrowing  bits  from  the  Host  ID  portion.  

  Each  time  we  borrow  a  bit  from  the  Host  ID,  we  split  the  network  into  a  different  number  of  networks.  
For  example,  when  we  borrowed  three  bits  in  the  Class  C  network,  we  ended  up  partitioning  the  network  into  
six  smaller  networks.  Let's  take  a  look  at  a  detailed  example  (which  we  will  break  into  three  parts)  so  we  can  
fully  understand  all  the  above.    We  are  going  to  do  an  analysis  using  the  Class  C  network  and  three  bits  which  
we  took  from  the  Host  ID.  The  analysis  will  take  place  once  we  convert  our  decimal  numbers  to  binary,  
something  that's  essential  for  this  type  of  work.  We  will  see  how  we  get  eight  networks  from  such  a  
configuration  and  their  ranges!    

  We  calculate  the  amount  of  partitioned  networks  in  our  example  above  as  follows:  3  bits  taken  means  a  
total  of  23  -­‐  2=  6  networks.  The  23  represents  the  8  different  ways  we  can  arrange  3  bits.  The  minus  two  is  a  
result  of  the  two  reserved  addresses.  A  000  in  the  Subnet  ID  portion  and  the  111  in  the  Subnet  ID  portion  are  
both  reserved.  The  rule  applies  to  all  types  of  subnets,  no  matter  what  class  they  are.  Simply  take  the  subnet  bits  
and  place  them  into  the  power  of  two  and  you  get  your  networks.    

  Now,  that  was  the  easy  part.  The  second  part  is  slightly  more  complicated.  The  Subnet  ID  and  Host  ID  is  
where  we  get  all  the  information  about  our  subnetworks.    Table  13.1  breaks  down  the  six  subnets.  

Determining the Subnets Determining the Hosts per Subnet

000 Reserved for the Network Address

first: 001 0 0001 to 1 1110

second: 010 0 0001 to 1 1110

third: 011 0 0001 to 1 1110

fourth: 100 0 0001 to 1 1110

fifth: 101 0 0001 to 1 1110

sixth: 110 0 0001 to 1 1110

111 Reserved to Broadcast the Network

Table 13.1: Analysis of Subnets


 

Note:    

• 0  0000  (first  Host  IP  in  each  Subnet)  is  reserved  as  the  Network  Address  for  the  Subnet.  
• 1  1111  (last  Host  IP  in  each  Subnet)  is  reserved  as  the  Broadcast  Address  for  the  Subnet.  
 

166
  When  we  want  to  calculate  the  Subnets  and  Hosts,  we  deal  with  them  one  at  a  time.  Once  that's  done,  we  
put  the  Subnet  ID  and  Host  ID  portion  together  so  we  can  get  the  last  octet's  decimal  number.    

  We  know  we  have  six  networks  (or  subnets)  and,  by  simply  counting  or  incrementing  our  binary  value  
by  one  each  time,  we  get  to  see  all  the  networks  available.  So  we  start  off  with  001  and  finish  at  110.  Next  we  
take  the  Host  ID  portion,  where  the  first  available  host  is  0  0001  (1  in  Decimal),  because  the  0  0000  value  is  
reserved  as  it  is  the  subnet  address,  and  the  last  value  which  is  1  1111  is  used  as  a  broadcast  address  for  each  
subnet.  The  formula  that  allows  you  to  calculate  the  available  hosts  is:  2X  –  2.  Where  X  is  the  number  of  bits  we  
have  in  the  Host  ID  field,  which  for  our  example  is  5.    When  we  apply  this  formula,  we  get  25  -­‐  2  =  30  valid  
(usable)  IP  addresses  for  Hosts  per  subnet.  If  you're  wondering  why  we  subtract  2,  it's  because  one  is  used  for  
the  subnet  address  of  that  subnet  and  the  other  for  the  Broadcast  Address  of  that  subnet.  Summing  up,  these  are  
the  ranges  for  each  subnet  in  our  new  network:  

First Subnet

First Subnet IP: 0010 0000 Last Subnet IP: 0011 1111

Full Range: 192.168.0.32 – 192.168.0.63

Second Network

First Subnet IP: 0100 0000 Last Subnet IP: 0101 1111

Full Range: 192.168.0.64 – 192.168.0.95

Third Network

First Subnet IP: 0110 0000 Last Subnet IP: 0111 1111

Full Range: 192.168.0.96 – 192.168.0.127

Fourth Network

First Subnet IP: 1000 0000 Last Subnet IP: 1001 1111

Full Range: 192.168.0.128 – 192.168.0.159

Fifth Network

First Subnet IP: 1010 0000 Last Subnet IP: 1011 1111

Full Range: 192.168.0.160 – 192.168.0.191

Sixth Network

First Subnet IP: 1100 0000 Last Subnet IP: 1101 1111

Full Range: 192.168.0.192 – 192.168.0.223

 
Note:   The  first  IP  Address  in  each  Subnet  is  the  Subnet  Address  for  that  Subnet.  The  last  IP  Address  in  each  
Subnet  is  the  Broadcast  Address  for  that  Subnet.  

167
13.3    Subnetting  Example    
  In  order  to  better  understand  subnetting,  Bancroft  hall  will  be  used  as  an  example.  In  Bancroft  hall  you  
have  a  squad  leader.  This  squad  leader  has  no  problem  managing  their  squad,  because  there  are  only  about  12  
Midshipmen  in  the  squad.  However,  there  are  over  4000  Midshipmen  in  Bancroft  Hall.  Imagine  if  you  only  had  
one  squad  and  you  were  the  squad  leader  in  charge  of  all  4000  Midshipmen.  It  would  be  impossible  to  manage  
unless  you  set  up  some  type  of  Hierarchical  structure.  In  Bancroft  Hall  the  Midshipmen  are  organized  within  the  
Brigade  into  Regiments,  Battalions,  Companies,  Platoons,  Squads,  and  finally  the  MIR.  

  Computer  Networks  are  no  different,  if  the  network  is  organized  as  one  large  squad,  then  the  router  
cannot  efficiently  route  the  traffic.  The  way  that  Network  managers  get  around  this  problem  is  by  organizing  the  
bits  they  own,  the  Host  ID  portion,  into  subnets  as  described  above.  The  subnets  are  like  the  Regiments,  
Battalions  and  Companies  within  Bancroft  Hall.  Subnetting  uses  the  Host  ID  portion  of  the  IP  address  and  splits  
them  into  Subnet  bits  and  Host  ID  bits.  The  number  of  bits  assigned  to  the  Subnet  and  to  the  Host  is  based  on  
the  needs  of  the  network.  For  example,  let’s  say  that  we  are  given  the  Network  ID  of  135.25.0.0  and  we  are  
tasked  with  subnetting  the  brigade:    

• We  are  tasked  with  providing  40  different  subnets.    


• We  need  at  least  150  hosts  per  subnet.      

Step  1:  Determine  the  number  of  bits  needed  for  the  Subnets.  We  need  40  different  subnets,  which  will  require  
n  bits  for  2n  subnets,  but  we  have  to  remember  that  the  first  address  is  lost  to  the  Network  ID  and  that  last  
address  is  used  to  broadcast  the  Network,  so  we  need  to  subtract  2.    As  we  solve    

    40  ≤  2n  −  2    

for  n,  where  n  is  the  number  of  bits  needed,  we  find  that  5  bits  are  not  enough  and  6  bits  provides  too  many  
 
subnets.  We  choose  six  bits  so  that  we  can  meet  the  requirement  of  40  subnets.  We  have  the  ability  to  have  26   −  
2  =  64  −  2  =  62  subnets,  of  which  only  40  are  used  and  the  rest  allow  for  growth.      

Step  2:  Determine  the  number  of  Hosts  per  Subnet.  We  were  given  a  class  B  address  (Identified  by  the  135  in  
the  first  octet)  which  has  16  bits  for  the  Network  ID,  and  16  bits  are  used  for  the  Host  portion  which  are  the  bits  
we  own  and  can  assign  them  the  to  meet  our  network  needs.  The  following  3  tables  summarize  what  we  have  
been  given  and  how  we  will  reassign  or  reallocate  the  bits  in  the  host  portion  of  the  Network  Address.    

Class  B  IP  Address  

Network Address: 135. 25. 0. 0

Network Mask: 255. 255. 0. 0

Conversion  to  binary  

Network Address: 1000 0111 0001 1001. 0000 0000. 0000 0000

Network Mask: 1111 1111. 1111 1111. 0000 0000. 0000 0000

Network ID Host ID

168
Conversion  to  Subnetting  

IP Address: 1000 0111 0001 1001. 0000 0000. 0000 0000.

Reassign Bits: 1000 0111 1100 1000 000000 00. 0000 0000

Network ID Subnet ID Host ID

  We  used  6  for  the  Subnet,  and  the  remaining  10  (16-­‐6=10)  will  be  used  for  the  Host  ID  on  each  subnet.  
Using  this  arrangement  we  can  have  210  =  1024  −2=  1022  assignable  Host  ID’s  for  each  subnet.  Remember  that  
we  lose  two  IP’s  per  subnet,  one  for  the  Subnet  ID  and  one  for  the  broadcast  on  that  Subnet.  We  will  only  use  
150  of  these  assignable  IP  addresses  and  the  rest  will  be  for  future  growth.    

Step  3:  Determine  the  Subnet  Mask.    In  order  to  identify  our  subnets,  the  router  needs  to  know  the  subnet  
Mask.  The  Subnet  Mask  has  all  ones  in  the  Network  ID,  all  ones  in  the  Subnet  ID  and  all  Zeroes  in  the  Host  ID.  
For  our  example  the  Subnet  Mask  will  be:  11111111.11111111.11111100.00000000  (Dotted  Decimal  
255.255.252.0).    We  have  now  broken  our  large  network  up  into  smaller  Subnets  which  are  easier  to  manage  
and  which  will  enable  us  to  make  our  network  run  more  efficiently.  Each  Subnet  is  like  a  separate  network  in  
the  eyes  of  the  Router  and  traffic  that  is  sent  from  one  host  on  a  Subnet  to  another  can  be  routed  using  a  router,  
bridge  or  switch.  This  arrangement  allows  packets  to  flow  efficiently  through  the  network.  Table  4  contains  the  
1st  and  last  Subnet  with  the  Subnet  Mask.  Please  take  some  time  and  study  this  table  and  understand  how  we  
assign  addresses  within  a  network.      

Network ID 135.25.0.0 10000111.00011001.00000000.00000000


Network Mask 255.255.0.0 11111111.11111111.00000000.00000000
Subnet Mask 255.255.252.0 11111111.11111111.11111100.00000000
1st Subnet ID 135.25.4.0 10000111.00011011.00000100.00000000
1st Host, 1st Subnet 135.25.4.1 10000111.00011011.00000100.00000001
Last Host 1st Subnet 135.25.7.254 10000111.00011011.00000111.11111110
Broadcast 1st Subnet 135.25.7.255 10000111.00011011.00000111.11111111
2nd Subnet ID 135.25.8.0 10000111.00011011.00001000.00000000
3rd Subnet ID 135.25.12.0 10000111.00011011.00001100.00000000
Last Subnet ID 135.25.248.0 10000111.00011011.11111000.00000000
1st Host Last Subnet 135.25.252.1 10000111.00011011.11111000.00000001
Last Host Last Subnet 135.25.255.254 10000111.00011011.11111011.11111110
Broadcast Network 135.25.255.255 10000111.00011011.11111111.11111111
Table C-13.2: Assigned Subnets within the 135.25.0.0

13.4    Plan  for  growth  


When  creating  subnets  for  your  network,  answer  the  following  questions:  

1. How  many  subnets  are  needed  today?    

169
 
Calculate  the  maximum  number  of  subnets  required  by  rounding  up  the  maximum  number  to  the  nearest  
power  of  two.  For  example,  if  an  organization  needs  five  subnets,  2  to  the  power  of  2  or  4  will  not  provide  
enough  subnet  addressing  space,  so  you  must  round  up  to  2  to  the  power  of  3  =  8  subnets.    

2. How  many  subnets  are  needed  in  the  future?  

You  must  plan  for  future  growth.  For  example,  if  9  subnets  are  required  today,  and  you  choose  to  provide  for  2  
to  the  power  of  4  =  16  subnets,  this  might  not  be  enough  when  the  seventeenth  subnet  needs  to  be  deployed.  In  
this  example,  it  might  be  wise  to  provide  for  more  growth  and  select  2  to  the  power  of  5  =  32  as  the  maximum  
number  of  subnets.    

3. What  is  the  maximum  number  of  hosts  on  a  given  segment?  

You  must  ensure  that  there  are  enough  bits  available  to  assign  host  addresses  to  the  organization's  largest  
subnet.  If  the  largest  subnet  needs  to  support  40  host  addresses  today,  2  to  the  power  of  5  =  32  will  not  provide  
enough  host  address  space,  so  you  would  need  to  round  up  to  2  to  the  power  of  6  =  64.    

4. How  many  hosts  will  there  be  in  the  future?    

Besides  planning  for  additional  subnets,  you  must  also  plan  for  more  hosts  to  be  added  to  each  subnet  in  the  
future.  Make  sure  the  organization's  address  allocation  provides  enough  bits  to  deploy  the  required  subnet  
addressing  plan.  When  developing  subnets,  Class  C  addresses  present  the  greatest  challenge  because  fewer  bits  
are  available  to  divide  between  subnet  addresses  and  host  addresses.  If  you  accommodate  too  many  subnets,  
there  may  be  no  room  for  additional  hosts  and  growth  in  the  future.    

  All  the  above  points  will  help  you  succeed  in  creating  a  well-­‐designed  network  which  will  have  the  
ability  to  cater  for  any  additional  future  requirements.    

170
Appendix  A:  Frequency  Spectra  and  Ideal  Filtering  
 

A.1  Amplitude  Spectrum  


  The  spectrum  of  a  signal  is  a  plot  that  gives  the  amount  of  each  sinusoidal  component  at  
each  frequency.  Sinusoids  have  both  amplitude  and  phase  and  so  also  do  spectral  plots.  Here  we  
will  concentrate  only  upon  the  amplitude  spectra  and  ignore  phase.  Ignoring  phase,  the  signal  
v1 (t ) = 2cos(2π 300t ) + 3cos(2π 500t ) Volts  has  the  same  amplitude  spectrum  as  
v2 (t ) = 2sin(2π 300t ) + 3sin(2π 500t ) Volts.  Spectra  vary  from  very  simple  for   v1 (t ) above,  to  more  
complex  for  periodic  signals  like  square  and  triangle  waves,  to  very  complex  for  typical  music  
signals.  Fourier  analysis  of  periodic  signals  will  not  be  considered  here.  We  will  concentrate  on  the  
idea  of  an  amplitude  spectrum  for  a  few  simple  signals  and  the  effect  of  ideal  filtering  on  those  
simple  signals.  A  good  text  book  that  surveys  electrical  engineering  can  be  consulted  for  details  on  
Fourier  series  and  Fourier  transforms.  

  The  amplitude  spectrum  of  a  signal  is  essentially  a  bar  graph  of  the  amplitude  present  at  
each  sinusoidal  frequency  component.  The  amplitude  is  plotted  on  the  vertical  axis  versus  
frequency  value  along  the  horizontal  axis.  The  idea  that  the  horizontal  axis  can  be  frequency  and  
not  time  is  new  and  must  be  kept  in  mind  at  all  times  when  dealing  with  spectra.  The  spectrum  of  a  
light  source,  which  can  be  viewed  using  a  prism,  is  the  intensity  of  light  present  at  each  color.  The  
amounts  of  different  colors  are  different  for  different  sources  of  light.  The  sun,  incandescent  bulbs  
and  florescent  lighting  all  have  different  color  compositions  and  spectra.  

  For  the  signal   v1 (t ) above,  we  see  that  the  amplitude  at  300  Hz  is  2  Volts  and  the  amplitude  
at  500  Hz  is  3  Volts.  There  are  no  other  components.  Each  of  these  two  components  is  located  at  
one  particular  value  of  frequency.  They  are  located  at  points  along  the  frequency  axis.  Because   v1 (t )  
is  composed  of  two  pure  tones,  its  spectrum  consists  of  two  lines  located  at  two  discrete  points  
along  the  horizontal  axis.  Music,  which  is  more  complex,  would  have  components  spread  out  along  
the  frequency  axis.  The  amplitude  spectrum  for   v1 (t )  is  labeled   V1 ( f )  and  is  shown  in  the  figure  
below.  The  spectrum  of   v2 (t )  is  the  same.  

171
 

Figure A-1: Amplitude Spectrum, for v1 (t )


 

  We  can  also  find  the  spectrum  of  a  signal  formed  from  the  product  of  two  sinusoids.  
v product (t ) = [4 cos(2π 100t )][3cos(2π 500t )] = 6 cos(2π 600t ) + 6 cos(2π 400t ) gives  the  sum  and  
difference  frequencies,  which  result  when  a  product  of  cosines  is  taken.  (Recall  the  trig  identity  for  
the  product  of  two  cosines.)  The  amplitude  spectrum  for  this  product  signal  is  shown  below.  

 
Figure A-2: Amplitude Spectrum for a Product Function
 

  Given  an  amplitude  spectrum,  the  corresponding  function  of  time  can  be  written  to  within  a  
phase  uncertainty.  If  we  assume  the  cosine  phase  for  each  component,  then  an  amplitude  spectrum  
consisting  of  a  1.5  Volt  at  300  Hz  plus  a  2.5  Volt  at  500  Hz  plus  a  3  Volt  at  700  Hz  corresponds  to
v(t ) = 1.5cos(2π 300t ) + 2.5cos(2π 500t ) + 3cos(2π 700t )  V.  

A.2  Ideal  Filtering  


  Ideal  filters  are  not  possible  to  build  in  practice  but  they  are  useful  in  simplifying  the  
concepts  when  discussing  communications.  Ideal  filters  have  vertical  sides  at  their  cutoff  frequency  
or  frequencies.  An  ideal  filter  is  essentially  a  window  which  allows  an  input  to  shine  through  within  
the  band  pass  of  the  filter  but  removes  all  components  that  fall  outside.  The  gain  of  the  filter  is  
positive  but  can  be  greater  than  or  less  than  1.  If  an  input  line  falls  within  the  window,  that  

172
frequency  component  appears  at  the  output  but  multiplied  by  the  gain  of  the  filter.  Ideal  band  pass,  
low  pass  and  high  pass  filters  are  illustrated  in  the  diagram  below.  

 
Figure A-3: Ideal Filters
 

  To  determine  the  spectrum  at  the  output  of  an  ideal  filter  knowing  the  input,  we  simply  
superimpose  the  filter  shape  over  the  input  spectrum  and  determine  which,  if  any,  frequency  
components  appear  in  the  window  and  therefore  at  the  output.  Any  input  components  outside  the  
window  are  removed  by  the  filter  and  do  not  appear  at  the  output.  The  amplitude  of  any  output  
frequency  component  is  given  by  the  amplitude  at  the  input  multiplied  by  the  filter  gain.  

  As  an  example  of  filtering,  we  will  determine  the  output  of  first  an  ideal  low  pass  and  then  
an  ideal  band  pass  filter  (each  filter,  one  at  a  time,  used  by  itself)  when  the  input  is  given  by:  

vin (t ) = 1.5cos(2π 300t ) + 2.5cos(2π 500t ) + 3cos(2π 700t ) Volts.  The  cutoff  frequency  of  the  low  
pass  filter  is  600  Hz  and  its  gain  is  0.5.  The  corner  frequencies  of  the  band  pass  filter  are  400  Hz  and  
600  Hz  and  its  gain  is  2.  First,  the  input  spectrum  and  superimposed  filter  response  are  shown  for  
the  low  pass  filter  along  with  the  output  spectrum.  

Figure A-4: Input Spectrum and Lowpass Filter Response


 

173
 

Figure A-5: Lowpass Output Spectrum


 

  The  output  spectrum  of  the  ideal  low  pass  filter  consists  of  the  two  lines,  one  at  300  Hz  and  
one  at  500  Hz,  which  are  within  the  low  pass  window.  Each  one  is  multiplied  by  0.5.  Ignoring  the  
phase  for  both  output  components,  the  corresponding  time  function  at  the  output  is  

vout (t ) = 0.5[1.5cos(2π 300t )] + 0.5[2.5cos(2π 500t )] = 0.75cos(2π 300t ) + 1.25cos(2π 500t )  Volts.  

  Next,  the  input  spectrum  and  superimposed  filter  response  are  shown  for  the  band  pass  
filter  along  with  the  output  spectrum.  Only  the  500  Hz  line  is  within  the  band  pass  window  and  it  
gets  multiplied  by  a  gain  of  two.  Thus,  the  output  function  of  time  is   vout (t ) = 5cos(2π 500t ) Volts.  

Figure A-6: Input Spectrum and Bandpass Filter Response


 

174
 
Figure A-7: Bandpass Output Spectrum
 

175
Appendix  B:  A  Typical  CW  Communication  System  
 

B.1  Introduction  
We  can  think  of  a  continuous  wave  (CW)  transmitter,  regardless  of  whether  it  is  AM  or  FM,  as  a  
device  which  modulates  a  continuous  wave  carrier  signal  with  an  information  signal.  This  
modulated  waveform  is  then  coupled  to  the  channel  using  an  appropriate  channel  matching  device.  
Figure  B-­‐1  shows  a  typical  CW  transmitter.  

 
Figure B-1: A CW Transmitter
 

  The  key  item  in  the  transmitter  is,  of  course,  the  modulator.   Besides  the  modulator,  an  
amplifier  is  necessary  because  most  transmissions  require  high  power  levels  if  they  are  to  travel  
any  distance.  For  example,  even  the  relatively  low  powered  citizen’s  band  transmitters  transmit  5  
Watts.  A  local  broadcast  station  may  transmit  a  hundred  thousand  Watts!  The  amplified,  modulated  
signal  is  now  ready  for  transmission.  To  do  so  usually  requires  a  channel  matching  device.  
Computers  talk  over  telephone  lines  using  a  modem.  A  modem  is  a  device  which  contains  both  a  
modulator  and  a  demodulator.  While  radio  and  TV  transmitters  require  an  antenna  to  transmit,  a  
light  beam  usually  requires  some  type  of  collimating  lens  prior  to  entering  a  fiber  optic  channel.  

  At  the  other  end  of  the  channel  we  find  the  receiver.  Basically  a  receiver  is  a  device  which  
extracts  the  signal  from  the  channel  and  prepares  it  for  delivery  to  the  output  processor  and  
transducer.  Since  the  received  signal  is  modulated  and  usually  very  weak  (attenuated)  we  would  
expect  a  receiver  to  contain  an  amplifier  and  demodulator.  

B.2  A  Citizen  Band  (CB)  Transceiver  


  The  popularity  of  the  citizen’s  band  has  been  made  possible  through  low-­‐cost,  reliable,  and  
portable  equipment.  The  minimal  licensing  requirements,  the  55  mph  speed  limit,  and  movies  such  
as  Smokey  and  the  Bandit  have  also  contributed  to  their  popularity.  The  FCC  has  allocated  a  small  
portion  of  the  11  meter  band  to  public  use.  The  spectrum  between  26.9  MHz  and  27.3  MHz  has  been  
divided  into  40  CB  channels  located  10  KHz  apart  

177
  The  CB  transceiver  is  a  single  system  with  dual  functions.  It  can  operate  either  as  a  
transmitter  or  a  receiver,  sharing  many  of  the  same  components  and  conserving  costs.  The  receiver  
is  usually  of  superhet  design  similar  to  the  one  studied  earlier  (note:  modern  CB’s  are  SSB-­‐SC  vice  
AM).  To  separate  the  closely  spaced  channels,  accurate  local  oscillators  are  required.  Originally,  
crystal  controlled  oscillator  were  used.  These  rock  solid  devices  based  their  accuracy  on  a  vibrating  
crystal  (like  today’s  quartz  timepieces).  Such  systems  were,  and  still  are,  expensive.  The  
introduction  of  a  unique  device  known  as  a  phase-­‐Locked-­‐Loop  (PLL)  has  made  the  concept  of  one  
crystal  oscillator  per  channel  almost  obsolete,  especially  in  the  realm  of  cost.  A  PLL  system  costs  
about  one-­‐third  that  of  a  comparable  crystal  controlled  device  with  very  little  loss  in  performance.  
The  transmitter  for  a  CB  system  is  conceptually  simple.  A  typical  configuration  is  shown  in  Figure  
B-­‐2.  

Figure B-2: A CB transmitter

  A  local  RF  oscillator  (different  from  the  receiver’s  local  oscillator)  signal  is  amplified  and  
modulated  by  the  information  signal-­‐  an  amplified  voice  signal  from  the  microphone.  The  
modulated  signal  is  input  to  the  antenna  for  radiation  into  space.  

  Most  CB  transmitter  and  receiver  sections  share  the  antenna  and  the  power  supply  
subsystems.  In  more  complex  systems  such  as  the  one  for  the  Craig  Model  4102  CB  Transceiver,  a  
synthesizer  oscillator  and  an  audio  amplifier  are  also  shared.  Can  you  follow  and  describe  the  
function  of  each  block  in  Figure  B-­‐3?  

178
 

Figure B-3: Block Diagram of a Craig Model 4102 CB Transceiver


 

179
Appendix  C:  The  Channel  
 

C.1  Introduction  
  The  channel  of  a  communication  system  is  the  medium  through  which  the  information  
flows  from  the  transmitter  to  the  receiver.  The  channel  can  take  many  forms,  but  three  encompass  
the  greatest  majority  of  all  communications.  These  are:  wires  or  cables  known  collectively  as  
transmission  lines;  free  space  which  includes  the  earth’s  atmosphere;  and  fiber-­‐optics,  a  small  
diameter  transparent  filament  used  to  propagate  light.  

  Another  channel  used  over  very  short  distances  (especially  in  radar  systems)  is  the  
waveguide,  a  hollow  rectangular,  circular,  or  elliptical  metal  tube.  Just  what  medium  is  used  in  a  
particular  communication  system  will  depend  on  a  number  of  factors  such  as  cost,  mobility,  
reliability,  channel  capacity,  distance  to  transverse,  signal  frequency,  noise  environment,  signal  
bandwidth,  and  signal  security.  

  Transmission  lines  are  used  when  the  distance  over  which  the  transmission  is  required  is  
short  and  the  spectrum  of  the  communication  is  below  a  few  hundred  KHz.  Transmission  lines  offer  
a  very  reliable  and  relatively  secure  channel.  For  example,  the  telephone  and  telegraph  systems  
were  originally  totally  on  wires  and  cables.  Today  they  still  are  in  a  large  part,  although  every  type  
of  channel  is  used  somewhere  in  the  current  complex  world-­‐wide  system.  When  the  spectrum  of  
the  information  exceeds  several  hundred  KHz  and  transmission  exceeds  a  few  miles  it  generally  
becomes  more  efficient  and  less  costly  to  transmit  the  information  via  radio  waves  in  free  space.  
Because  it  is  so  efficient  to  transmit  information  over  free  space,  it  has  become  the  primary  means  
of  all  communications.  We  shall  discuss  this  channel  at  length  later.    The  following  table  points  out  
how  much  more  efficient.  

 
Type of Channel Required Transmitter Power
 

 
Transmission Line 10600MWatts
Waveguide 1020 MWatts
  Free Space 10 −7 MWatts

Table C-l. Power needed to transmit a 109 Hz signal at a distance of 30 miles so that a signal level of 10 −9
Watts arrives at the receiver
 

  A  newcomer  to  the  channel  market  is  fiber-­‐optics.  The  invention  of  the  laser  in  the  late  
fifties  provided  us  with  an  electromagnetic  oscillator  which  operates  at  optical  frequencies.  The  
output  of  a  laser  is  an  electromagnetic  sinusoidal  wave.  Just  like  its  lower  frequency  counterpart,  it  
can  be  modulated  and  used  to  carry  information.  The  significance  of  using  a  laser  is  that  it  can  be  
multiplexed  to  carry  many  more  signals  that  any  other  type  of  carrier.  The  reason  is  simple.  

181
Suppose  audio  information  needs  to  be  transmitted.  Table  C-­‐2  compares  the  number  of  20  KHz  
audio  bandwidth  signals  that  can  be  carried  at  various  frequencies.  An  assumption  made  is  that  
each  frequency  has  a  useable  bandwidth  of  1%  of  its  center  frequency.  

Spectrum Band Center Frequency Bandwidth Number of Signals


 
Radio Waves
  MF 1 MHz 0.01 MHz 0.5
HF 10 MHz 0.1 MHz 5
  VHF 100 MHz 1 MHZ 50
UHF 1000 MHz 10 MHz 500
  Radar Waves
 
SHF 10 4 MHz 10 2 MHz 5 × 103
EHF 105 MHz 103 MHz 5 × 104
  Laser 108 MHz 106 MHz 5 × 107
Table C-2: Number of 20 KHz bandwidth signals that can be carried at various frequencies
 

  From  the  above  table  it  is  evident  that  the  higher  the  frequency,  the  larger  the  number  of  
channels  that  can  be  carried.  Lasers  offer  the  ability  to  greatly  reduce  the  number  of  cables  needed  
to  carry  the  enormous  number  of  channels  necessary  for  telephone  conversations  between  major  
US  cities.  The  savings  in  copper  alone  will  more  than  pay  for  the  cost  of  installing  fiber-­‐optic  links  
between  major  communication  centers.  The  first  commercial  televised  picture  to  be  carried  on  
fiber-­‐optics  was  of  the  Winter  Olympics  at  Lake  Placid,  New  York  in  1980.  Bell  telephone  has  
installed  a  major  fiber-­‐optics  link  in  Chicago  and  more  are  being  installed.  

C.2  Propagation  of  Signals  in  Free  Space    


  A  brilliant  British  physicist,  James  Clerk  Maxwell,  theorized  the  existence  of  electromagnetic  
waves  in  1855.  Although  Maxwell  drew  from  the  efforts  of  Coulomb,  Faraday,  Gauss,  Ampere  and  
other,  his  equations  represent  a  scientific  achievement  of  unparalleled  proportions.  It  is  through  his  
equations  that  we  have  an  understanding  of  the  behavior  of  electromagnetic  radiation.  

  An  electromagnetic  wave  is  composed  of  a  time-­‐varying  electric  field  and  a  corresponding  
time  varying  magnetic  field.  These  fields  are  interdependent;  that  is,  one  cannot  exist  without  the  
other.  A  principal  property  of  electromagnetic  fields  is  that  they  can  propagate  in  space.  The  
velocity  with  which  an  electromagnetic  wave  propagates  is  very  nearly  equal  to  the  speed  of  light  
(light  itself  being  an  electromagnetic  wave)  and  equals  the  speed  of  light,   3 × 108 m/sec,  in  a  perfect  
vacuum.  We  have  already  seen  that  the  product  of  the  frequency, f ,  of  a  particular  electromagnetic  
wave  and  this  wavelength, λ  ,  is  the  speed  of  light.  The  wavelength  of  a  particular  electromagnetic  
wave  determines  the  efficiency  with  which  the  wave  is  transmit  and  received  at  the  transmitting  
and  receiving  antennas.  It  also  plays  an  important  part  in  the  wave’s  absorption  or  reflection.    

182
  Electromagnetic  waves  travel  in  a  straight  line  away  from  the  originating  source  much  like  
ripples  in  a  pond  radiate  away  from  the  point  where  a  pebble  is  dropped  in.  As  the  waves  spread,  
they  become  weaker  and  weaker,  and  to  an  observer  located  at  a  fair  distance  from  the  source,  the  
waves  appear  not  only  weak  in  amplitude  but  also  parallel  to  each  other.  When  these  waves  
encounter  a  different  medium  they  can  be  bent  (refracted),  reflected,  scattered,  or  greatly  reduced  
in  amplitude  (attenuated).  For  example,  light  waves  are  obviously  bent  when  entering  water  and  
the  light  is  greatly  attenuated  by  the  water.  In  this  section  we  will  concentrate  our  study  of  
electromagnetic  radiation  on  the  portion  of  the  electromagnetic  spectrum  most  used  for  
communication,  radio  frequency  waves.  

C.3  Radio  Waves    


  As  illustrated  in  the  Figure  C-­‐1,  the  entire  electromagnetic  spectrum  occupies  an  extensive  
range  of  frequencies.  Although  there  are  similarities  among  all  electromagnetic  phenomena  (such  
as  free  space  propagation  velocity),  certain  characteristics  or  functions  are  confined  to  definite  
bands  within  the  spectrum.  These  major  subdivisions  are  identified  in  the  figure  and  are  often  
classified  as  complete  spectra  in  themselves  (e.g.,  the  optical  spectrum  refers  to  the  range  of  visible  
light).  Note  that  at  some  frequencies  these  various  spectra  overlap.  Thus  a  radiated  wave  with  a  
frequency  of  1020  Hz  exhibits  behavior  characteristic  to  both  X-­‐rays  and  gamma  rays.  This,  of  
course,  is  because  we  are  dealing  with  a  continuum  and  the  spectral  bounds  are  necessarily  chosen  
somewhat  arbitrarily.  Some  sources,  such  as  the  sun,  radiate  several  types  of  electromagnetic  
energy  and  such  radiations  have  extremely  broad  relative  bandwidths.  Note  that  frequency  and  
wavelength  in  the  figure  are  inversely  related  in  accordance  with f λ = c .  Thus  the  values  for  these  
parameters  pertain  to  propagation  through  free  space.  

 
Figure C-1: Electromagnetic and expanded radio spectra

183
  In  the  study  of  signal  propagation,  our  primary  interest  is  in  radio  waves.  These  waves  
occupy  that  portion  of  the  electromagnetic  spectrum  from  10  KHz  to  1000  GHz  (G  =  giga  = 109 ).  An  
expanded  radio  spectrum  is  also  shown  above.  The  ITU  (International  Telegraphic  Union)  
designation  of  frequency  bands  and  the  frequency  ranges  for  some  of  the  more  common  uses  of  
radio  waves  are  indicated.  Since  the  transmitted  signal  in  a  communication  system  usually  has  a  
narrow  relative  bandwidth,  the  propagation  characteristics  are  determined  almost  exclusively  by  
the  carrier  wave.  Thus,  in  this  section  we  will  extend  the  properties  of  the  carrier  wave  (such  as  
frequency  and  wavelength)  to  the  entire  transmitted  signal.  

  As  a  radio  wave  propagates  over  its  transmission  path,  several  things  can  alter  it.  It  can  be  
attenuated,  reflected  or  scattered.  Attenuation  of  a  radiated  signal  is  caused  primarily  by  the  
spreading  out  of  the  wave  as  it  propagates  from  its  source.  As  a  result,  only  a  small  fraction  of  the  
transmitted  power  is  intercepted  by  a  receiving  antenna.  Additionally,  there  is  a  small  loss  of  signal  
power  due  to  interaction  of  the  wave  with  the  medium  through  which  the  wave  passes.  If  a  wave  
strikes  the  boundary  of  a  conductive  medium,  it  may  be  reflected  and/or  refracted.  Reflection  of  
incident  waves  increases  with  the  conductivity  of  the  medium  such  that  a  perfect  conductor  results  
in  total  reflection.  Ionized  portions  of  the  atmosphere  are  conductive  and  therefore  can  cause  a  
radio  wave  to  be  reflected  or  refracted.  Scattering  of  a  propagating  wave  may  also  occur  if  there  are  
non-­‐homogeneities  within  the  medium.  The  characteristics  of  both  the  medium  and  the  radio  wave  
determine  whether  the  wave  will  be  significantly  altered  by  scattering,  reflection,  refraction,  or  
attenuation.  The  most  significant  wave  characteristic  is  the  frequency.  Likewise  the  direction  of  
propagation  may  have  to  be  considered  since  reflection  and  refraction  are  dependent  upon  the  
incident  angle  at  which  a  wave  strikes  a  conductive  medium.  A  wave  may  reach  a  particular  point  
by  any  of  several  paths.  A  wave  can  be  reflected  or  refracted  or  scattered  several  times  and  arrive  at  
the  same  point  it  may  reach  by  line  of  sight  (although  the  waves  may  arrive  at  slightly  different  
times).  

  Within  the  earth’s  atmosphere  there  are  two  regions  or  layers  that  greatly  affect  the  
propagation  of  radio  waves.  The  first  of  these  is  the  troposphere,  which  extends  from  the  surface  to  
about  33,000  feet.  Clouds  are  formed  and  most  weather  phenomena  occur  in  this  region.  Within  the  
troposphere,  there  are  sharp  discontinuities  in  temperature,  water  vapor  content  and  air  density.  
The  resulting  blobs  of  air  can  scatter  radio  waves.  The  second  region  is  the  ionosphere,  which  
consists  of  several  ionized  layers  at  heights  between  30  and  250  miles.  These  ionized  layers  are  
formed  by  radiant  energy  from  the  sun  and  display  variations  in  both  ion  density  and  height.  
Generally,  the  ionosphere  is  denser  and  lower  during  the  day  and  in  summer.  At  night  and  during  
the  winter,  ion  density  decreases  whereas  the  bottom  of  the  ionosphere  rises.  The  ionosphere  is  
also  affected  by  sunspot  activity  which  occurs  in  11  year  cycles.  As  might  be  expected,  the  
ionosphere  is  quite  unstable  and  is  not  uniform  around  the  earth  at  anyone  time.  Since  the  
ionosphere  is  a  conducting  medium,  it  may  reflect  and/or  refract  radio  waves  which  strike  it.  

C.4  Propagation  of  Radio  Waves    


  The  propagation  characteristics  of  radio  waves  vary  greatly  from  one  frequency  to  another.  
This  is  not  surprising  when  it  is  realized  that  wavelengths  within  the  radio  spectrum  range  from  
30,000  meters  (about  19  miles)  to  0.3  millimeters  (about  0.012  inches).  There  are  four  primary  

184
ways  in  which  radio  wave  propagate.  They  are  direct  or  line  of  sight  (LOS),  surface  wave,  sky  wave,  
and  forward  scatter.  Each  of  these  involves  a  propagation  path  and,  as  previously  mentioned,  it  is  
possible  for  a  signal  to  be  transmitted  over  several  paths  simultaneously.  However,  one  path  in  
multipath  transmission  is  usually  predominant  and  produces  a  much  stronger  received  signal  than  
the  others.  Sometimes  it  is  necessary  to  control  the  characteristics  of  the  transmitted  signal  so  that  
multipath  interference  is  eliminated.  

C.4.1  Line  of  Sight  (LOS)  


  All  electromagnetic  energy  will  propagate  by  line  of  sight,  although  atmospheric  conditions  
will  attenuate  some  frequencies  more  than  others.  As  can  be  seen  from  the  following  figure,  LOS  
communications  on  earth  are  limited  by  the  curvature  of  the  earth  and  the  heights  of  transmitting  
and  receiving  antennas.  (Of  course  transmitter  power  is  also  a  limiting  factor.)  Thus,  in  Figure  C-­‐2  
the  signal  transmitted  from  point  T  reaches  the  receiving  antennas R1 ,  and R2 .  However, R3  is  not  in  
the  LOS  path  of  the  signal  and  hence  cannot  intercept  any  of  the  radio  energy.  

  For  communication  and  data  telemetry  between  earth  and  space  vehicles,  LOS  propagation  
is  used.  As  a  vehicle  begins  its  reentry  into  the  earth’s  atmosphere,  its  surfaces  are  heated  by  
friction.  The  heated  surfaces  ionized  the  surrounding  air  and  plasma,  many  times  denser  than  the  
ionosphere,  is  formed  around  the  vehicle.  The  ion  density  is  so  great  that  the  plasma  becomes  an  
almost  perfect  conductor  for  a  wide  band  of  radio  frequencies  and  these  waves  cannot  propagate  
through  it.  This  blackout  condition  persists  until  the  vehicle  has  decelerated  to  a  point  where  the  
heat  caused  by  friction  is  not  intense  enough  to  cause  ionization  of  the  air.  We  may  view  the  plasma  
as  a  line  of  sight  obstruction  in  the  communication  channel.  Short  distance,  secure  military  
communication  occurs  through  use  of  LOS  microwave  links.  

 
Figure C-2: Line of sight propagation
 

C.4.2  Surface  Wave  


  In  surface  wave  propagation,  the  radio  wave  travels  from  the  antenna  to  the  receiving  
antenna  with  the  bottom  of  the  wave  touching  the  ground.  This  is  illustrated  in  Figure  C-­‐3.  The  
surface  wave  is  actually  guided  along  and  around  the  surface  of  the  earth,  and  distances  greater  
than  line  of  sight  can  be  obtained.  In  this  guiding  process,  minute  currents  are  induced  in  the  
ground  directly  beneath  the  surface  wave.  Although  the  ground  is  a  fairly  good  conductor,  it  does  

185
have  some  resistance  and  the  energy  required  for  these  currents  to  flow  is  absorbed  from  the  wave.  
As  frequency  is  increased,  the  losses  due  to  the  resistivity  of  the  ground  also  increase  and  the  
surface  wave  is  greatly  attenuated.    

  Surface  waves  are  very  effective  for  signal  propagation  in  the  VLF  and  LF  bands.  However,  
as  frequency  is  increased  through  the  MF  band,  this  effectiveness  decreases  rapidly  and  surface  
waves  are  not  used  above  3  MHz.  Commercial  AM  broadcasting  stations  typically  use  surface  waves  
to  transmit  their  signals.  All  electromagnetic  waves  have  the  ability  to  penetrate  into  a  conductor  to  
a  small  fraction  of  their  wavelength.  For  all  but  very  low  frequencies  (VLF)  this  depth  is  
inconsequential.  At  VLF  a  radio  wave  can  penetrate  even  ocean  water  to  a  depth  of  several  meters.  
The  navy  uses  this  phenomenon  to  communicate  with  its  submarines.  Unfortunately,  VLF  
transmission  requires  very  long  antennas  (miles)  to  transmit  and  receive  waves.  Hence  there  is  
always  a  trade-­‐off  between  lowest  frequency  and  shortest  antenna.  

 
Figure C-3: Surface wave propagation
 

C.4.3  Skywave  
Radio  energy  reflected  or  refracted  from  the  ionosphere  back  to  the  earth  such  as  illustrated  in  
Figure  C-­‐4  is  known  as  a  sky  wave.  To  understand  how  the  ionosphere  affects  radio  waves  of  
different  frequencies,  let  us  consider  it  as  a  huge  sieve  surrounding  the  earth.  

186
 
Figure C-4: Sky wave propagation
 

  Whether  or  not  a  wave  passes  through  this  sieve  depends  partially  upon  the  relative  
dimensions  of  the  wavelength  and  of  the  mesh  openings.  Thus  radio  energy  with  a  long  wavelength  
(low  frequency)  is  more  likely  to  be  reflected  back  to  earth  than  that  with  a  short  wavelength  (high  
frequency).  Additionally,  the  angle  of  incidence  ( α in  Figure  C-­‐4)  is  the  complement  of  the  angle  of  
incidence)  with  which  a  wave  strikes  the  ionosphere  must  be  considered.  In  general,  the  smaller  the  
angle, α ,  the  greater  the  probability  that  the  wave  will  be  reflected.  However,  if α  is  made  too  
small,  the  wave  will  effectively  remain  in  the  ionosphere  and  not  be  returned  to  earth.  By  now  it  
should  be  obvious  that  there  can  be  a  trade-­‐off  between  this  angle  and  frequency.  That  is,  for  a  
given  angle,  there  is  some  maximum  frequency  that  can  be  used  for  sky  wave  propagation.  
Likewise,  for  a  given  frequency  (within  certain  limits),  there  is  some  maximum  angle  that  will  
produce  a  sky  wave.  Notice  that α  determines  the  distance  from  the  transmitter  that  the  reflected  
waves  may  be  received.  Recall  from  our  discussion  in  the  previous  sections  that  the  ionospheric  
sieve  is  neither  uniform  nor  stable.  Thus,  frequencies  and  incident  angles  are  dependent  upon  
season  and  time  of  day.  Maximum  useable  frequencies  usually  lie  in  the  HF  band;  waves  whose  
frequencies  are  above  this  are  refracted  slightly  by  the  ionosphere  but  propagate  through  it.  A  
peculiarity  of  the  ionosphere  is  that  its  lower  layers  readily  absorb  energy  in  the  MF  band.  Thus  sky  
waves  in  this  band  are  possible  only  at  night  when  the  lower  ionospheric  layers  dissipate.  

C.4.4  Forward  Scatter  


  When  a  radiated  signal  strikes  the  discontinuous  blobs  of  air  in  the  troposphere,  it  is  
scattered  in  various  directions,  such  as  shown  in  Figure  C-­‐5.  Some  of  this  scattering  is  in  the  
forward  direction  and  the  resulting  radio  signal,  although  relatively  weak,  can  be  received  at  a  point  
that  is  beyond  the  horizon  from  the  transmitter.  Although  the  scatter  is  dependent  upon  
atmospheric  conditions,  it  is  possible  to  achieve  reliable  communications  using  highly  elaborate  
and  sensitive  transmitting  and  receiving  equipment.  Tropospheric  forward  scatter  is  effective  in  the  
VHF,  UHF,  and  SHF  bands.  Much  in-­‐theater  military  communication  is  accomplished  using  these  
types  of  propagation.  

187
     

Figure C-5: Forward scatter propagation


 

C.4.5  Summary  
  The  following  remarks  summarize  the  propagation  characteristics  and  usefulness  of  radio  
waves  within  the  various  bands  of  frequencies.  Line  of  sight  propagation,  although  not  mentioned  
specifically  for  all  bands,  is  possible  throughout  the  entire  spectrum.  Of  course  the  limited  distance  
obtainable  by  LOS  is  the  main  drawback  of  this  type  of  propagation.  Line  of  sight  from  a  20  foot  
height  on  a  flat  portion  of  the  earth’s  surface  is  only  about  5.5  miles,  and  other  methods  of  
propagation  are  often  used  to  overcome  this  restriction.  

  VLF  and  LF  (very  low  frequencies  and  low  frequencies)  -­‐  At  these  lower  frequencies,  surface  
waves  are  attenuated  very  little  and  may  be  used  for  signal  propagation  of  a  thousand  miles  or  
more.  This  maximum  distance  gradually  decreases  with  increasing  frequency  and  is  about  400  
miles  at  300  KHz.  The  sky  wave  does  exhibit  slight  fluctuations  with  changes  in  the  ionosphere  but  
is  still  fairly  reliable.  It  can  be  used  for  communication  over  distance  from  about  500  miles  to  8000  
miles  in  the  LF  band.  In  the  VLF  range,  the  combination  of  the  surface  wave  and  sky  wave  
mechanisms  makes  possible  world-­‐wide  signal  propagation  with  radiated  power  levels  of  about  1  
MW.  

  MF  (medium  frequencies)  -­‐  In  this  band  the  maximum  distance  for  surface  wave  
propagation  varies  from  about  400  miles  at  300  KHz  to  about  20  miles  at  3  MHZ.  Ionospheric  
absorption  of  electromagnetic  energy  in  this  band  (maximum  absorption  occurs  at  1.4  MHZ)  makes  
sky  wave  propagation  impossible  during  the  day.  At  night  sky  waves  furnish  reception  at  distances  
from  about  100  to  3000  miles.  

  HF  (high  frequencies)  -­‐  The  attenuation  of  surface  waves  above  3  MHz  is  so  great  that  the  
surface  wave  has  effectively  no  use  for  communication.  Sky  waves  are  used  extensively  in  this  band  
and  their  behavior  is  mostly  governed  by  ionospheric  conditions.  Although  sky  wave  propagation  is  
not  always  reliable,  it  is  possible  over  distances  of  12,000  miles  and  more.  For  distances  such  as  
this,  frequencies  from  4  to  20  MHz  have  proven  most  effective.  

188
VHF  (very  high  frequencies)  -­‐  Although  sky  waves  may  occur  at  lower  VHF  frequencies,  their  
reliability  is  so  poor  that  they  are  virtually  useless  for  communication.  The  predominant  form  of  
propagation  in  this  band  is  line  of  sight.  The  effectiveness  of  forward  scatter  becomes  increasingly  
important  as  frequencies  reach  50  MHz  and  above.  

UHF  and  SHF  (ultrahigh  frequency  and  super  high  frequency)  -­‐  Line  of  sight  propagation  is  widely  
used  at  these  frequencies  since  excellent  low-­‐static  reception  is  possible.  Forward  scatter  ranges  of  
a  few  hundred  miles  can  be  realized  up  to  about  10  GHz.  Most  applications  use  frequencies  well  
below  this.  

EHF  (extra  high  frequency)  -­‐  Direct  radio  waves  at  these  frequencies  attenuate  rapidly  in  space  but  
the  short  wavelengths  permit  very  precise  measurements  in  uses  such  as  radar.  Most  applications  
of  propagated  EHF  energy  are  in  the  experimental  stage.  Figure  C-­‐6  summarizes  which  frequencies  
of  the  electromagnetic  spectrum  are  best  suited  to  the  four  types  of  propagation  of  radio  waves.

C.5  Multiple  Path  Propagation  and  Skip    


  As  shown  in  Figure  C-­‐7,  it  is  possible  for  a  signal  to  be  transmitted  to  a  receiver  via  a  
combination  of  different  types  of  propagation.  This  is  called  multiple-­‐path  propagation.  In  the  case  
of  Figure  C-­‐7,  the  signal  arriving  at  the  receiving  antenna  consists  of  a  direct  (LOS)  wave,  a  surface  
wave  and  a  ground-­‐reflected  wave.  These  signals  will  not  be  in  phase  and  this  will  result  in  mutual  
interference.  Just  how  detrimental  this  interference  might  be  depends  upon  the  wave  length,  
distance  and  terrain  involved.  At  commercial  AM  broadcasting  frequencies,  the  wavelength  is  long  
enough  so  that  a  small  phase  difference  caused  by  time  delay  presents  no  great  problem.  However,  
even  a  small  phase  difference  in  a  commercial  television  signal  which  is  transmitted  on  a  higher  
carrier  frequency  produces  annoying  interference  that  appears  as  video  ghosts.  The  planned  
propagation  of  television  signals  in  LOS.  To  eliminate  any  possible  ground  wave  propagation,  the  
signals  are  horizontally  polarized  and  the  direct  wave  is  usually  the  only  one  received.  There  are  
times,  however,  when  ghosts  appear  as  a  result  of  a  television  signal  being  reflected  from  the  
ground,  clouds,  airplanes,  buildings,  etc.  

 
Figure C-6: Propagation within the radio spectrum
 

189
 

Figure C-7: Multiple path propagation


 

  Another  phenomenon  of  importance  occurs  when  transmission  of  a  signal  is  accomplished  
by  sky  wave  propagation.  At  the  point  where  a  sky  wave  returns  to  earth,  a  very  strong  signal  can  
be  detected.  However,  between  this  point  and  the  transmitter,  there  is  essentially  no  energy  from  
the  sky  wave  at  all.  The  distance  from  the  transmitting  antenna  to  the  spot  where  the  reflected  
wave  strikes  the  earth  is  called  the  skip  distance.  At  all  points  less  than  the  skip  distance  from  the  
transmitting  antenna,  only  that  portion  of  the  signal  propagated  by  surface  wave  (and  of  course  
LOS)  can  be  received.  Especially  in  the  HF  band,  where  surface  wave  propagation  is  somewhat  less  
than  20  miles,  there  is  often  a  considerable  distance  in  which  essentially  no  radiated  energy  from  
either  the  surface  wave  or  the  sky  wave  is  present.  This  region  is  called  the  skip  zone.  

  At  frequencies  in  the  MF  band,  the  surface  wave  often  propagates  further  than  the  skip  
distance  and  hence  there  is  no  skip  zone.  For  all  frequencies,  severe  fading  of  the  signal  can  occur  at  
point  beyond  the  skip  distance.  This  fading  is  caused  by  mutual  interference  between  multiple-­‐hop  
sky  waves  or  between  sky  and  surface  waves.  

190
Figure C-8: Skip distance and skip zone
 

191
Appendix  D:  Overview  of  the  USNA  SATCOM  Communication  
System  
 

  The  YP  SATCOM  communications  system  provides  a  radio  link  between  the  Yard  Patrol  Craft  (YP)  and  
the  Satellite  Earth  Station  in  Rickover  Hall,  room  122.  This  system  is  used  primarily  for  the  reporting  of  YP  
location  and  for  short  messages  between  the  YP’s  and  the  USNA  facility.  This  system  involves  a  number  of  
concepts  and  techniques  discussed  in  the  communications  portion  of  our  EE  course.  The  primary  mode  of  
transmission  is  by  HF  at  4,  6,  and  12  MHz  as  shown  in  Figure  D-­‐l  below.  

Figure D-l: HF YP to USNA Communications


 

  HF  carriers  can  reflect  from  the  earth’s  ionosphere  and  can  therefore  carry  radio  messages  over  the  
horizon.  Because  the  condition  of  the  ionosphere  varies,  the  communications  results  vary.  Use  of  three  different  
carriers  enhances  the  chance  that  one  of  them  will  work  at  anyone  time.  VHF  radio  is  propagated  along  the  line  
of  site  and  can  be  used  for  ship  to  ship  communications  within  a  squadron.  The  YP’s  can  also  use  a  satellite  link  
to  communicate  with  the  Naval  Academy  but  satellite  time  is  limited  and  is  therefore  not  the  usual  mode.  

  The  primary  function  of  the  HF  packet  radio  communications  system  is  for  location  reporting  by  the  
YP’s.  Each  YP  determines  its  position  regularly  and  reports  back  to  the  USNA.  Position  is  determined  by  
traditional  methods  or  by  use  of  GPS.  The  use  of  GPS  (Global  Positioning  System)  is  rapidly  becoming  the  
standard  method.  It  uses  very  inexpensive  equipment  to  determine  location  by  receiving  and  processing  signals  
from  GPS  satellites.  This  positioning  system  is  very  accurate  and  will  soon  find  widespread  use  in  automobiles.  
The  HF  packet  radio  link  between  YP’s  and  the  USNA  is  also  used  to  convey  other  information  such  as  weather  
conditions,  equipment  problems,  etc.  

  The  packet,  in  packet  radio,  refers  to  a  group  of  binary  l’s  and  0’s  used  to  convey  a  short  message,  in  
other  words,  digital  communications.  Amateur  radio  operators  have  been  using  packet  radios  for  years.  Modern  
electronics  has  made  digital  radio  transmission  efficient,  inexpensive,  flexible  and  more  and  more  widespread.  
Robert  Bruninga,  retired  CDR  U.S.  Navy,  has  been  responsible  for  adopting  and  adapting  the  packet  radio  
concept  to  the  needs  of  the  YP  program.  

  The  heart  of  a  packet  radio  system  is  the  Terminal  Node  Controller  (TNC).  The  TNC  is  shown  in  relation  
to  the  radio  and  data  source  in  Figure  D-­‐2  below.  

   
193
 
Figure D-2: Bi-directional Data Flow in Packet Radio
 

  When  data,  such  as  current  YP  location,  is  entered  on  the  computer  keyboard,  a  steam  of  bits  is  fed  into  
the  TNC  from  the  computer.  The  TNC  then  processes  the  binary  data  with  the  necessary  destination  and  source  
addresses  and  the  necessary  protocol  and  error  correcting  codes.  The  TNC  then  acts  as  a  modem  representing  
the  binary  information  as  two  audio  tones,  one  at  1600  Hz  and  the  other  at  1800  Hz.  Representation  of  1’s  and  
0’s  as  two  different  tones  is  called  frequency  shift  keying  (FSK)  and  is  the  same  technique  used  in  computer  
modems.  The  sequence  of  two  tones  is  fed  into  the  modulator  input  of  a  packet  radio.  In  the  radio,  the  two  audio  
tones  are  amplitude  modulated  onto  an  HF  carrier  (4,  6,  and  12  MHz  frequencies  are  used  by  the  YP’s).    

  Single  side  band  modulation  is  used  to  conserve  bandwidth  and  allow  as  many  channels  as  possible  to  
operate  in  the  same  area.  The  maximum  baud  rate  for  this  technique  is  about  300  baud  (about  30  characters  per  
second).  The  baud  rate  for  HF  carriers  is  limited  by  the  HF  mode  of  propagation  through  the  earth’s  
atmosphere.  The  HF  carrier  is  reflected  from  the  ionosphere  along  its  path  from  transmitter  to  receiver,  but  the  
ionosphere  does  not  act  like  a  mirror  with  a  sharply  defined  surface.  Rather,  the  reflection  takes  place  over  a  
certain  depth  which  tends  to  form  ghost  1’s  and  0’s.  This  means  the  bits  can  not  be  spaced  too  close  together  or  
they  will  begin  to  overlap  and  interfere  with  one  another.  Using  VHF  or  UHF  along  the  line  of  sight  (for  example,  
to  a  satellite)  this  problem  does  not  exist  and  higher  bit  rates  are  possible.  A  baud  rate  of  300  bits  per  second  is  
too  slow  for  real  time  voice  but  OK  for  data.  It  should  be  pointed  out  that  a  voice  signal  could  be  digitized  at  a  
high  rate  and  then  buffered  before  transmission  over  packet  radio  at  a  much  slower  rate.  This  would  require  
buffering  again  at  the  receiver  before  playback.  The  sampling,  quantization  and  coding  of  an  analog  signal  does  
not  occur  in  this  packet  radio  system  because  the  input  data  from  a  computer  is  already  in  digital  form.  

  Data  is  transmitted  and  received  in  small  packages,  called  packets,  of  about  80  characters  at  a  time.  The  
amount  of  overhead  is  about  20  characters,  leaving  about  60  for  information.  Overhead  includes  the  addresses  
of  the  source  and  destination  and  error  detecting  bits.  If  a  packet  is  made  more  than  one,  80  character  line,  the  
probability  of  making  an  error  starts  to  become  substantial.  Much  less  than  80  characters  and  not  much  can  be  
said,  so  the  optimal  is  about  80.  

  Each  packet  transmitted  by  a  YP  contains  its  most  current  position,  speed  and  heading.  The  position  can  
be  determined  using  GPS  or  by  some  other  means  and  then  manually  keyed  into  the  computer.  One  of  the  next  
improvements  to  the  system  will  be  to  automate  the  position  determination  and  reporting  by  directly  
interfacing  the  GPS  receiver  to  the  computer  and  TNC.  If  a  YP  is  out  of  radio  range  from  the  USNA,  it  is  still  
possible  to  communicate  by  using  TNC’s  on  other  YP’s  as  relays.  Typically,  the  relay  TNC’s  are  on  YP’s  located  
closer  to  the  Naval  Academy  and,  therefore,  within  radio  range.  The  relay  function  can  be  automatic  with  no  
operator  intervention  if  the  correct  routing  addresses  are  included  from  the  source.  This  potential  for  increase  
in  range  is  one  advantage  of  a  digital  packet  radio  system  over  traditional  direct  voice  HF  communications.  
194
  When  a  YP  is  receiving  a  packet  of  information  the  process  of  transmission  described  above  is  reversed.  
The  radio  demodulates  the  two  tones,  representing  binary  data,  from  the  HF  carrier  and  passes  them  along  to  
the  TNC  which  in  turn  converts  them  into  voltage  levels  which  a  computer  can  understand  as  1’s  and  0’s.  

  What  does  the  TNC  do  and  how  does  it  do  it?  A  TNC  uses  an  on  board  microprocessor  to  keep  track  of  all  
that  is  required.  The  computing  power  of  a  TNC  confers  many  advantages  to  packet  radio  communication,  such  
as  error  detection.  A  calculation  is  done  on  every  packet  received.  If  the  answer  is  not  correct,  it  means  the  data  
was  probably  corrupted  and  an  acknowledgment  of  receipt  is  not  sent  back  to  the  transmitter.  This  means  the  
transmitter  will  try  again  until  the  packet  is  received  error  free.  This  is  another  advantage  of  packet  radio.  A  
complete  description  of  the  protocol  used  by  a  TNC  is  very  long  and  complicated.  Listed  below  are  some  of  the  
functions  performed  by  the  TNC.  I  think  you  will  begin  to  appreciate  why  a  computer  is  needed  inside  the  TNC.  

1.  The  TNC  must  monitor  both  its  data  port  connected  to  the  terminal  (is  data  going  to  be  transmitted?)  
and  its  connection  to  the  radio  (is  data  being  received?).  

2.  The  TNC  must  check  the  addresses  of  all  packets  as  they  are  received.  Only  those  with  the  correct  
address  will  be  processed.  If  the  intended  recipient  is  somewhere  else,  those  packets  must  be  ignored,  
but  if  a  relay  of  a  packet  is  requested,  the  TNC  must  determine  that  fact  and  then  retransmit  the  packet  
to  pass  it  along  to  its  intended  destination.  

3.  The  TNC  must  acknowledge  packets  received  correctly  and  complain  about  those  in  error.  The  TNC  
must  send  its  own  packets,  keeping  track  of  those  that  have  been  acknowledged  and  those  that  haven’t.    

4.  While  it  is  transmitting,  the  TNC  must  know  whom  it  is  talking  to  and  tell  other  TNC’s  that  it  is  busy,  
and  to  try  again  later.    

5.  The  TNC  must  listen  to  the  radio  and  wait  to  transmit  if  another  signal  is  on  the  air.    

These  are  some  of  the  functions  of  the  TNC  in  a  packet  radio  system.  Their  descriptions  above  should  give  you  a  
better  idea  of  what  packet  radio  is  and  how  it  works.  

195

You might also like