Professional Documents
Culture Documents
Kuo-NetworkAnalysisSynthesis Text PDF
Kuo-NetworkAnalysisSynthesis Text PDF
AND SYNTHESIS
FRANKLIN
F.
KUO
SECOND EDITION
1.50
i-
->
PETER CLARKE
SECOND HAND BOOKS
Tl.
Rochdale 50514
Network
Anal/sis and
Synthesis
Second Edition
Network
Analysis and
Synthesis
Second Edition
by Franklin
Bell
F.
Kuo
Telephone Laboratories,
John Wiley
&
Sons,
Inc.,
New York
London
Inc.
Sydney
Tokyo, Japan
which
it
is
Number
66-16127
To My
Preface
on computer
computers have
brought about many significant changes in the content of
engineering
subject matter concerned with both analysis and
design. In analysis,
computation has become an important adjunct to theory. Theory
estate
lishes the foundation of the subject matter; computation
provides clarity,
depth, and insight. In design, the computer has not
only contributed
precision and speed to existing procedures but has
made practicable
design methods that employ iteration and simulation. The
importance of
computer-aided design cannot be overemphasized. In Chapter 15
I have
attempted to survey some digital computer applications in the
areas of
network analysis and design. I strongly encourage all students
applications (Chapter 15).
In the past
to read
It
By putting
general-
vii
Preface
viii
Chapters 1 and 2
brief description of the subject matter follows.
characteristics of
general
certain
deal with signal representation and
and includes the
analysis,
Fourier
linear systems. Chapter 3 deals with
and 5
impulse method for evaluating Fourier coefficients. Chapters 4
domain.
time
the
in
equations
differential
discuss solutions of network
two preceding
In Chapters 6 and 7, the goals are the same as those of the
domain.
frequency
the
of
that
here
is
viewpoint
chapters, except that the
system function.
Chapter 8 deals with the amplitude, phase, and delay of a
Chapter
synthesis.
network
with
concerned
are
chapters
The final seven
realizability theory
of
elements
the
10,
Chapter
In
two-ports.
9 deals with
elementary drivingare presented. Chapters 11 and 12 are concerned with
Chapter 13, some
In
procedures.
synthesis
function
transfer
and
point
Chapter 14
fundamental concepts in modern filter design are introduced.
design.
and
analysis
deals with the use of scattering matrices in network
digital
of
survey
brief
And, as mentioned earlier, Chapter 15 contains a
five
are
there
addition,
computer techniques in system analysis. In
functions,
filter
approximation.
intended for a two-semester course in network theory.
or
Chapters 1 through 8 may be used in a one-semester undergraduate
analysis.
system
linear
or
analysis
transient
beginning graduate course in
The book
is
network
Chapters 9 and 15 are to be used in a subsequent course on
synthesis.
The second edition is largely a result of the feedback from the professors
who have used this book and from their students, Who have discovered
express my sincere
errors and weaknesses in the original edition. I wish to
appreciation to those
A.
I
J.
and Robert Tracey of Illinois for editorial advice and to Donald Ford
of Wiley for help and encouragement.
Zicchino, and
addition, I wish to thank Elizabeth Jenkins, Lynn
In
tj
u.
Heights,
Berkeley
New Jersey,
December, 1965
efficient
Kuo
transition
from network
be accomplished with
relative ease.
He
Chapters
and 2
deal with signal representation and certain characteristics of linear net3, 4, 5, and 6 discuss transient analysis from both a time
domain viewpoint, i.e., in terms of differential equations and the impulse
response, and a frequency domain viewpoint using Fourier and Laplace
transforms. Chapter 7 is concerned with the use of poles and zeros in
both transient and steady-state analysis. Chapter 8 contains a classical
works. Chapters
in network analysis
and
synthesis.
and proofs
covering the rudiments of matrix algebra, complex variables,
of Brune's realizability theorems.
The book is intended for a two-semester course in network theory.
or
Chapters 1 through 7 can be used in a one-semester undergraduate
analysis.
system
beginning graduate course in transient analysis or linear
Chapters 8 through 13 are to be used in a subsequent course on network
synthesis.
upon
clarity
intuitive understanding.
unit
of a generalized function is first introduced. It is stressed that the
whose
functions
of
sequence
a
actually
but
function
a
not
really
is
impulse
notion of an
limit point is undefined. Then, the less rigorous, intuitive
the
impulse "function" is presented. The treatment then proceeds along
nonrigorous path.
There are a number of topics which have been omitted. One of these
at present. I have puris network topology, which seems to be in vogue
that
posely omitted topology because it seems out of place in a book
analysis.
network
de-emphasizes the form and structure approach to
In an expository book of this nature, it is almost impossible to reference
field of
adequately all the original contributors in the vast and fertile
La
xi
enjoyable undertaking.
Murray
Hill,
January, 1962
New Jersey,
F. F.
KUO
Contents
Chapter
I:
1.1
Signal Analysis
1.2
Complex Frequency
Network Analysis
Network Synthesis
1.3
1.4
Chapter 2:
Signals
14
and Waveforms
20
20
24
28
33
46
3.1
Introduction
3.2
Orthogonal Functions
Approximation Using Orthogonal Functions
46
47
48
50
52
2.1
2.2
2.3
2.4
Chapter 3:
3.3
3.4
3.5
3.6
Fourier Series
Evaluation of Fourier Coefficients
Evaluation of Fourier Coefficients Using Unit
Impulses
3.7
The Fourier
3.8
Integral
xiii
58
63
67
xiv
Contents
Chapter 4:
Differential Equations
75
4.1
Introduction
75
4.2
76
4.4
82
85
4.5
Integrodifferential Equations
91
4.6
93
4.3
Chapter 5:
Network
Analysis:
100
100
103
106
5.1
Introduction
5.2
5.5
Network Elements
Initial and Final Conditions
Step and Impulse Response
Solution of Network Equations
5.6
Analysis of Transformers
114
122
134
6.2
6.3
134
135
137
6.4
5.3
5.4
Chapter 6:
6.1
6.5
6.6
6.7
6.8
Chapter
7:
7.1
7.2
7.3
7.4
7.5
7.6
Transform Methods
in
Network Analysis
HI
144
148
155
162
165
175
175
180
187
194
197
201
Contents
Chapter 8:
212
8.1
212
8.2
Bode
221
Plots
8.3
Single-Tuned Circuits
8.4
Double-Tuned Circuits
On Poles and Zeros and Time Delay
229
238
245
Network
253
8.5
Chapter 9:
Analysis:
II
9.1
Network Functions
9.2
9.4
Interconnection of Two-Ports
9.5
Incidental Dissipation
9.6
253
264
266
271
276
279
10:
290
9.3
Chapter
xv
10.1
10.2
Hurwitz Polynomials
Positive Real Functions
Elementary Synthesis Procedures
290
294
299
308
315
10.3
10.4
Chapter
11:
11.1
11.2
11.3
11.4
11.5
11.6
Synthesis of Certain
R-L-C Functions
315
319
325
329
331
333
xvi
Contents
341
12.1
341
12.2
Zeros of Transmission
Synthesis of Yn and ZS1 with a 1-Q Termination
Synthesis of Constant-Resistance Networks
345
347
352
365
365
365
368
373
388
392
Chapter
12:
12.3
12.4
Chapter
13:
13.1
13.2
13.3
13.4
13.5
13.6
13.7
Approximation
Synthesis of Low-Pass Filters
13.9 Magnitude and Frequency Normalization
13.10 Frequency Transformations
13.8
Chapter
14:
395
397
402
404
413
14.1
14.2
14.3
The
The
14.4
14.5
Insertion Loss
413
415
419
426
429
14.6
431
Chapter
15:
Power Flow
Network
a Two-Port Network
Computer Techniques
in Circuit Analysis
in Circuit Analysis
15.1
15.2
15.3
15.4
438
438
450
453
457
Contents
Appendix A:
xvii
461
Fundamental Operations
Elementary Concepts
Operations on Matrices
Solutions of Linear Equations
461
A.2
A.3
A.4
A. 5
References
A.1
Appendix B:
462
464
468
469
on Matrix Algebra
470
B.l
Generalized Functions
B.2
470
476
481
C.l
C.2
C.3
C.4
Analysis
481
483
486
487
Appendix C:
Appendix D:
Appendix E:
Proofs of
Functions
An Aid
Some Theorems on
Positive
Real
490
mation
493
493
494
495
496
498
502
504
E.l
Introduction
E.2
E.3
E.4
Contour Drawings
E.5
Correction Procedure
E.6
E.7
Conclusion
Bibliography
05
Name Index
$09
Subject Index
si I
chapter
and
three key words: the excitation, the network, and the response as depicted
in Fig. 1.1. Network analysis is concerned with determining the response,
given the excitation and the network. In network synthesis, the problem
to design the network given the excitation and the desired response.
In this chapter we will outline some of the problems to be encountered
in this book without going into the actual details of the problems.
is
We
I.I
some
basic definitions.
SIGNAL ANALYSIS
For
electric
voltages
Response
Excitation
Network
>-
FIG.
I.I.
The
>
Network
m
Ao-
FIG.
Sinusoidal signal.
1.2.
is effected by the
and the Laplace transform. We shall
have ample opportunity to define and study these terms later in the book.
At the moment, let us examine how a signal can be described in terms of
both frequency and time. Consider the sinusoidal signal
other.
s(t)
where
is
the amplitude,
=A
O is
sin (a>
the phase
0.1)
O)
shift,
and
<w
is
the angular
where
<
L2>
in Fig. 1.2.
aD Ao
io.
E
b>0
Angular frequency
FIG.
1.3a. Plot
of amplitude
o>o
to
Angular frequency
<o.
<o.
6>4 6)3
FIG.
a>2 6>i
&>0
1.4a. Discrete
fc>4 W3 ft)2
FIG.
let
signal
is
is
is
+<<>
ft>4
0)2
0>3
ft>4
+0)
phase spectrum.
described in terms of ^4
amplitude
ft>i
1.4b. Discrete
co
"3
amplitude spectrum.
b)i
ft)
">1
>
w o>
anl
^0.
as
and in
shown in
Fig. 1.36,
where
where phase shift
Fig. 1.3a,
plotted.
=%A
<
sin
(<v
sinusoidal components
(1.3)
fy)
xo
where A{m)
is
known
0(to)]
dm
and
(1.4)
0(co)
as the phase
spectrum.
As we
Network
anal/sis
and synthesis
Mo>)
<a
FIG.
1.5a.
FIG.
1.5b.
1.2
COMPLEX FREQUENCY
In this section,
we
we
complex frequency. As
= a + ja>
(1.5)
S(f)
FIG.
1.6.
aJ*
Ae
Triangular signal.
(1.6)
ReS
FIG.
1.7.
Rotating phasor.
ReS(f)
ImS(0
(1.7)
]<b)
(b)
Rotating
(see
Appendix C).
Network
analysis
and synthesis
/Envelope = Ae~
FIG.
1.9.
Damped
sinusoids.
/ ^~ Envelope = Ae**
FIG.
1. 10.
FIG.
I.I I.
Fig. 1.9.
Exponential signals.
real
From
this discussion, it is
S(0
= Ae" = Ae {a+Mt
(1.8)
0,
= Ae*'
(1.9)
Finally, if a = ja> = 0, then the signal is a constant
S(0
as
shown in Fig.
A. Thus
1.3
we
1.11.
complex frequency
description.
NETWORK ANALYSIS
time and frequency, and determining how the network behaves as a signal
processer. Let us turn our attention now to a brief study of the properties
of linear networks and the general characteristics of signal processing by
a linear system.
Network
analysis
and synthesis
BASIC DEFINITIONS
Linear
By
[eg(0, r 2(f)l
e[t)
if,
and
= ex(t) + e2(t),
would be
the response
r(t)
= r,(f) + ri).
By
the
Cx
tionality
is
linearity are
Passive
Reciprocal
Cieift)
C2 e2(t)
..
System
System
CuiM
C2 r^t)
Cm(t) *
Ci*t(0+Ck*gt)
System
FIG.
1.12.
CW<)
linear system.
1
G. Raisbeck, "A Definition of Passive Linear Networks in Terms of Time and
Energy," /. Appl. Phys., 25 (Dec. 1954), 1510-1514.
>-
System
.M.
3<
Ht-Tit
-Ti)
7\
TihTi
Tl
FIG.
excitation
System
1.13.
Time-invariant system.
for any
Causal
=
KO =
(0
then
Time
<T
'
<r
(1.10)
if
before an excitation
< <
t
is
applied at
T.
invariant
T)-*-r(t T),
where the symbol - means "gives rise to." To understand the concept
of time invariance in a linear system, let us suppose that initially the
excitation is introduced at f = 0, which gives rise to a response r(t). If
the excitation were introduced at / = T, and if the shape of the response
waveform were the same as in the first case, but delayed by a time T
(Fig. 1.13), then we could say the system is time invariant. Another way
of looking at this concept is through the fact that time-invariant systems
contain only elements that do not vary with time. It should be mentioned
here that linear systems need not be time invariant.
Derivative property
if e{f)
at the input
gives rise to r(i) at the output (Fig. 1.14), then, if the input were
e'(f),
Network
10
analysis
and synthesis
System
de(t)
System
>
*
dr(t)
System
d*Ht)
'
dt'
it
d*e(t)
dt*
FIG.
1.14.
Some
k
>
System
e(T)dr
dt
C'.{,\
l'(T)
J
1
r'(t).
The proof is quite
where e is a real quantity. By the
time-invariant property, the response would be r(t + e). Now suppose the
i.e.,
e)
excitation were
e1 (r)--W*
)-)]
(1-11)
= - [r(I + e) -
ri (0
e ->- 0,
we
K0]
(112)
see that
limc1(0
e-0
= 7-c(0
at
(1.13)
lim rx(0
-o
= ^ K0
at
We
can extend this idea to higher derivatives as well as for the integrals of
e(t)
and
Ideal
r(f),
as
shown
in Fig. 1.14.
models
Km.
e(t)
Amplifier
FIG.
1.15. Amplifier.
fit)
K dtd
K
FIG.
>K i
FIG.
Amplifier:
1.
An
1.18.
Time-delay network.
1.19.
2. Differentiator:
amplifier scales
The input
Excitation function.
up
up or down
ff VaT
/Tt-JX
uyif
lit)
dt
FIG.
Kdfltt
.16. Differentiator.
FIG.
i.e.,
1.15).
signal is differentiated
(Fig. 1.16).
3. Integrator:
The output
is
shown
in
Fig. 1.17.
4. Time delayer: The output is delayed by an amount
same wave shape as the input (Fig. 1.18).
T,
Then
12
synthesis
m
1
-1
<b)
tit)
_L
Ti
Ti
Ti
(d)
(c)
(ft)
(rf)
Delayed output.
Ideal elements
network.
The energy sources that
are ideal
polarities
Two-port
network
FIG.
1.21.
Two-port network.
*a Response
'/ measurement
13
w Q)
p
FIG. 1.22a. Voltage source.
is
last definition,
we then
an
define
R, L, and
t>(0
C elements,
shown
= L -J9
or
i(t)
= -v(t)
or
iQ)
=i
(0
=7
I'
CJo
v(x)
L *o
J
at
(*)
dx
+ v(0)
or
i(t)
=C
by the equations
dx
j(0)
(1.14)
^
dt
i'(0)
and
be
m*v(t)
m
v(t)
M
FIG.
Ut)
*
v(t)
(b)
(b) Inductor,
^=C
(e)
(c) Capacitor.
Network
14
Ks)
1(3)
>
V(s)
]sL
V(s)
defining the R, L,
initial
(c)
(b)
(a)
FIG.
sC
V(s)
and
(b) Inductor,
shown
elements,
(c)
Capacitor.
moment)
V(s)
V(s)
= sLI(s)
RI(s)
or
I(s)
= ^V(s)
or
I(s)
V(s)
(1.15)
sL
7(s)
= -^/(s)
or
/(s)
sCV(s)
sC
We see that in the time domain, i.e., where the independent variable is
t,
and networks
When
= H(s) E(s).
(1.16)
In network analysis, we are given E(s), and we can obtain H(s) directly
from the network. Our task is to determine R(s).
1.4
NETWORK SYNTHESIS
In network synthesis,
excitation E(s),
we
and the
from the
m <b
FIG.
Z{s)
VM
1.25. Driving-point
15
impedance
R.
system function
aw-
(1.17)
(s)
Since R(s) and (5) are voltages or currents, then H(s) is denoted generally
as an immittance if R(s) is a voltage and E(s) is a current, or vice versa.
9
driving-point immittance is defined to be a function for which the
is
same
Thus a
port.
driving-point impedance
the function
TO-
(1.18)
I(s)
shown
is
in Fig. 1.25.
a current
I(s)
When we
where
Z(s)
= ^i = R
(1.19)
I(s)
Now suppose the resistor in Fig. 1.25 were enclosed in a "black box."
We have no access to this black box, except at the terminals 1-1' in Fig.
1.26. Our task is to determine the network in the black box. Suppose we
are given the information that, for a given excitation /(*), the voltage
response V(s) is proportional to I(s) by the equation
V(s)
= KI(s)
is
(1.20)
resistor
ns)
IRE
_M.
+ 4j
is
a voltage
(1.21)
Network
16
analysis
and synthesis
Us)
hM
'
lo-
>i
=r 4
Two-port
network
Vi(s)
V&)
l'o
FIG.
1.27.
Network
FIG.
realiza-
1.28.
Two-port network.
Our task is to
synthesize
+ 4s).
Before
we proceed with
the
Ztt(s)
(1.22)
)=
tf(S
Fi()
4
Loe.
eit.
17
\H(ju)\
FIG.
As
1.2?. Ideal
filter.
transfer function
must
satisfy in
order to be realizable.
We
shall study
synthesis
is filter
design.
a certain portion of a frequency spectrum and blocks the remainder of the spectrum. By the term
"blocking," we imply that the magnitude response \H(ja>)\ of the filter is
approximately zero for that frequency range. Thus, an ideal low-pass
a network which passes all frequencies up to a cutoff frequency
and blocks all frequencies above w c as shown in Fig. 1.29.
One aspect of filter design is to synthesize the network from the transfer
function H(s). The other aspect deals with the problem of obtaining a
filter is
mc
realizable transmittance H(s) given the specification of, for example, the
1.29.
Why
the
word "approxiand C
characteristics of the R, L,
made to
we can realize low-pass filters
<o c
characteristics
of Fig.
1.30.
In connection with
\H(J<*)\
FIG.
1.30. Realizable
Network
18
analysis
and synthesis
we
problems in magnitude
a filter, we deal with
element values such as R 0.5 ohm and C = 2 farads instead of "practical"
element values of, for example, R 500,000 ohms and C = 2 picofarads
(pico = 10-12). Also we will study a method whereby low-pass filter
designs might be transformed into high-pass, band-pass, and bandelimination filters. The mathematical basis of this method is called
the
filter
design problems,
that, in designing
frequency transformation.
We next discuss some aspects of analysis and synthesis in which the
excitation and response functions are given in terms of power rather than
and
reflected
Problems
1.1
1.2
JtTir/4
3 sin (t
+ ^1 + 4sin
(it
- 1| +
6 sin it
Find the response to the excitation sin / into a sampler that closes every
= 0, 1, 2,
seconds where
Draw the response for ^ / ^ 2*.
13 Find the response to the excitation shown in the figure when the network
is (a)
an
ideal differentiator;
(ft)
an ideal integrator.
Mt)
_L
PROS.
1.4
1.5
1
1.3
a network
is
given as
H{s)~
(s
+ 2X* +
3)
19
network
realiz-
is
s
1.5
ations.
their simplest
Z(s)
()
3*
(*)
Y(s)
= 2s
Z(s)
id)
r(s)
-3i+
s*+2
2,
3s
s*
+4
1.6 For the network shown, write the mesh equation in terms of (a) differential
equations and (b) the complex-frequency variable s.
R
l
wv
npfir*
-0
PROB.
1.7
1.6
For the network shown, write the node equation in terms of (a) differential
PROB.
1.8
r(t)
1.7
(/)
were
chapter
Signals and
in this chapter
is
waveforms
signals
functions of time. In previous studies we have dealt with d-c
of
sinusoids
were
which
signals
a-c
or
time,
with
that were constant
practice,
engineering
In
sin
(cot
A
0).
s(t)
as
such
constant amplitude,
than
the class of signals encountered is substantially broader in scope
of
simple a-c or d-c signals. To attempt to characterize each member
signals
of
variety
infinite
almost
the
of
view
in
foolhardy
is
the class
be
encountered. Instead, we will deal only with those signals that can
building
characterized in simple mathematical terms and which serve as
will concentrate on formublocks for a large number of other signals.
than deal with
lating analytical tools to aid us in describing signals, rather
We
signals.
Let us
first
2.1
is
periodic, then
= s(tkT)
k
20
0,1,2,...
(2.1)
Signals
and waveforms
21
-2T -r
T 2T 3T AT 5T
-l
FIG.
where
2.1.
Square wave.
T is the period of the signal. The sine wave, sin is periodic with
T = lit. Another example of a periodic signal is the square wane
in Fig. 2.1. On the other hand, the signals given in Fig. 2.2 are
/,
period
given
do not repeat
an
after
a certain
finite
infinite period.
is
s(i)
sin
t is
(2.2)
-s{-t)
(2.3)
t is
even.
is
odd
The square
(Fig. 2.26).
Observe that a signal need not be even or odd. Two examples of signals
of this type are shown in Figs. 2.3a and 2.4a. It is significant to note,
however, that any signal s(t) can be resolved into an even component 5,(0
and an odd component s9(t) such that
*(0
= *.(0 + *o(0
(2.4)
For example, the signals in Figs. 2.3a and 2.4a can be decomposed into
odd and even components, as indicated in Figs. 2.3b, 2.3c, 2.4b, and 2.4c.
m
o
t
2
M
FIG. 12.
(b)
(a)
Even
function, (b)
Odd function.
22
Network
analysis
and synthesis
s(t)
1
i
-1
-t
(a)
(a)
sjt)
sM
-1
i
1
(b)
(b)
soft)
oW
-1
-i
(0
(O
FIG.
2.3.
Decomposition into
Original
Even
From
part,
Eq. 2.4
(c)
function,
(6)
Odd part.
we observe
that
s(-t)
= st(-t) + * (-0
- sXO - *o(0
(2.5)
Consequently, the odd and even parts of the signal can be expressed as
(2 .6;
Consider the signal s(t), shown in Fig. 2.5a. The function s(-t) is equal
axis and is given in Fig. 2.5ft. We then
to s(t) reflected about the t
shown
in Figs. 2.5c
and
d, respectively.
Signals
and waveforms
13
rf-tf
-1l
(a)
*Jt)
lift)
-]I
-i
(d)
(e)
s(t)
and s(t).
The
is
f(T+) -f(T-
/(T+)
where
=A
(2.7)
= lim/(T+c)
-0
f(T-)
and
is
(2.8)
- lim/(T- )
discontinuities in the
{
T
FIG. 24. Signal with discontinuity.
Network
24
analysis
and synthesis
Tl
FIG.
/(0+) and/(0-)
2.7. Signal
are
/(0+)
- lim/(e)
-0
/(0-)
(2.9)
= lim/(-e)
For example, the square pulse in Fig. 2.7 has two discontinuities, at
Tt The height of the discontinuity at Tx is
Tx and
s(Tx +)
- s(T -) = K
2.1
(2.10)
Ta is K.
Time constant
important to know how quickly a
of
the decay of an exponential is the
useful measure
waveform decays.
waveform described by
exponential
the
time constant T. Consider
In
many
physical problems,
it is
r(t)
From a plot
Also
of r(t) in Fig.
2.8,
= Ke~*IT u(t)
we
see that
when
(2.11)
T,
r(T)
= 0.37KO)
(2.12)
r(4T)
= 0.02r(0)
(2.13)
Signals and
waveforms
25
1.00
075
r(t)
0.50
037
025
O02
5
FIG.
2.8.
1.
Observe that the larger the time constant, the longer it requires for the
waveform to reach 37% of its peak value. In circuit analysis, common
time constants are the factors RC and RjL.
RMS Value
The rms or root mean square value of a periodic waveform e(t) is defined
as
Crms
i?f>*r
(2.14)
where T is the period. If the waveform is not periodic, the term rms does
not apply. As an example, let us calculate the rms voltage for the periodic
waveform in
Fig. 2.9.
Inns
4^r(T-)M>]r
(2.15)
-fair
= J2J3Av
D-C Value
The
d-c value of a
is
oc
= -1
fT
TJ
e{t)dt
p
(2.16)
26
Network
analysis
and synthesis
2T
-A
FIG.
2.9. Periodic
waveform.
waveform in
[AT
ATI
(2.17)
Duty cycle
The term duty cycle, D, is defined as the ratio of the time duration of the
positive cycle t9 of a periodic waveform to the period, T, that is,
(2.18)
T
becomes important in dealing with waveforms of the type shown in Fig. 2.10, where most of the energy is conThe rms voltage of the waveform
centrated in a narrow pulse of width f
The duty
in Fig. 2.10
is
,h
*--(iM'
(2.19)
= AJ7JT
An
(o
cycle.
Signals and
waveforms
27
TT
FIG.
2.1 1. Periodic
cycle.
Wc see that the smaller the duty cycle, the smaller the rms voltage. The
square wave in Fig. 2.1 has a 50% duty cycle.
Crest factor
Crest factor1
CF = -^-
-^-
or
rms
whichever
voltage
is
is greater.
(2.20)
rms
defined as
e, p
(2.21)
<?
eJ*
= et(T-t )
(2.22)
(2.23)
Also,
eb
= eD
and
ea
= ePP(l -
D)
(2.24)
is
= ( e( l ~ D? +
f
enWr-
t )\
*
(2.25)
G.
Justice,
'The
eppy/D(l
D)
No. 5
Network
28
CF =
ejerm*,
we have
CF =
e(\
D)
ep jD(l -
D)
(2.26)
-Vi/J>-i
For example,
if
D=
CF
(2.27)
= ^100 -
UD =
10
10,000'
CF=VlO,000- 1~100
(2.28)
A voltmeter with high crest factor is able to read accurately rms values of
whose waveforms differ from sinusoids, in particular, signals with
low duty factor. Note that the smallest value of crest factor occurs for
the maximum value of D, that is, X>mx = 0.5,
signals
CFnto
= Vl/iW =1
(2.29)
Z3
The
shown
(/)
=
=
in Fig. 2.12
is
defined as
<
f0
(2.30)
The
which closes at
u(t)
^o
Switch
lv
Network
T.
FIG. 2.12. Unit step function.
step.
Signals
and waveforms
29
m
*<t-a)
a
FIG. 2.14. Shifted step function.
argument
argument
(/)
> 0, is defined by
vit
and
is
occurs
(t) is
- a) =
=
is
negative,
Thus
/
1
<a
{Zil)
t^a
shown in Fig. 2.14. Note that the jump discontinuity of the step
when the argument within the parentheses is zero. This forms the
basis of the shifting property of the step function. Also, the height of the
jump discontinuity of the step can be scaled up or down by the multiplication of a constant K.
With the use of the change of amplitude and the shifting properties of
the step function, we can proceed to construct a family of pulse waveforms.
For example, the square pulse in Fig. 2.15 can be constructed by the sum
of two step functions
sif)
= 4u(t -
1)
The "staircase"
by the equation
<
s(t)
=>J,u(t
(-4)
u(t
- 2)
function,
shown
(2.32)
in Fig. 2.17,
- kT)
*t)
-4
FIG.
2.16. Construction
is
(2.33)
Network
30
analysis
and synthesis
3
2
1
1
ZT
FIG.
shifting
property,
s(t)
we
see
m(0
simpler
way
2u(t
T)
2u(t
2T)
wave
2k(*
is
- 3D +
'
(2.34)
is
s(0
is
2.18.
(2.35)
-(*?)
S (0
is
negative, as seen
= (sin^)-u(-sin^)
(2.36)
Another method of describing the square wave is to consider a generalization of the step function known as the sgn function (pronounced signum).
is
defined as
sgn [/(f)]
=1
=
>o
fit) < o
fit)
fit)
s(t)
T
FIG.
2. IS.
2T
The
3T
AT
signal u(sin7rr/r).
(2.37)
Signals
and waveforms
31
ft)
ZT
3T
Fig. 2.1
s(0
is
simply expressed as
sgn
sin
(2.38)
we
s(r)
3T)]
(2.39)
The
tu(t)
(2.40)
FIG.2JM.
Network
32
analysis
and synthesis
s(t)
FIG. 2.21.
First, let
When
of 5(0 versus
we
in Fig. 2.20.
t',
If,
t,
= a.
t'
f u(t')
a.
Then
(2.41)
then have
p(f)
shift
by a new variable
p(r')
time
(t
- a) u(t -
(2.42)
a)
Fig. 2.21.
From the preceding discussion, it is clear that if any signal /(/) u(t)
delayed by a time T, the delayed or shifted signal is given by
f{t')=f{t-T)u{t-T)
For example,
let
is
(2.43)
s(t'),
s(t')
shown
[sin
in Fig. 2.22,
- (t -
is
T)] u(t
T)
(2.44)
Signals
and waveforms
33
m2-
As a final example, consider the waveform in Fig. 2.23, whose component parts are given in Fig. 2.24. For increasing t, the first nonzero
component is
1) u(t
1),
straight
line
of slope 2 at
2u(t
term
s(t)
1.
= 2{t -
1) u(t
1)
down
2(t
to zero. Thus,
2) u(t
2)
2u(t
2)
(2.45)
Dirac
2(t-\)u(t-\)
P.
Press, 1930.
Network
34
anal/sis
and synthesis
by the equations
=
<J(0 =
6\t) dt
f:
most important property
Its
is
(2.46)
for
(2.47)
f(t)d(t)dt=f(0)
f
J
(2.48)
approach.
At
first
..
o(0
is
if
/0 ao>l
(2.49)
,.
(0
doubtful. After
all,
it does
not even exist at that point! However, consider the function g t (i) in Fig.
2.25. It is clear that as e goes to zero, g ( (t) approaches a unit step, that is,
unit step
is
lim ge(t)
(0
(2.50)
-o
is
defined by the
equations
g'(0
as
shown in
>
e 4+1 .
-e
= 0;
t<0, t>
Now let take on a sequence of values e
such that
(
Consider the sequence of functions {g' c< (0} for decreasing
Fig. 2.26.
*V0
et (t)
FIG.
(2.51)
US.
0.
FIG.
1M.
Derivative of
g((t)
in
Fig. 2.25.
1955, 175-190.
Signals and
waveforms
35
h-\W>
8it
s s
values of c { , as
shown
in Fig. 2.27.
property:
J't>0g' (t)dt
(i
*1<0
where
tx
and
/,
As
(2.52)
approaches zero,
g'(o+)-
00
,-0
(2.53)
so that the limit of the sequence is not defined in the classical sense.
Another sequence of functions which obeys the property given in Eq.
2.52
is
We
particular,
we
define
r*t>o
J*i>o
t)A-lim
i<0
g' t((t)dt
e<-0/i<0
a
4hm
r* >o
UQdt
(2.54)
Network
36
analysis
and synthesis
It
stated previously,
From
is
is
we can
heuristic one.
<5(f)
=
=
oo
(2.55)
for
t^O
waveforms
Signals and
Continuing with
the impulse
is
d(t)dt
Thus
we
6
6{i)
dt
/*o-
S(t) dt
we have
(2.56)
is
"concentrated" at
r+oo
(5(f)
dt
=
is
0.
Con-
zero,
s(t)
J0+
J-oo
# 0,
Jo
<x>
as seen by
37
s(t)
= Au(t- a)
(2.57)
A,
The
s(t)
yields
= Au(t a)
(2.58)
an impulse function
B'(t)
s\t)
= A6(t- a)
(2.59)
A written
Note that A is the
A d(t - a).
multiplier
s'(t)
= Ad(t-a)
Consider the implications of Eqs. 2.58 and 2.59. From these equations
see that the derivative of the step at the jump discontinuity of height A
we
yields
At
T
FIG. 2.30. Function with discontinuity at
T.
38
Network
which
is
analysis
and synthesis
given as
A =/(r+) -f(T-)
Let us define
same shape
fx(t)
(2.60)
<
T,
>
T, that
is,
Ut)=f(t)-Au{t-T)
The
derivative
A0
is
then
A0 -AC) + ^
The following example
the function /(f)
(2.61)
illustrates this
<K*
point
T)
(2.62)
more
clearly.
In Fig. 2.31a,
is
fit)
= Au(t-a)-A u(t - b)
(2.63)
Its derivative is
fit)
= Ad(t - a) - Ad(t - b)
(2.64)
f(b+) -fQ>-)
- -A
(2.65)
m
A
8(0
4
(a)
r<*>
(a)
g'(t)
A'
2
2
~0]
'-A
(b)
(b) Derivative
FIG .2.32.
(a) Signal,
(*)] derivative.
Signals and
waveforms
39
As a second example,
We
0+)-(l-) =
2,
(2.66)
f
This integral
1
5* T.
is easily
evaluated if
that
T)
for all
f{t)d(t-T)
If/(r)
we
(2.67)
is
we
single-valued at
aXLtitT
(2.68)
integral so
obtain
/(T)
d(t
J O0
-T)dt= f(T)
is
(2.69)
continuous at i
T.
+00
Jf(t)Ht-T)dt
Example
is
2.1
/(0
- 1**
T)dt
- e* r
+0O
Example
Z2
e
~&*Hf
(2.70)
-00
/(/)
= sin t
>''('- 5)*-^
(2.71)
Network
40
analysis
and synthesis
f(t)
f(t)d(t-T)dt=f(T)
oo
< <
t
Let
oo.
(2.72)
-oo to +oo,
which holds for all / in this case. If T were varied from
of this sort
operation
An
entirety.
its
in
reproduced
then/(0 would be
paper with
of
sheet
a
moving
by
function/(f)
the
scanning
corresponds to
across a plot of the function, as shown in Fig. 2.34.
higher order derivatives of the unit step function.
Let us
we represent the unit impulse by the function ft(i) in Fig. 2.28,
a thin
slit
now examine
Here
which, as
Thus,
6'(t)
r
The
dt
I
1
/() d'(t
(2.73)
is
-T)dt= -f\T)
rt (t)
f2
(2
<=
-* 0.
(2.74)
Signals and
waveforms
41
*/
f(t)d'(t-T)dt=f(t)d(t-T)\
=
It
(t)d(t-T)dt
-00
/ oo
(2.75)
-f'(T).
f
r
oo
/(O
<B,
<5
(<
(2.76)
where (5< n and/ <"> denote nth derivatives. The higher order derivatives of
can be evaluated in similar fashion.
>
d(t)
Problems
2.1
Resolve the waveforms in the figure into odd and even components.
s(t)
s(t)
-v
+1
(a)
a>)
s(t)
PRO B. 2.1
Network
42
analysis
and synthesis
waveforms
fit)
2
1
A
2
(b)
f(t)
fit)
1.5
1
(d)
(c)
PROB.
2.2
23
for a value of
>
f(r)dr
T.
f(t)
f(t)
2*Wl
EMt)
>*
-E6(t-T)
PROB. 2.4
PROB. 2.5
Signals and
2^ For
shown
waveforms
43
what value
P/(0<//-0
()
J 00
j"7<f)<fr-o
<*)
J0+
2.6
is,
/",(/
- Q, d(t -
/,),
CU.1.1.
Sketch *u_
r
*
rL\
j * x
..
the ..
waveforms
for (6) and (c) neatly.
JW
m
20
"*
At
(0
(ii)
PROS.
2.7
2.4
Prove that
(a)
<*'(*)
(6)
--<}'(--X)
-<(a;)
= x d'(x)
/oo
(c)
;*)
&
-o
J 00
2.8
/W-^(/-e)*,
= 0,
Show that
as e
;!
o<;/
elsewhere
J" 5
Network
44
analysis
and synthesis
Plot
2.9
(a)
%s
(ft)
2.10
sgn (cos 0;
t 2n
'
(fl )
<*t
Ti <T^
J CO
/go
a,)
<5(a>
a> )
cos
o>f rfo>
J 00
OO
J
2.11
(c)
00
f " sin
2.12
(t
- l\ +
S'
('
"
f)
r(t)
Plot
tU
KO for
<, t <.
sin
n s(t
*'<'
>]
is
K = 0,
-Kj\dt;
1, 2, 3,
IT.
2.13
system is r s (0
due to a
2e
'
u(t),
staircase
j(0-i(/-*r)
*=o
Plot both excitation
and response
functions.
If the
e(t)
Plot
2S(t
1)
2d(t
e(t).
PROB. 2.16
2)
is
W) = ~* "(')
Signals and
2.16
The
(a)
(b)
Make
(2e-
2t
waveforms
45
is
1) (/)
Show
all
pertinent
chapter
3.1
INTRODUCTION
One of
signals.
the most
If
common
= s(tnT)
7i
= 0,l,2,...
"T
I
is finite
(where a
is
an arbitrary
\s(t)\
(3.1)
number of discon-
finite
dt
can be expanded
s(t)
=^+
+
+ a t cos 2<ot +
b x sin cot + b t sin "hot +
a x cos
tot
2\
m=
where
s(t)
= ?* + 2 ( cos not +
2
b n sin
ntot)
(3.3)
apparent from Eqs. 3.2 and 3.3 that, when s(t) is expanded in a
Fourier series, we can describe s(t) completely in terms of the coefficients
It is
46
analysis
47
its
stitute
ORTHOGONAL FUNCTIONS
3.2
Then
x (f)
if
r,
f
JTx
/i(0/*(0dt
=O
(3.4)
we
say that/i(r) and/^/) are orthogonal over the interval [Tlt TJ. For
example, the functions sin t and cos t are orthogonal over the interval
nlit
r <;
<f>&)
(
<j>
H(t)}.
l)2n.
(*.
4>i)
V<(0
4>M dt =
i*j
0,
(3.5)
then the set {<f>t } forms an orthogonal set over the interval [7\, TJ. In
Eq. 3.5 the integral is denoted by the inner product (^ 4 , <f> ). For convent
ience here, we use the inner product notation in our discussions.
The set {^J is orihonormal over [Tu 7",] if
=
The norm of an element
l&l
<f>
k in
the set
W. H.
t-.j
{^J
is
defined as
(3.6)
1
set
{<f>x, <f>t,
. . ,
set,1
is
Theory, CT-1,
(3-7)
IRE on
Circuit
Network
48
analysis
and synthesis
- e~**
*i(0
Ut) -
To show
- 2(fl01
*,(/)
4>t(t)
= e-'[l -
far
(3.8)
+ 6(flf)* - Ka/)8
f"
Letting t
e-*[l
2(0*] dt
(3.9)
= at, we have
i,-;;J'V a-4T+2,*>rfr
,,
(3.10)
,_(J[-^.*)
-
11*1
-^
a
') ]
Uo
(3,.,
*}
J
(3.12)
V2a
It is
not
difficult
each element fa by
3.3
set
orthonormal,
we
divide
1/ V2a.
In this section
we
explore
= lim
n-oo
*W) - /(0]
f
JTi
dt
(3.13)
When
Eq. 3.13
we
is satisfied,
49
analysis
fit).
To examine
mean more
>
we
closely,
for which
T,
\f(t)\*dt<ao,
(
JTi
we
[T
Tt
is
integrable
L*
in [7\,
TJ, and
we
write /(/)
e L
in
].
Definition
12
tions integrable
If/(f)
that if
Bm.P|/(0-/^)r*-0
then
{/.(*)}
The
converges in the
2 we say that
4>i,<l>
/(0
<f>
n-
= I *<&(<)
(3.14)
<-i
Our problem
is
squared error
11/
(3.15)
JTx
is
Proof.
must
set
= ]Kt)Mt)dt
ct for every
H/-/.II*
1, 2,
. .
,
.
\\f
(3.16)
/|| s to be minimum
we
n.
11/11*
-22a,(/,&)+2a,W
(3.17)
Network
50
(/,&).
analysis
{&}
and synthesis
orthonormal, |\| a
is
1,
and by
definition, ct
We thus have
I/-/-I'
i/-/.!"
l/l'
- 22A + i
(318)
'*
2 c* gives
8
ii/n
- i*"
- 2iA
+ i*'
+ i^
_1
<-l
<-l
<_1
(3.19)
set
= c<.
coefficients
The
minimum
attain
we must
coefficients c t ,
Parseval's equality
Consider fn(t) given in Eq. 3.14.
set
{+&)}.
We see that
ft
I L/j(or*-i'
since
320>
equality,
3.4
<
and
is
FOURIER SERIES
aft)
_ s + V (a , cos nmt +
2
b n sin nmt)
(3.21)
Ti
where a
is
any
real
r^MO-s-O)]*^
number.
||j(0
322>
Ja
is finite,
are the Fourier coefficients of s(t) with respect to the orthonormal set
/cos koat
sin k<ot \
.__,
the
the constants a{ , b t
51
analysis
fYVtW\
-3x
-2x
-x
FIG.
2x
3x
wave.
earlier, are
*)A
fl0=s
rJ
2 r
s(t)
-If
mean, when
s(t)
cos kmt dt
(323)
/f
(3.24)
(3.2S)
contains a
= ^IJ(o=)
S .0,)
( 3. 26)
At any point
*(*i)
T=
s{t)
Let us take a
derived,
= A |sin
we have
(3.27)
t\
and
ir.
K = ~ Jo (0 sin 2nt dt =
|
(3.28)
TT
2A C*
&A
=
sin tdt =
\
IT
Jo
2[>
(3.29)
IT
Jo
-An*
AA
(3.30)
Network
52
analysis
of the
series
5(/)
and synthesis
wave
rectified sine
- i(l + f
is
cos 2nt)
(3.31)
3.5
In this section we will consider two other useful forms of Fourier series.
In addition, we will discuss a number of methods to simplify the evaluation
of Fourier coefficients. First, let us examine how the evaluation of
coefficients is simplified by
that
is,
rr/t
T
ss
ro
s(t)
cos mot dt
J
"
J-T/S
T/a
JO
2 rf
-i
T,t
nmt dt
s(t) sin
TUo
s(f) sin
(3.32)
nmt dt
J-T/i
an
bn
(T/2;
= 2
T/ *
[s(x)
TJ
T Jo
T/
2 C
I
Then we have
0).
*
[s(a:)
for all n,
s(x)
s(x), then we
see that
and
T/2
-if*
nmx dx
s{x) sin
tJo
On
if a
function is odd,
its
= s(x), then b n =
an
(3.34)
is
even, that
is,
and
TJo
s(x) cos
nmx dx
(3.35)
anal/sis
53
FIG. 3.2
s(t)
,(,
l)--<0
as given
and
= bn = 0;
-if
TJo
T Jo
s(t)
(3.36)
neven
cos nmt dt
(3.37)
n odd
dt,
C cos (na>i + 8J
^|
(3.38)
Network
54
anal/sis
and synthesis
m
A
^
^
^*^T
-T--
^"'
(a)
m
A
"*-..
-r'- "~^
f~"-~.
-A
(6)
m
A
**-
-T)
t
*
I**"
-A
(e)
FIG. 3.4. (a) Even function cosine terms only, (ft) Odd function
Odd harmonics only with both sine and cosine terms.
(c)
= C cos 0
bn = C sin 0
a
and
0,
(3.40)
of an and bn, as
c,-(fl.'
Ca =
(3.39)
a
(3.41)
(-!:)
combine the cosine and sine terms of each harmonic in the original
we readily obtain from Eqs. 3.38-3.41 the Fourier cosine series
If we
series,
f(t)
= C + Ct cos (mi + 00 +
+ C.cos (3fl>f + 0,) +
C, cos (2o>f
0.)
+ Cn cos(nmt + J +
(3.42)
55
anal/sis
If however,
3cos2a>f =
For example, the Fourier
was shown to be
of the fully
series
= ^(l + f
s(0
2ir\
3 cos (2<u/
n=i
+ tt)
(3.43)
rectified sine
wave in
Fig. 3.1
-^
cos 2nt)
An
(3.44)
series, s(t) is
(3.45)
can be written as
(3.46)
fl
" + ^* *-'"')
n-l\
n-l
If we define
"~ n
'
'
is
= A + i(/ffe*"" + /*--'"")
5(0
"^
- 2
n
34)
PSoo
Pn
= l
~ Jb
C
T Jo
=i
f
T Jo
Equation 3.49
is
is
s(0(cos nott
s(t)e-
j sin na0 dt
(3.49)
mt dt
i'
/?.
(/)
56
Network
analysis
and synthesis
Cb
Ci
Ci
C2
Ch
<h
Ci\
c*.
-Au -3 -2w -w
FIG.
3.5.
2(i>
Amplitude spectrum.
-2T
2T
HF -T 4
-A
FIG.
3d)
3.6.
Square wave.
,C*
4a>
is
usually complex
/*n
The
real part
of fi
Re /?,
- -1
T Jo
+yIm,
(3.50)
(0 cos tuot dt
(3.51)
is
T Jo
It is clear that
CT
JImfl,-^f
function in n.
57
is
Re Pn
Re/8B
anal/sis
s(0sinn>r<fc
(3.52)
ImPM
|/JJ-(Re"fc.+
and the />/uue spectrum
is
(3.53)
denned as
6,
= arctan^
(3.54)
It is easily seen that the amplitude spectrum is an even function and the
phase spectrum is an odd function in n. The amplitude spectrum provides
us with valuable insight as to where to truncate the infinite series and still
maintain a good approximation to the original waveform. From a plot of
the amplitude spectrum, we can almost pick out by inspection the nontrivial terms in the series. For the amplitude spectrum in Fig. 3.5, we see
that a good approximation can be obtained if we disregard any harmonic
above the third.
As an example, let us obtain the complex Fourier coefficients for the
square wave in Fig. 3.6. Let us also find the amplitude and phase spectra
of the square wave. From Fig. 3.6, we note that s(t) is an odd function.
Moreover, since s(t
T/2) = s(t), the series has only odd harmonics.
From Eq. 3.49 we obtain the coefficients of the complex Fourier series as
Ae-'Mi
--
TJo
dt--\
Ae' ,nat dt
TJt/
(3.55)
(I
2e~ linmT/i)
-)-
jnmT
Since ncoT
e~
,tuaT
)
- rr- a - 2*"*" +
"*"*)
( 3 - 56 >
58
synthesis
Amplitude
-3
-5
<a)
Phase
-5
-3
-1
2
(b)
FIG.
Simplifying
/5
one step
we
further,
obtain
n odd
(3.57)
jnir
=
The amplitude and phase
n even
Fig. 3.7.
34 EVALUATION
In this section
is
simplify the calculation of complex Fourier coefficients. This method
restricted to functions which are made up of straight-line components only.
Thus the method applies for the square wave in Fig. 3.6. The method is
f" /(*)(-
J 00
TO* -/(TO
(3.58)
Let us use this equation to evaluate the complex Fourier coefficients for
,fmt
e~
, we have
the impulse train in Fig. 3.8. Using Eq. 3.58 with/(f)
'
~ *~ 2V,,
e
d%
T fH' - 2 V'"" '~T
<
(359>
Al
A..
A,i
_L
1
2
FIG.
3.8.
59
analysis
_L
3r
T
2T
Impulse
sr
train.
We
see that the complex Fourier coefficients for impulse functions are
obtained by simply substituting the time at which the impulses occur into
the expression, e~,m" t .
In the evaluation of Fourier coefficients, we must remember that the
T, as
wave
is
wave to
can
give
s'(t),
now evaluate
as
shown
of the square
in Fig. 3.96.
We
made up of impulses
s(t) is
>(t)
which clearly
is
alone. Analytically, if
-A
given as
s(0
(<)
- 1
inert
Pe
(3.60)
00
W
is
A|
(3.61)
Here,
we
define a
new complex
coefficient
(3.62)
or
jnm
(3.63)
-2A
a function which
consists of impulse components alone, then
If the. derivative s'(i)
we
is
/?
from Eq.
(6)
[0, T\.
Network
60
and synthesis
analysis
s'(t)
[0,
A6(t)-
d(t
2^4
- |) + A d(t -
T)
(3.64)
coefficients are
Tvo
(3.65)
= (\ 2e- linmT/* +
}
e~
inmT
)
The Fourier
coefficients
y
jnco
(3.66)
(i
2e~
linaTI * )
e~
inmT
)
jnmT
which checks with the solution obtained in the standard way in Eq. 3.55.
If the first derivative, /(f), does not contain impulses, then we must
differentiate again to yield
= f
n
s (f)
(3.67)
(3 - 68)
For the
[0,
T]
is
s(0
The
inmt
K =jno>Yn = (P )* P
where
period
Xne
oo
coefficients
= ^Uf) - 2d(f - |) +
A n are
Xn
now
=-
**
- d]
( 3 - 69 >
obtained as
T
( s"(i)e-
inat
dt
TJo
(3.70)
2A
which
,j
_ 2e-OfK27*>
-fnmT\
simplifies to give
X =s
"
&A
T*
.,
n odd
(3.71)
n even
analysis
61
A&1
3T
T
(a)
24 Y0
2"
2A
T
2A
(b)
[>"<'>
2A
(C)
From
A_
we
its
derivatives.
obtain
P-r.
(Jmnf
2A
(3.72)
nodd
nV
neven
terms
we must
differentiation,
differentiate
once more.
However, from
we
this additional
T)
J GO
-s\T)
(3.73)
so that
f " d'(t
J 00
T)e- im"dt
This
= jna>e-"u T
'
(3.74)
Network
62
analysis
and synthesis
S '(0
The second
M (0
|[
u(t
f )]
+ 0 - 2(l - f)
0.75)
doublets as given by
mas
shown
|[aco
in Fig. 3.11c.
*(t
f )]
+ *W -
2d'(t
1)
We therefore evaluate A as
,-inmt
s"(t)e
dt
tJo
(3.77)
.,
'
(3.76)
(W
(o)
-*'('-!)
"W
pit)
?(*-i)
to
FIG. 3.11
t-M(l-f)
The complex
coefficients /S are
now
63
analysis
obtained as
Qtonf
janT
(jconT)*
Simplifying,
(3-78)
'
we have
1
Pn=-
mr +
t
"odd
j2irn
(3.79)
n even
j2irn
In conclusion,
it
/3
In this section
we
CT
s(i)e-
M"'*dt
TJ-i
-T/2
and the inverse
(discrete)
transform was
*(0
From
= 2
n=
ln,'
Knfo)e
(3.81)
oo
(3.80)
is
A/ -
(n
l)/
=i
(3.82)
As the period T becomes larger, the spacing between the harmonic lines in
the spectrum becomes smaller. For aperiodic signals, we let T approach
infinity so that, in the limit, the discrete spectrum becomes continuous.
64
Network
and synthesis
anal/sis
lim
Tx>
W = r
Jo
A/-0
The
inverse transform
-"
s(()c
it
(3.83)
J oo
is
<0-f"
J
S(J)e
iU,t
(3.84)
df
00
Equations 3.83 and 3.84 are sometimes called the Fourier transform pair.
5~x denote
denote the operation of Fourier transformation and
If we let
inverse transformation, then
*</>-*-<0
s(t) = ?-i-S(f)
In general, the Fourier transform S(f)
S(f)
The
real part of
S(f)
is
is
(3 . 85)
= ReS(f)+ j Im S(f)
(3.86)
ReS(/)
J[S(/)
|"
cos 2-nft dt
S(-/)]
(3.87)
s(t)
J CO
and
= [SCO -
Im S(f)
S(-/)]
2/
(3.88)
=_
J CO
is
s (0 sin 2tt/(
dt
defined as
[Re SC/) 2
+ Im (/)*]*
(3.89)
is
"-" Sin
**
Using the amplitude and phase definition of the Fourier transform, the
inverse transform can be expressed as
s(t)
A(f) cos
J 00
Let us examine some examples.
[lirft
<Kf)] df
(3.91)
65
anal/sis
<Kf)
-f
+/
(a)
((,)
Example
/,).
3.2.*
sit)
= A6\t-td
S(f)~ ("
A^t- t9)er^t* dt
J 00
- Ae-i**">
Its
amplitude spectrum
(3.92)
is
A(f)=A
while
its
phase spectrum
(3.93)
is
<Kf)
(3.94)
-2^//o
Example 33.
Formally,
HSyW
\x\
>
-1
3r
[rect/]
(3.95)
-W
recta;/
The
3.13.
(pronounced
sink),
sine t
W/2
e>*"*df
(3.96)
-I-Jf/2
sinirWr
Vt
* It should be noted here that the Fourier transform of a generalized function is also
a generalized function. In other words, if e C, (^ <f>) C. For example, J" <5(f ) =
1, where 1 is described by a generalized function 1(/). We will not go into the formal
details of Fourier transforms of generalized functions here. For an excellent treatment
of the subject see M. J. Lighthill, Fourier Analysis and Generalized Functions, England,
Cambridge University Press, 19SS.
<f>
Network
66
1.0
-*
"
0.8
0.6
sine
0.4
0.2
\\
0.0
-0.2
-0.3
~w
3.
~w
1.
~w
W
/
2.
3.
4.
5.
curve.
r(t)
Excitation
System
Narrow pulse
Wide band
*(0
Excitation
i.
-L
~w
Narrow pulse
r(t)
System
Response
10 Ti
10 Tx
Wide pulse
Narrower band
Wide pulse
FIG. 3.15. Illustration of the reciprocity relationships between time duration and
bandwidth.
67
anal/sis
From the plot of sine t in Fig. 3.14 we see that sine t falls as does |r| _1 ,
with zeros at t
n\W, n
We also note that most of the
1, 2, 3, .
energy of the signal is concentrated between the points
\\W t
\\W.
< <
Let us define the time duration of a signal as that point, t , beyond which
the amplitude is never greater than a specified value, for example, e
We
can effectively regard the time duration of the sine function as t
1/ W.
The value W, as we see from Fig. 3.13, is the spectral bandwidth of the
.
rect function.
We
see that if
W increases,
decreases.
The preceding
narrow
is
quite fundamental.
pulses,
i.e.,
It illustrates
why
in pulse transmission,
mitted through filters with large bandwidths; whereas pulses with longer
time durations do not require such wide bandwidths, as illustrated in
Fig. 3.15.
3.8
In this section
we
some important
consider
properties of Fourier
transforms.
Linearity
The
linearity property
transform of a
sum of two
transforms, that
is,
Hcx *i(0 + c
= Cl S (/) + c
t sjflj]
St(f)
(3.97)
Differentiation
This property states that the Fourier transform of the derivative of a
signal isjlnf times the Fourier transform of the signal itself:
J-s\i)=j2-nfS{f)
(3.98)
?- s (t) = (j2*frS(f)
(3.99)
or more generally,
transform definition,
00
s'(0
= 7;f
S(J)e*<*df
at J -oo
(3.100)
-f
J
CO
j2nfS(f)e*"df
Network
68
Similarly,
it is
and synthesis
analysis
easily
shown
f
5(t)
'
dr\
= ^-
(3.101)
S(J)
j2nf
LJ-oo
is
j(0
Its
Fourier transform
= -'(0
( 3 - 102>
is
s(/)
r^vHtyr^dt
f"CO
J
(3.103)
The
derivative of s(t)
s '(t)
Its
Fourier transform
+ j2irf
is
3(0
- aer** (0
(3.104)
is
?['')]
-1-
j2nf
a+j2wf
a+j2irf
(3.105)
= j2nfS(f)
Symmetry
The symmetry property of Fourier transforms
states that if
= X(f)
^ X(t) = *(-/)
3"
then
(3.106)
x(t)
(3.107)
This property follows directly from the symmetrical nature of the Fourier
transform pair in Eqs. 3.83 and 3.84.
Example
3.4.
^
It is
section,
we know
that
= rect/
(3.108)
(3.109)
sine /
&
rect
property
&
8(t)
1.
From
Consider next
the symmetry
8(f)
(3.110)
as
analysis
69
HD
-/
-t
1.0.
Scale change
The
-M*0
[.(*)]
Proof.
We prove
and bandwidth
It states that
reciprocity relationship.
this
(3.111)
form
J-\\a\ S{af)]
\a\
f" S(af)e* df
(3.112)
J oo
Let/'
= af;
then
*^ll
S(f')]
\a\
f " S(/V*"'
J co
( '/o,
a
(3.113)
As an example,
consider
T[e* "(01
(3.114)
J2vf+a
then
Sy
\a\
(01
j2iraf
a
(3.115)
j2nf+\
ifo>0.
Folding
The
J-W-01
= s(-f)
(3.116)
Network
70
anal/sis
An
example
and synthesis
directly
from the
is
3=V u(-t)]
(3.117)
l-j2vf
Delay
is delayed by an amount t in the time domain, the corresponding effect in the frequency domain is to multiply the transform of the
undelayed signal by er iz"ft that is,
If a signal
<>,
(3.118)
For example,
&[e- aU
-' )
u(t-t
)]
-}2rftt>
=
a
(3.119)
+ j2nf
Modulation
The modulation or frequency
states that if
signal
by e iufot that
,
is,
3^[S(f - /<,)]
mM
(3.120)
s(t)
Example
3.5.
Given S(f) in Fig. 3.17a, let us find the inverse transform of
5"" 1
Si(f) in Fig. 3.176 in terms of s(t) =
S(f). We know that
Sxif)
S(f -f )
+ 5(/ + /)
(3.121)
-/
+/
-/
*f
(b)
FIG.' 3.17.
analysis
71
(3.122)
Thus we see that multiplying a signal by a cosine or sine wave in the time domain
corresponds to shifting its spectrum by an amount
f In transmission terminology / is the carrier frequency, and the process of multiplying s(t) by cos
.
2itfy
is
Parseval's
theorem
An
"
i(0 St(0 dt
I
J
00
= f"
f"
si(0 i(0 dt
J no
Sx(/) S,(-/) df
as follows:
-f"
*
(3.123)
SiC/V""* d/
Si(/)4rf"
s&Vu"dt
(3.124)
Jao
oo
sMSA-fldf
-f"
J
00
f " aty)
J 00
* = J
f"
ISC/)!
4T
(3.125)
00
energy
1-ohm
is
(0 dt
We see from Eq. 3.125 that the total energy is also equal to the area under
the curve of \S(f)\*.
or energy spectrum.
Thus
\S(f)\* is
Problems
3.1 Show that the set {1, sin nnt/T, cos nnt/T}, n = 1,2, 3,
, forms an
orthogonal set over an interval [a, a + 2T], where a is any real number. Find
the norms for the members of the set and normalize the set.
.
Network
72
3.2
analysis
and synthesis
Fourier
Given the functions/^) and/i(0 expressed in terms of complex
senes
/l(/)=2^
n oo
"
B,,
/K0-2A.'*-'
where both/i(0 and/j(0 have the same period T, and
show
= Kl e**\
A.
e*
that
oA>
+ 22 lgl cos(flB -
Note that
For the periodic
4>J
m * -n
\0,
33
lftl
f(f>
--
+T
-T
(h)
fit)
T
5
(e)
PROB.3.3
A
T
analysis
3.4 For the waveforms in Prob. 3.3, find the discrete amplitude
spectra and plot.
73
and phase
3.5 For the waveforms in Prob. 3.3 determine the complex Fourier coefficients
using the impulse function method.
3.6
shown
figure.
A AA
ST,
3T.
~r
Tj
T,
-t
PROB.
3.7
L,
3T,
51
~T
~i
3.6
figure.
f(t)
/
/
-T
^C*
*T
(b)
(>
f(t)
f<t)
V
.1
3V
(c)
Id)
PROB.
3.8
3.7
(a)
/ - A <3(0
(*)
/(f)
A sin atf
in the
74
Network
anal/sis
and synthesis
3.9 Prove that (a) if /(f) is even, its Fourier transform F(ja>) is also an even
function; (6) if /(/) is odd, its Fourier transform is odd and pure imaginary.
F(ju)
+0
(i)
PROS.
3.10
shown
domain?
as
3.10
in the figure.
= P 6(fo -
Wo)
+ P 8{(o +
eo )
chapter
Differential equations
INTRODUCTION
4.1
This chapter
is
equations.
5.
The differen-
tial
/MO, *'('),
where
t.
*< B >(0,
t]
(4.1)
t is
The
respect to
t,
namely,
>/a
x(t)
d(i)
x(t)
~df~
(42)
F
in Eq. 4.1 is x(t) and must be obtained as an
of /. When we substitute the explicit solution x(t) into F,
the equation must equal zero. If fin Eq. 4.1 is an ordinary
linear differThe
solution of
explicit function
ential equation,
it is
(B>
* (0
+ a^x^Xt) +
Network
76
Ordinary.
anal/sis
and synthesis
our case,
(in
t).
for
partial derivatives.
is
a linear
+ 2z(0 = sin
differential equation.
3[*
is
(4.4)
and
5t
(4.5)
by the
An
respectively,
as
JUt) and/,(0 such
AO-afM + btff)
(4-6)
x(r)-ai(0
(4 - 7 )
+ &*40
that the
where a and b are arbitrary constants. It should be emphasized
kept in
be
should
and
important
extremely
is
superposition property
mind
4.2
in
differential equations.
HOMOGENEOUS LINEAR
DIFFERENTIAL EQUATIONS
homogeneous,
This section deals with some methods for the solution of
let us find
First,
linear differential equations with constant coefficients.
the solution to the equation
x'(t)
- 2*(<) =
(4.8)
we
Ce"
(4.9)
Ce2 *
obtain
2Ce* f
- 2Ce* =
tA
4.8,
tn
(4.10)
77
Differential equations
It
differential
vt
obtain the solution of any differential equation, we substitute Ce for x(t)
for
which
the
equation
is
in the equation and determine those values ofp
an *<">(*)
we
let x(t)
+ +
4.11
+ a, x(f) =
! x'(t)
(4.1 1)
becomes
(4.12)
+ a# + a,) =
+ a^tf*- +
pt
Since e cannot be zero except at/? = oo, the only nontrivial solutions
1
Ce'Kanp"
H(p)
- <W" + a-tf"-1 +
'
W+
(4.13)
(4.14)
solutions
t .... Cw_ 1c*-' are all
of Eq. 4.11. By the superposition principle, the total solution is a linear
combination of all the individual solutions. Therefore, the total solution
of the differential equation is
x(t)
Q**'
<:_!-
(4.15)
, C
,
x,
B_i are generally complex. The solution x(t) in Eq.
, Cn_x are uniquely
not unique unless the constants C, Cls
specified. In order to determine the constants Cit we need n additional
pieces of information about the equation. These pieces of information are
usually specified in terms of values of x(t) and its derivatives at t
0+,
where
C C
= CV' +
4. IS is
= 0+.
x'(0 ),
to solve
t = 0
*(0 ),
=
values
at
In
order
determine
the
f
0+
we
must
^"-^(O ),
for the constants C{
This problem arises when the forcing
If the initial specifications are given in terms of
is
x(0+)
= Ce*=C
(4.16)
Network
78
so that x(t)
is
analysis
and synthesis
uniquely determined to be
x(t)
Example
4.1
*'(/)
given the
= 4e*
initial
5x\t)
+ 4a</) =
(4.17)
conditions
a<0+)-2
= -1
x'(0+)
Solution.
= p* +
H(p)
which factors into
+ 4X/> +
(/>
5p
1)
(4.18)
(4.19)
characteristic values)
= Qe-* +
x{t)
From
(0+)
x(0+)
2,
we
(4.20)
C^e-**
Q+C
=2 =
(4.21)
In order to solve for Cx and C, explicitly, we need the additional initial condition
x'(0+) = -1. Taking the derivative of x(i) in Eq. 4.20, we have
= -Qe-' - 4Q*-4
X '(f)
Atf =0+,a;'(0is
a:'(0+)
Thus the
final solution is
Next,
we examine
multiple roots.
root p
=p
(4.23)
we find
that
=i
a*/)
the case
(4.22)
-Q -4C
= -1 =
Ci
C% = J
= %r* - \e~**
when
Specifically, let
(4.24)
of multiplicity k as given by
H(p)
= an(p -
k
)
(p
-pd---(p-pn)
(4-25)
be left to the reader to show that the solution must then contain k
terms involving e" * of the form
It will
x(i)
CV"
+ C01 te*>* +
+
+ C^*** +
Cnt'e***
+ C^t*-^"*
Cje "' +
+ Cne*'*
(4.26)
where the double-scripted terms in Eq. 4.26 denote the terms in the solution
due to the multiple root, (p #,)*
79
Differential equations
Example 4.2
- Sx'(t) +
16*(r)
a<0+)2 and
with
Solution.
H{p) ~p*
Since H(p) has a double root at/>
x(t)
In order to determine
=4,
final solution is
"=4
is
16
= (p - 4)
(4.28)
and C we evaluate
a<r)
x(f)
(4.29)
and
'(') at r
= 0+
to give
= C1 =2
(4.30)
=4Q +C =4
*'(<>+)
Another
*'(<>+)
(4.27)
C^* + Cy*"
(0+)
Thus the
ip
= 2c*' - 4te*'
(4.31)
roots.
H{p)
- a (p -pMp-pJ
(4.32)
Px,Pi
The
e*"
is,
ojo>
(4.33)
form
Ci<*W">*
<V'>'
by Euler's equation,
x(t)
(4.34)
can be expressed as
Cje^cos
cor
y sin cat)
(4.35)
which reduces to
x(t)
= (Q +
C,y* cos
cor
+j(C1
CJe'*
sin cor
and Af so that
(4.36)
x(r)
may be
expressed in the
x(t)
M! =
d + C,
^. =y(C! - C2)
The
constants Afj
initial
conditions.
and
(4.37)
(4.38)
manner from
Network
80
and synthesis
analysis
Ml = MsiD *
= M cos
we
M and
<f>
+ 2*'(/) + 5x(t) =
=
a<0+)
The
*'(<>+)=
H(p)
characteristic equation
+ 2/7 +
H(p) =/>*
= (p +
is
+j2Kp +
x\t) is
The
= 0+
a>
2.
(4.43)
(4.44)
= Mx( -"' cos 2/ - 2e"' sin 20 + M -e~* sin 2/ + 2c" cos 2t)
(4.45)
'(<)+)
x(t)
we had used
+ 2MS
-Afj
If
(4.42)
derivative of x(i) is
*'(*)
At
-j2)
x(t)
At
we have a = -1 and
Then
(4.41)
conditions
initial
Solution.
(4.40)
<f>)
x\t)
with the
<f>
x{i)
Example 4.3
(4.39)
<r*(cos 2/
we
find
(4.46)
and
+ sin 2t)
M, =
+$.
(4.47)
we would have
obtained the
solution
x(t)
Vfe-* sin
[2t
tan"1 (2)]
The
*<(/)
The
initial
differential
+ 9*
(4
>(f)
equation
(4.48)
illustrates everything
is
conditions are
a!
U)(0+)=0
x'(0+)
= -1
5c
(0+)
x'(0+)
(S,
we
a<0+)
(4.49)
Differential equations
Solution.
H(p)
81
is
(4.50)
H(p)
- (p +
+yi)0
- Mi*-
cos
-;1)0>
+ 2)ty +
(4.51)
as
3)
Qte"**
+ Qe-3
(4.52)
a<0+)
a;'(0+)
x'(0+)
a;)(0+)
*<((> +)
- Mx + C + C8 = 1
- -Mx + Mt - 2C - 3Ct + Q - -2M, + 4C - 4Q + 9Ct = -1
= 2Mj, + 2Mt - 8C + 12Q - 27C, =
= -4M1 + 16C + 81C, - 32CX =
Mi -
C =
-1
(4.53)
1
we obtain
Q -i
=0
C.
is
a</)
(4.54)
upon
the roots of
its
characteristic
TABLE
4.1
Forms of Solution
Roots of H(p)
1.
2.
3.
=p
e p,t
Ce*i*
Cje'i*
M C* cos mt + Mge"*
+ CfcV'-W
sin ait
+ +)
Network
82
4.3
NONHOMOGENEOUS EQUATIONS
a nonhomogeneous
equation is one in which the forcing function f(t) is not
identically zero for all t. In this section, we will discuss methods for
obtaining the solution x(t) of an equation with constant coefficients
differential
an *<>(*)
+ a^ x-(0 +
+a
=f(t)
x(t)
(4.55)
Let x p (f) be a particular solution for Eq. 4.55, and let x e (t) be the solution
in Eq. 4.55.
of the homogeneous equation obtained by letting f(t)
It is readily
seen that
<t)
is
(4-56)
is
= 0+. 1
nonhomogeneous
equation if it
In Eq. 4.56, x p(t)
differential
t
= xjf) + x (0
satisfies
is
is
the specified
initial
conditions at
is
the comple-
example,
For
derivatives.
if/(f)
x P(t)
let
us take/(0 to be
/(0=<*e"
where a and
/S
is
the
is,
xjf)
and
We
(4.57)
unknown
coefficient.
= Ae"
To
(4.58)
determine A,
we simply
substitute
AefiXaJ"
a,,.^"-
+ aj +
= oe"
(4.59)
ed.),
Differential equations
83
Example 4.5
0.
**(r)
with the
initial conditions,
Solution.
The
a<0+)
1,
x'(0+)
= 1.
+ 3/> + 2 = (p + 2Xp +
1)
is
= cie-* + c*-*
= 4,
=> 1.
^ = _i_ = ?
Then
H(l)
Thus we obtain
xp(t)
= |e*
total solution is
*(')
To
(4.61)
characteristic equation is
Hip) -/>
The
(4.60)
(4.62)
initial conditions,
namely,
40+) =
*'(0+)
Solving Eq. 4.63,
we find
"Q+Cg+f
(4 ' 63)
= -1 - -Ci - 2C, + f
that
Q=
a</)
. _ e-' +
1,
C,
= $.
Consequently,
+ fe*
(4.64)
initial
Next,
let
= a.
/(r)
=a=
ae*
(4.65)
Network
84
that
we
is, j8
= 0.
analysis
For the
and synthesis
differential
*(0
and
= Cltr* + C*r + 2
(4.67)
When
the
= 4,
see that
method of undetermined
fit)
coefficients
and Eq.
4.60.
we can still
make use of
Suppose
(4.68)
x Plit)
From the
Re
MO + J Im MO
we can show
superposition principle,
=
fit) =
fit)
if
if
o>t
then
a sin tot
then
a.
cos
(4 ' 69)
that
= Re MO
*,(0 = Im * rt(f)
x(0
mt or a
Consequently, whether the excitation is a cosine function a cos
iat
<xe
driver
exponential
;
an
fit)
use
can
we
or,
sine function a sin
integral.
particular
resulting
the
of
part
imaginary
or
real
the
then we take
Example 4.6
x'it)
Solution.
First, let
(4.70)
= 2e*
(4.71)
rt(0
- Ae*
(4.72)
we determine
the coefficient
H(j2)
Then
MO
=/>*
to be
_
-5 +jlS
=,
L^tan-Hs)--]
5 VlO
1^ o^ ^*2
is
+5p +4
* -'5VI0
(4.73)
474>
<
Differential equations
and the
85
= 2 sin 3/ is
xjf)
5V10
*]
(4.75)
= A; constant.
= A(t n + Vi' +
+ V + b );
pt
3. /(f) = e
real
or
complex.
p
!
/(0
2. /(f)
'
n, integer.
4.
linear
network
analysis, the
1, 2,
method
or
3.
more than
is
adequate.
The
particular integral
xp(t)
where the
4.4
AND
In this section
p = a+jm
can be written as
= (AJ* +
coefficients
STEP
= At*e pt
A& A_lf
4^ +
.
. ,
...
A A
lt
+ A 1t + A )e*
(4.76)
are to be determined.
IMPULSE RESPONSE
we
As
physical
initial
conditions at
From Chapter 2,
= 0.
=
=
^
f <
f
Network
86
analysts
and synthesis
oo
=
and
tj*0
<3(0 df
JoJo-
in addition,
we have
the relationship
d(t)
= i*@
dt
As
the definitions of d(t) and u(t) indicate, both functions have dis-
continuities ait
0.
drivers,
solution x{t)
x'(t),
= 0.
and
its
derivatives
In other words,
may be
it
that
*<>(()-) ji z<>(0+)
x (-i)(0_)
x(0-)
In
many
^ a;<"-"(0+)
5*
*(0+)
conditions at
is
= 0
we must
discussed here
The method
is
often
Consider the
differential
a n *<>(t)
(4.77)
To
insure that the right-hand side of Eq. 4.77 will equal the left-hand
x0) must contain an impulse.
one of the terms z (B, (f), * (n_1) (0>
The question is, "Which term contains the impulse ?" A close examination
shows that the highest derivative term x ln) (t) must contain the impulse,
because if * <n-1) (f) contained the impulse, x (B) (f) would contain a doublet
C d'(t). This argument holds, similarly, for all lower derivative terms of
*-u
(n>
(f) contains the impulse, then s
x(i). If the term 3
(0 would contain a
{n~*
that,
We
conclude
therefore
step and x
for an impulse
(0> a ramp.
at i = 0.
discontinuous
derivative
terms
are
highest
the
two
forcing function,
side,
'The Green's function is another name for impulse response; see, for example,
Morse and Feshbach, Methods of Theoretical Physics, McGraw-Hill Book Company,
New York,
1952, Chapter 7.
87
Differential equations
For a step forcing function, only the highest derivative term is discontinuous
at
0.
0 , our task is to
t
determine the values a; (B) (0+) and * <n - 1, (0+) for an impulse forcing
function. Referring to Eq. 4.77, let us integrate the equation between
Since initial conditions are usually given at
= 0
and
)xM
an
(t)
dt
Jo-
AP
==
0+, namely,
+ a^ f ^'-"(O dt +
Jo-
a ]x{t)dt
Jo-
= A f %*)<*'
Jo
(4-78)
.
After integrating,
aB [*(-(0+)
we
obtain
a,
- *<-(0-)] +
(0+)
=A
(4.79)
are continuous at
1)
= 0.
an [x<-1 >(0+)
xl
so that
- ^"(O-)] = A
^v(0+) = + * *-"(0-)
(
At
all
(4.81)
is
+ a^ *<"-(0+) + + Oo *(0+) =
derivative terms below ( 1) are continuous, and
an *<">(0+)
Since
(4.80)
(4.82)
since
we
- - -- [0^x^X0+) +
ai x'(0+)
*(0+)]
flo
<*n
(4.83)
For a step forcing function A u(t), all derivative terms except x<">(f), are
(B)
continuous at t
0. To determine a;
(0+), we derive in a manner
similar to Eq. 4.83, the expression
*<>(0+)
= - - -
-1>
[a^1 x(B
initial
(0+)
conditions
ao*(0+)]
when the
(4.84)
forcing function
simplified
by the visual
Network
88
shown in Eqs.
process
its
analysis
and synthesis
4.85
OnX^Kt)
ax<\t)
It
oB_1a: (B 1 W
OaV- ^) + (Wf^O +
1
if a derivative
doubletit
for example, a
+ a^^-'KO + + ao^O =
<*'(')
+ flo^W = *0
(4.85)
(4.86)
For
*'(/)
we assume
3'(0
2a</)
(4.87)
4<5'(/)
= 0:
= A S'(t) + B 8{f) + C (0
x'(f) - A 8(t) + B u(t)
x(t) = A u(t)
*"(0
A d'(t) + B (5(0 +
Or
in a
Cu(t)
we
(4.88)
obtain
+ 2A u(f) - 4d'(t)
(4.89)
(4.90)
+ I A 8(t) +
3B (r)
A d'(t) +
Equating
(B
3A)
like coefficients
<J(0
on both
A=
=
C + 3B + 2A =
B+
3A
4
(4.91)
89
Differential equations
B = 12
C=
and
Therefore, at
28.
= 0,
it is
true that
**(*)
we
evaluate the A, B,
conditions at
example,
(4.92)
Once
For
if
= -2
*'(0-) - -1
*(<)-) = 7
*(<)-)
=-2 + 4 = 2
= -1 - 12 ex -13
**(()+) = 7 + 28 = 35
x\0+)
(4.93)
is
evaluated the
initial
conditions at
= 0+,
we can
= 0.
Having
z(0
From
Eq. 4.93
we
= C <r* +
x
C*tr
(4.94)
readily obtain
*(0
= (-9er* +
ll<r-) u(t)
(4.95)
function
90
Network
Example 4.7
analysis
and synthesis
+ 4x'(t) +
2x"(f)
where /(f)
<5(f)
and/(f)
I0x(0 =/(f)
(f), respectively.
The
initial
conditions at
are
a<0-)
Solution.
Let us
contains an impulse
a ramp, and
we use Eq.
*'(0
Note that we
x"(0-)
and a step
Thus x(0+)
0.
x term contains
= x(0 ) = 0. To
4.81
- - + *'(0 -) -
+)
=\
a2
+) and
x'(0
Next,
- 2(p* + lp +
- lip +
5)
is
+jlKp +
(4.96)
2.
2.
H(p)
We
therefore continuous at
is
obtain a;'(0+),
first
*'(0-)
-j2)
a<0+)
x'(0+)
(4.98)
= J = 2Mcos $ -
M=
**) =
\.
(4.99)
sin
<f>
Next we must solve for the step response xu{t). For convenience,
complementary function as
The
particular integral
constant /(f)
is
(4.100)
let
us write the
(4.101)
so that
a,
The
for
=0 = Afsin^
and
xe(t)
(4.97)
jt) _!_ = J-
(4.102)
(4.103)
Since x'(t) and x(t) must be continuous for a step forcing function,
a
<0+)=a<0-)=
a;'(0+)-a; , (0-)
=0
(4.104)
Differential equations
*(*)=
d/dt
C*
cos 2,)]*)
WO
L-
*,>
(4.105)
>*,(/)
FIG.
Note
4.1.
and the
by the equation
x,
(0
i.
*(<)
at
(4.106)
> *j(0
C*
L_
--*(0.5sin2/+
0.1[1
d/dt
> *(0
*,(')
is
91
Let us substitute
_|
.(*)]
equation,
we
+ 4 ^ x.(0 +
.(*)
. h(0
10*(0
(4.107)
we have
+ 4 1[| *(o] +
io[^ *(*)]
<H0
(4.108)
follows.
we can
We
INTEGRODIFFERENTIAL EQUATIONS
4.5
In this section,
we
an
will consider
form
a *"(t)
a,.!
*-\t)
x(t)
a_t f x(r) dr
= /(/)
(4.109)
Jo
a n x (n+1 \t)
+ a^ xM (t) +
consists
a x\t)
a_t
of a change of variables.
x(t)
-/'()
(4.U0)
a, y(n+1)(t)
(n
*_! y \t)
y'(t)
a_x
y(t)
= f(t)
(4.111)
Network
92
Note that from Eq. 4.110 we obtain x(t) directly. From Eq. 4.111, we
obtain y(t), which we must then differentiate to obtain x(t). An important
point to keep in mind is that we might have to derive some additional
the
initial conditions in order to have a sufficient number to evaluate
unknown
constants.
Solve the integrodifferential equation
Example 4.8
x'(t)
+2f
3a<0
*(t)
dr
(4.112)
5(0
is x(0 ) = 1.
Since the characteristic equation of Eq. 4.112 is of second degree,
obtain *'(<)+) from the
need an additional initial condition x'(0+).
The
initial
condition
Solution.
we
We
given equation at
= +
+ 3<0+) + 2
Since x(t) is continuous at / = 0,
a< T) dr
*'(0+)
(4.113)
<H
a<T)</T=0
(4.114)
(0+)-a<0-)-
(4.115)
I
and
x'(0 +)
Therefore,
-Method
1.
+ 3*'(0 + 2a<0 =
is
initial
conditions for
= Car* +
a<0+) and
x(t)
2.
(4.1 16)
Letting y'(t)
becomes
We know that
obtain
(4.117)
5<K0
the
MO +
y'(P+)
C*r*>
*'(<>+),
- 4e~' -
= x(t),
y"(0
we
then
*.(0
Method
- 3s(0 +) = 2
Using the
(4.118)
we
3e-*
original
2y(t)
= (0+) -
differential
equation then
5(0
1
(4120)
,,,,
(4.121;
j,'(0+)= *'(<>+)= 2
From Eq.
4.120, at t
= 0+, we obtain
y(0 +)
i[5
j<0
(4.122)
can be determined as
- -4e- + |e-* + \
93
Differential equations
Differentiating y{t),
we
obtain
<i)
4.6
(4.124)
with
dependent variable
In this section,
x{t).
variable.
we
The methods
described here,
x *'(0
first
the
fi,
(4.125)
<$!
where a
H(p)
is
+
+ Yo>
(a^
oto)
(ftP
is
+ A)
(<V + to)
<fc/>
(4.126)
The
that
is,
- OV + PoXYiP + Vo) =
It is seen that a nontrivial solution of H(p) =
exists only if
(tf>
(*
)(V
*)
+ KV +
<>a)
* OV + &XftP + y<0
(4.127)
(4.128)
Assuming that the preceding condition holds, we see that H(p) is a seconddegree polynomial tap and can be expressed in factored form as
H(p)
where
C is a constant multiplier.
The complementary
= K^e** + K***
y(t) + Kf
x(t)
(4.129)
functions are
(4.130)
Network
94
x,
9,
If
/>i
if H(p)
i.e.,
then
x(t)
(#,
y(t)
(tf,
+ Kt t)e*<*
K^e***
(4.131)
Pi
= a jto
Pi*)
then
x(t)
M^e"*
sin (cot
<^)
y(f)
<f>J
(4.132)
Example 4.9.
2x'(t)
x\t)
with the
initial
2x(0
+ y'(0 + y(t) =
conditions
*'(0+)2
y '(o+)=-3
a<0+)=0
Solution.
The
(4.134)
1
characteristic equation is
#0)
so that
i/(0+)
+4
+2
we find
that
=/>*
5/>
(4.135)
+ 6 = (p + 2Xp + 3)
y{t)
a</)
- KtfT* + J>-
Kjfi-"
+ A^-8
(4.136)
'
(4.137)
-s
Differential equations
Next,
95
differential equations.
+ aye + piy + p = Ne
Yi*' + y<P + b y' + d& =
We first assume that
xp(t) = Ae"
alX '
'
iff
(4.139)
yv(t)
(4.140)
= Be9
+ xJA + (^0 +
(.YJ + YM + O*i0 +
Qtfl
The determinant
H(8)
+
+ Y*
i0
= A(0) =
where H(6)
is
(4.141)
is
PiO
<,
YiO
=N
d )B =
o)B
+ Po
+\
0.
(4.142)
We now determine
#4he undetermined coefficients A and B from A(0) and its cofactors, namely,
A=
*A
1X (0)
A(0)
(4.143)
JVA ia(e)
B=
A(0)
where
A is
Example
4.10.
+ Ax + y' - y = 3e**
(4.144)
x'+2x+y'+y=0
The
*,(/)
- Ae**
y,(t)
- Be**
(4.145)
= 4 is
2(4)
H(A)~
(4)
+4
+2
(4)-l
(4)
42
(4.146)
Network
96
analysis
and synthesis
A The incomplete
B - -t
solutions are
*(f)
y(t)
(4.147)
we
obtain
finally
s
y(0=K-4" '-34')
(4.148)
Example 4.11.
+ 4x + y' + ly =
x' +x +y' +3y =
lx'
given the
initial
Solution.
First
a<0-)
we find
A(/,)
2/>
/>
+4 / +
(4.150)
p +
simplifies to give
H(p)
=(pt +2p+5)=(p +
The complementary
and
-'
functions x c (t)
xe(t)
y e (t)
The
(4.149)
56(f)
conditions
J5T(Cp>
which
5(/)
= ^e
-'
= jBiC
j/
(f)
+j2)(p
-j2)
(4.151)
are then
cos It
+ A&-* sin 2/
cos 2/
+ B^r* sin 2f
(4.152)
>
0,
namely,
+ 4x + y' + ly =
*' +
+ y' + 3y -
2x'
5
(4.153)
a:
as,
= Q;
yv
= Ca
#(0)
A(0)
(4.154)
Differential equations
5AU(0)
n
c
Dd
= 5(3)
^--5(or
97
3
(4.155)
c,
The
general solution
is
y(t)
In order to find
The
then
x(t)
y'(0+).
A lt At Blt B* we
,
values for
0 and f
+ 4* + y' + ly)dt -
0+. Thus,
5(f)<//
Jo-
Jo/0+
(*'
We know
/VH-
(2*'
impulses at
(4.156)
+ x + y' +
3y*<ft
(4.157)
fH-
5d(t)dt
continuities at t
r
r
(4a;
After integrating,
+ ly) dt =
(4.158)
(x
3y)rff0
we obtain
2a<0+)+y(0+)=0
(4.159)
*(0+)+y(0+)5
Solving,
we find
a<0+)
To
find x'(0+)
and y'(0+), we
= -5
y(0+)
10
(4.160)
= 0+.
Thus,
- 20 + j/'(0+) + 70 = 5
x\0+) - 5 + y'(0+) + 30 =
x'(0+) = -20
y'(0+) = -5
2z'(0+)
so that
(4.161)
(4.162)
x(t)
- ( -%e~* cos It -
y(/)
14e~* sin It
3) u(t)
(4.163)
1) (f)
Network
98
analysis
and synthesis
Problems
Show
4.1
that
and
3^(0
= Mxe~* cos It
xjj)
m,
M&-* sin It
+ 2x'(t) + 540 =
x'(t)
Show
that
x1
xt
is
also a solution.
4.2
+ 4x'(0 + 340
+ 8x'(t) + 540
*"(/) - 540
x'(0 + 540
x"(t) + 6x'(0 + 2540
z'(0 + 6x'(.0 + 940
conditions 40+) = 1,
**(0
(a)
x"(t)
(6)
(c)
(d)
(e)
(/)
4.3 Given the initial
solutions for
=
=
=
=
=
=
*'(0+)
1, determine the
+ 2540 =
&r'(0 + 16x(t) =
x\t)
x'(t) + 4.8lx'(t) + 5.7640 =
(a)
+
+
x'(t)
(b)
(c)
4.4
fo'(0
x"(t)
(b)
(c)
(d)
(e)
(/)
x"(t)
(.b)
x"{t)
(c)
*"(/)
4.6
(a)
initial
(b)
'
determine the
The
= e-8
= 2 sin 3f
= e~* sin It
= <r"/2
=6
= 2e-* + 3e~at
40+) = 1, x'(0+) =0,
+ 7x'(t) + 1240
x'(t) + 3x'(t) + 240
*'(0 + 2x'(t) + 540
x"(t) + 2*'(0 + Sx(t)
x'(t) + 5.0x'(0 + 6.25x(t)
**(/) + fo'(0 + 540
(a)
3y'(/)
Given the
differential
1;
- a(0
y'(0-) = 2. Find
2y(t)
equation
as<*>(0
14*'(0
+ 840 -
6<K0
Differential equations
= 6,
and x"(0-)
= **(0-) = 0,
(a)
99
-7, find
x'{t)
x"{t)
(*)
4.8
Given the
initial
the integrodifferential
equations
+ 4 f a<r) </t =
()
x'(t)
(b)
x'(t)
+ 2tft) + 2 f x{r) dr = | e-
(c)
*'(')
4.9
Given the
with s(0-)
y'(0+).
4.10
set
5a</)
(/)
<K0
*C)+y'(0 + 2y(0
- *'(0-) = y(o-) = y'(0-) = 0,
+ 3*(0 + y'(t) +
*' + *(') + y'W +
All initial conditions at f = 0 are zero.
6y(t)
6y(0
= */)
= (')
2s'(0
x'(0
The
of equations
2*'(0
4.11
2 sin
initial
2<5(0
3e~* uSj)
conditions are
a#)-)
*'(0-)=-l
y(0-)=0
j,'(0-)=0
and
chapter
Network
analysis:
5
I
INTRODUCTION
5.1
We
and with methods for writing mesh and node equations for a-c or d-c
1
We will, therefore, consider only briefly the problem of writing
circuits.
mesh and node equations when the independent variable is time t. The
problems in this chapter have the following format: Given an excitation
signal from an energy source and the network, a specified response that
is a current or voltage in the network is to be determined. When relating
laws,
is
the response.
unknown
variable x(t)
differential
correctly, or
whose
initial
closing at
is
Before the
1
For a comprehensive treatment, see H. H. Skilling, Electrical Engineering Circuits
(2nd Edition), John Wiley and Sons, New York, 1965.
100
Network
analysis:
101
*0
/\
Energy
source'
FIG.
switch
Network
5.1.
Switching action.
differential equations.
only for
Since
= 0-
all
^ 0+.
in
mind
FIG. 5 Jo
current sources.
102
Network
FIG. 5.2b
By
circuits.
the superposition
Network
principle, the total current
algebraic
T(t) due to
of the sources
all
is
103
equal to the
sum
h = \ + '.,+
5.2
analysis:
I.,
,,+
(5.1)
NETWORK ELEMENTS
In this section,
we
it is
important to assign
Before
we examine
these
first
the voltage across an element, and a reference direction of flow for the
current through the element.
5.3.
is
discussion,
at the tail
we
of the current
current relationships for the resistor, the inductor, and the capacitor,
which we
first
discussed in Chapter
1.
Resistor
The
resistor
shown
i(t),
namely,
K0 - r
i(0
where
m
(5.2)
= Gt*0
Capacitor
For the capacitor shown
dt
(5.3)
<t)
= -\
i(r)
CJo-
dr
+ o(0-)
-o
'
+
Ut)\
tR
ZZC
**>
v(t)
i(*>\
v(t)
={=C
(a)
FIG.
5.3. Resistor.
(b)
(A)
Capacitor
Network
104
<
ml\
v(t)
(J).-)
C is given in farads.
v(t)
(b)
(a)
FIG.
(6)
Inductor with
initial current.
where
The
derivatives
of impulses.
Inductor
-'5
(5.4)
f ir)dr+iL(0-)
LJo-
There are basically two ways to write these network equations. The first
way is to use mesh equations and, the second, node equations.
Mesh equations are based upon Kirchhoff's voltage law. On the mesh
basis, we establish a fictitious set of loop currents with a given reference
direction, and write the equations for the sum of the voltages around the
loops. As the reader might recall from his previous studies, if the number
of branches in the network is B, and if the number of nodes is N, the
+ 1.*
number of independent loop equations for the network is B
cit.
Network
anal/sis:
105
must, in addition, choose the mesh currents such that at least one mesh
current passes through every element in the network.
We
F.mpl 5.1.
In Fig.
5.6,
We therefore need 7 - 5 +
of the mesh currents
ilt
Vi(t)
Li
</W-
c,
Ci"
'1
biwQ
fCi'
R 2
vatQ
0^9-r Qhfi-i
FIG. 5.6
capacitors in the circuit have associated initial voltages. These initial voltages
are assigned reference polarities, as shown in the figure. Now we proceed to
write the
mesh equations.
Meshix
i(0
Mesh
c
I
if*
dr - J^t) dr
it :
vCl(0-)
Mesh
- *1 h(0 +
l
*C,(0-)
- vcfii-) = -
C*
^ J^r) dr+Li
di.
it :
-vJLO
+ vCt(0 -) =
^P
i/r) dr
^r
J\(t) dr+Ki tf
(5.5)
After
we find the
currents
to find
following relationships
*<>*-'*-*
(5.6)
'c,
- '
v*t)
is
- tff)4
(5.7)
Network
106
anal/sis
and synthesis
Datum
FIG. 5.7
the network
is
datum node.
Consider the network in Fig. 5.7. Let us write a set of node equations
for the network with the datum node shown. Since the number of nodes
1
2 independent node equations.
in the network is
3, we need JV
These are written for nodes vx and vs , as given below.
positive with respect to this
Node
v^.
Node
j
B (t)
L(0~)
L Jo-
ty.
i(0-)
. --
!
*
v t (r)
L Jo-
dr
+ 7 Jf
L
' V 2(r) dr
o-
+C
^& + G
t> 2
()
(5.8)
dt
5.3
INITIAL
AND
In this section,
FINAL CONDITIONS
solution.
cit.
Network
analysis:
107
network.
Initial
c(0+)
If
i(t)
at
= 0+
is
+ vc(0-)
(5.9)
Jo-
o(0+)
vc<0-)
(5.10)
clude that
circuit at t
= 0+ is a short circuit.
When
there
Since
we
instantaneously. Therefore at
voltage source
initial
if
an
initial
charge
exists,
or by a short
circuit if there is
no
charge.
Example 5.2. Consider the R-C network in Fig. 5.8a. The switch is closed at
t = 0, and we assume there is no initial charge in the capacitor. Let us find
the initial conditions i(0 +) and i'(0 +) for the differential equation of the circuit
Vu(.t)=Ri{t)
e',(t)/t
(5.11)
i(t)
>B
R-C network
R-C network,
'
>+))
T
m
(b)
>R
A-
Equivalent circuit at /
- 0+
0+.
Network
108
The
and synthesis
analysis
equivalent circuit at
- 0+
is
we
obtain
>+>-5
To
obtain /'(u+),
we must
V8(t) -Ri'(t)
At
(5.12)
i(0
(5.13)
(0+)
-Jt/'(0+) +'
= 0+ we have
(5.14)
We then obtain
i'(0+)-
-i(0+)
RC
(5.15)
Fig. 5.8a is
final condition, or steady-state solution, for the current in
obtained from our knowledge of d-c circuits. We know that for a d-c excitation,
is
a capacitor is an open circuit for d-c current. Thus the steady-state current
The
i(0
Initial
= '() -
relationship at
= 0+
is
iKr)dT+iL(0-)
(5.16)
If there is no
iL(0+) i(0-).
which corresponds to an open circuit at
_ o+. This analogy can also be obtained from the fact that the current
t
through an inductor cannot change instantaneously due to the conserva-
L(0+)
= 0,
L^+JI*)
L
o.c.
-o
9
(a)
FIG.
5.9. (a)
R-L network.
s=:
o-
i(o+)J
W
(*) Equivalent circuit at / *=
0+.
initial
Network
From
5.9b,
analysis:
we
i(0+)=0
+)
/'(0
109
see that
(5.18)
(5.19)
V R
- - - - /(0 +)
(5.20)
as
The
V
L
steady-state solution for the circuit in Fig. 5.9a is obtained through the
/(')
=/(>)
is
a short
circuit.
(5.21)
Example 5.4. In this example we consider the two-loop network of Fig. 5.10.
As in Examples 5.2 and 5.3, we use the equivalent circuit models at t +
and t = oo to obtain the initial conditions and steady-state solutions. At
f =
the switch closes. The equivalent circuits at t = 0+ and / = <x> are
shown in Fig. 5.1 la and b, respectively. The initial currents are
*i(0+)
ig(0+)
The
V
^~
*i
(5.22)
) - j^r^ = ")
<5 - 23 >
When
is
KOo)l
sin
K< - <K<*J\
(5.24)
where
is
Network
10
analysis
and synthesis
*2
VWV
h(0
-VW\r
^=:C
3
kM
FIG. 5.10
JVW\r
V!=
ww*
2(0+)
i(0+)J
Equivalent circuit at
O.C.
= 0+
WW-
-WWo
o.c.
Equivalent circuit at t
FIG.
5.1
1,
= 0+.
itfO
Y
FIG. 5.12
(6)
>
Equivalent circuit at
oo.
'
Network
form shown
analysis:
III
in Eq. 5.24,
IHM>)I
|rOo)l
(5.26)
h
(G a
#o>
and
so that
We
5.4
vjLO
refer to this
^w
tan"
co
2/~*\W
*C*j
(G>+
problem in Section
(5.27)
"
sin (>o<
tan"
^)
(5.28)
5.5.
STEP
As an
equations, let us consider the important problem of obtaining the step and
As we shall see in
Chapter
7,
is
stated as
we
initial
F,e * 5 * 3
due
If the excitation
is
the circuit
is
KO =
We assume Vq(0) =
x'(t) for j'(0 in
0.
*)
= R 1(0 +
l(r) dr
7
J~
CJi>C
I
ii
(5.29)
we substitute
= Rx'(t) + z(t)
(5.30)
Network
12
and
0+
gives
*>+>-The
(5.31)
characteristic equation is
H(p)
and with
little
effort
- Rp +
(5.32)
we have
_ L ,-t/RC n(0
*(0-
m-
so that
R^'RC
6"'*
(5.33)
"]
(5 ' 34)
We
which
is
response
shown
i(t)
m
'
aro
R*C
(*)
R-C circuit.
(A)
Network
analysis:
113
R-C circuit.
(A)
Jt-C circuit.
Assuming zero
initial
Jo (0
Gt<0
steady-state value
C^p
(5.35)
characteristic equation
H(p)
The
is
of KO
Cp
+G
(5.36)
is
(5.37)
H(0)
m
/o
Network
14
analysis
and synthesis
FIG.
5.17.
From the
initial
to pulse excitation.
is
(5.38)
= 0, we obtain K =
IRC u(t)
t>(0 = JoR(l - e-*
)
condition (0+)
I^R
so that
(5.39)
f(0
= l* e -tlRC (0
are plotted
(5.40)
on
Figs. 5.15a
T)]
and
b.
Suppose
KQ = IMt) shown
in Fig. 5.16.
u(t
K0 =
which
5.5
is
V*[(l
shown
(5.41)
- (1 -
e-
(i
would be
- T)IBC
)u(t
T)]
(5.42)
in Fig. 5.17.
this section
will apply
Network
instance, if the
mesh
write
unknown
quantity
On
equations.
the choice
quite arbitrary.
is
a branch
is
if
current,
for example,
it is
we wish
If,
analysis:
115
preferable to
to find a voltage
In many cases
to find the voltage v
better.
we wish
Example
network
source
in Fig.
is e(t)
when
5.18,
the voltage
Solution.
*(/)
5.5.
differential
= Rl(t) + ^ j
equation
i(r)dr
is
+ v^O-)
=fcC-if
(5.43)
we
have
2~" u{t)
= i(0 + 2
(t)
dr
(5.44)
2(5(0
To
obtain the
the limits
initial
= 0
and
we
obtain
dm + 2i(t)
- e- u u(t) =
dt
condition i(0+),
t
FIG. 5.18
= 0+
we must
to give (0+)
(5.45)
= 2. From
the characteristic
equation
H(p)=p+2*0
we obtain
/<#)
If
(5.46)
we assume the
A = The incomplete
= JCr
solution
_1
//(-0.5)
(5.47)
= Ae~
bt
,
then
we
obtain
(5.48)
is
(5.49)
(5.50)
is
Network
16
alone and
is
analysis
and synthesis
On
independent of excitation.
is
sinusoids
= ^
/!
This
a response that
network. If a
network
are zero or negative, and if theyo> axis roots are simple, then the
unstable*
is
network
the
which it describes is said to be stable; otherwise,
will be discussed
Stability is an important property of passive networks and
in greater detail later.
Example
5.6,
p^O-) -
v(t)
it is
by
0.
Solution.
Now
let
node
basis is
We
C -r + Gv = I
(5.51)
sin mtiKt)
at
-Cp+G
(5.52)
vJfl-KT* *
so that
solution
(5-53)
is
we
t<0
)
0,
condition
From the initial
*/)
a>C\
Fig. 5.20
we see that,
Consequently,
but
it
(5.54)
moment.
(555)
Network
analysis:
117
lm
A
A
o
C
1
Be
MG.SM
From the complementary function
the circuit is T - C\G - RC.
in Eq. 5.53,
Differentiating
The
LJ*-
dt
S.S8,
we have
(5.59)
H(p)
- Cp + Gp + 1 =
c(j>
+ 1 P + j^)
(5.60)
is
#(p)
= G(p-J>i)0>-/>i)
(5.61)
-)
nasai
118
Network
and synthesis
analysis
v(t)
+ K2
AB
Pi
where
2C
P*
LCI
2l\c)
(5.62)
case
B is
or imaginary.
Bis
1.
a,>
lc
(1
then the free response
is
vdt)
(5.63)
case
2.
5=
0, that is,
(-T-\Cl
LC
= p = A
At
vc(t) = (^ + K t)e-
then
p1
so that
When B
(5.64)
0,
the response
is critically
v(t)
o
FIG. 5.23. Critically damped response.
5.23.
Network
analysis:
119
case
3.
Letting
B is imaginary,
damped
is,
B = jfl, we have
c(0
In
that
response
is
LC
(*i sin
said to
fit
+ Kt cos
(5.65)
fit)
be underdamped, and
is
shown by the
Example
5.7.
In this example we discuss the solution of a set of simultaneous
network equations. As in the previous examples, we rely upon physical reasoning
rather than formal mathematical operations to obtain the initial currents and
voltages as well as the steady-state solutions. In the network of Fig. S.25, the
switch S is thrown from position 1 to position 2 at t = 0. It is known that
prior to t = 0, the circuit had been in steady state. We make the idealized
assumption that the switch closes instantaneously at / = 0. Our task is to
find i(0
and
/,(/)
Vl and V% are Vx = 2 v, Vt =
3 v;
L=
C=
The mesh equations
for
ix{t)
Rt = 0.5 n
= 2.o n
lh,
jig
if,
and
it(i)
after
are
Vt =Li\{t)+Rl i1{i)-R1 i)
-vc(0-)= -/Mito +
(5.66)
cL ^T)dr+{R
k
+R^i^t)
(5.67)
Network
120
analysis
and synthesis
t.o^onnnr^
?2
v2 ^:
hit))
FIG. 5.25
we differentiate it to
give
(5.68)
Using Eqs. 5.66 and 5.68 as our system of equations, we obtain the characteristic
equation
Lp +'*!
-Ri
H(p)-
-+(R1+ RJp
-RiP
<569>
dement values
H(p)
The
into H(p),
- 2.5p* + 4p +
1.5
we have
(5.70)
(5.71)
The
mesh
is
oo
by
then a short
(5.72)
Now let us determine the initial currents and voltages, which, incidentally, have
sources are not
the same values at / - 0- and t - 0+ because the voltage
Network
impulses. Before the switch is thrown at
analysis:
121
(0-)-K1 2v
ro
0-)-^-4amp
(5.73)
0-)-0
We next find i\(0+) from Eq. 5.66 at / - 0+.
Vt ~Li\{0+) + 4*i<P+) - RiW+)
Substituting numerical values into Eq. 5.74,
'
From Eq.
5.68 at / <-
amp/sec
*!+*. /'i(0+)
values of
initial
we find
*'>+)
With these
1(0+)
(5.74)
it(t)
and
i/t),
- 0.2 amp/sec
we can
(5.75)
solutions
/i(r)
i/f)
5.26.
6.0
5.5
h(t)
5j0
45 4.0,
o.io'
0.08
\w
0.06
0.04
O02
r
(>
0.5
1.0
IS
2.0
2.5
FIG. 5.26
3.0
33
40 43
Network
122
and synthesis
analysis
ANALYSIS OF TRANSFORMERS
5.6
may
for
inducing the
sufficient
conditions
current
it
Ox
in
the coil
to the
In
this section
we
containing a device
made up of two
known as a
equations
VO-L^ + M^
at
at
(5.77)
, x
du
di t
at
at
flux linking
Lt to Lt
M = k4IT
x
The constant
bounded by the
K in
limits
Eq. 5.78
(5.78)
is
It is
Ox in
K=
For
circuits
with transformers,
we must
dijdt.
Network
analysis:
123
Voltmeter
ttransformeri
An experiment
FIG. 5.28.
which we connect the positive lead of the voltage source. The dot reference
is placed on the secondary terminal at which the
voltmeter indicates a
positive voltage. In terms of the primary current i the positive
voltage at
lt
the secondary dot is due to the current i flowing into the dot
on the
t
primary side. Since the positive voltage at the secondary dot corresponds
to the current
If
and
linkages of Ly
and
it
If
are the
N&
2 $ linkages = N&i + N^
however, one of the currents, for example,
other ig flows out of the other dot, then
If,
ilt
(5.79)
2 $ linkages = N& - N Q
t
(5.8O)
An
i\(t)
R* i8(0
FIG. 5.29
i'tf)
+ L, I'M
(581)
Network
124
analysis
and synthesis
= 0
and
= 0+
results in
the determinant
Za&(0+)
By
(I^L,
If
t
M)[i1(0+)
K<
obtain
*1 (0-)][i,(0+)
/,(0-)]
>
M*, that
is,
fl
(5.83)
1,
(0+)
*(0+)
and
K<\
*<1
- /,((>-),
-
ig(0-),
5.8.
Solution.
The
differential
6(0
The
L^Lx
we
(5.82)
(5.84)
(5.85)
Lt 6ii(0-
1A,
2h,
Let us find
(5.86)
by the determinant
Wj
+8
= 0> + 3)(2p + 8)-/>
= (^+14/ + 24)
-
(/>
2^
+ 2)(p +
(5.87)
12)
= Kx<r + A^e-"'
= K*r + JV"m
Ie (0
ile{t)
(5.88)
To
V_
hJt
Oamp
we
rely
upon
amp
(5.89)
Network
analysis:
125
Since the excitation does not contain an impulse (and since I^L,
jt 0, as we shall see later), we can assume that
d(0+) and ^(0+) are
Af1
also zero.
5.86
+ f,(0+)
o -r<p+) + 2i',(o+)
1(0+) - 12 and i'0+) - 6.
6'"i(0+)
Solving,
ditions,
(5.90)
'
we find i'
With
we obtain the final solutions of ix(t) and %(/)
(5 ' 91)
0 - (-Ir* + *e-0 (0
Suppose, now, I^L, - M, that
K - then i,(0 and (/) need not be
continuous at 0. In fact, we will* show that the currents are discontinuous at >
for a unity-coupled transformer. Assuming that K = 1,
is,
1,
consider the
mesh equation
*t 'i(0+)
= -M i'^O-H) - L, .',(0+)
(5.92)
(5.93)
V - Rt
it
(0+)
[L,.
= 0+
i\(0+)
side then
becomes
(5.94)
We
iPi(0+)
it
+ MM>+) - i^O-)] =
5.82. Since x (0-) = i^0~) = 0,
(0-)}
is
(5.95)
we
solve
^3
1^(0+)
R\JL%
wo+)
=RL
1
i=4h,
At-8 0,
M = 2h,
+ R^Li
(5.96)
+ R^
L,= lh
R9 =*3Q
K=10v
in Fig. 5.29
Network
126
anal/sis
and synthesis
The
/2 (0-
is
closed at
on mesh
basis are
10-4^
+ 8^ + 2^
dt
dt
(5.97)
dt
The
characteristic equation
dt
is
4(p
2p
2)
H{p)
2p-
- 20p +
H(p)
which yields
(p
24
(5.98)
3)
= 20(p + |)
(5.99)
(5.100)
The
we
(r)
= 1 = 12 =
(5.101)
hit)
The
initial
conditions are
*i(0+)
i*(0+)
We then find Kx =
VL,
=
RtLi
+ R^
-0.75 and
20
(5.102)
-VM
-!2-a5
=
K = - 1.0 so that
t
= (-O.lSe-11 +
W) = -er^u{t)
*
ijff)
We
see that as
state value
of
approaches
1.25.
-1.0
infinity
it(t)
1.25) u(t)
- 0, while
(5.103)
it (t)
goes
its
steady-
Network
analysis:
127
Problems
5.1
nnnp
nnnp
Cl <0->
Pi
*^\{=
'l(0-)
2<0-)
*~-,
Ci
*1
QWb(
te1
(0-)=J;C2
PROB.
5.2
Write a
set
J8 2
5.1
figure.
W
ii
wm{(T)
SJlo
Ci
*2>
= = G>
2(0
c <0-)
PROB.
5.2
53 The network shown has reached steady state before the switch S is
opened at t = 0. Determine the initial conditions for the currents
1-fy) and
4(0 and their derivatives.
R
o^f o
vW-
^WV
2WJ
1
PROB. 5.3
5.4
from a to
b.
derivatives.
moves
Determine the initial conditions for iL(t) and vdt) and their
Determine also the final values for iL(t) and vc(t).
first
128
Network
1
10 v
lh_
vc (t)
*t
5v-
==
If
:io
T
PROB.
5.4
5.5 The network shown has reached steady state before the switch moves
from a to b. Determine the initial conditions for the voltages vx(,t) and vt(t)
and their first derivatives. Determine also the final values for v^t) and vjj).
t0
vi(t)
J:-
*i
=C
R2 *
v 2 (t)
i
PROB.
5.6
(a)
Find
(A)
Show
(c)
5.5
= 0.
- 0.
ItfOQ)
PROB. 5.6
(r)
d(t)
- tr*u(t)
and t<0-)
= 4.
Find
Network
10
I
v/W-
analysis:
129
o
m
\k*t)
1Q
W(J)
PROB.
5.8
5.7
b, steady-
6<?
PROB.
5.9
For the
been in steady
5.8
circuit
state.
at
had
>Tnnp
/^
PROB.
U\
5.9
5.10 Find the current /(f) in the network shown when the voltage source is a
unit impulse. Discuss the three different kinds of impulse response waveforms
possible depending upon the relative values of R, L, and C. All initial conditions
are zero at/ 0.
130
Network
anal/sis
and synthesis
nnRp
Qm
*>J
PROB.
5.10
5.11 An R-C differentiator circuit is shown in the figure. Find the requirements for the R-C time constant, such that the output voltage v^t) is approxi-
uoW
PROB. 5.
5.12 An R-C integrator circuit is shown in the figure. Find the requirements
for the time constant such that the output t> (f) is approximately the integral of
the input voltage.
S
A/V
O"
ci
")
PROB.
i(f)
vo(t)
5.12
in the figure
shown
for (a)
a current
Network
analysis:
131
Source
PROB. 5.13
At t = 0, the switch goes from position 1 to 2. Find /(/), given
= e~* sin It. Assume the circuit had been in steady state for t < 0.
5.14
e(t)
PROB.
5.15
initial
that
5.14
For the circuit shown, switch S closes at t = 0. Solve for l(t) when the
conditions arc zero and the voltage source is e(t) = sin (cat + 0). What
-VNAr
I"
^)
PROB.
5.15
at
= 0.
state at
Find
f
= 0.
Network
132
analysis
and synthesis
MA
->
Xs
1
PROB.
5.17
For the
circuit
shown,
i(r)
5.16
- 4r (0-
Find
i<f);
- </<
JQ^
lf=t=
PROB.
ih
5.17
in Prob. 5.4.
5.18 Find the complete solution for i L(t) and vdO
Find
t - 0.
5.19 For the circuit shown, the switch is closed at
energy.
initial
zero
oo.
Assume
<t <
ijj) for
10Q
irff)
ana
lh
10Q
PROB. 5.19
For the transformer circuit shown, the switch
and i2(f). Assume zero initial energy.
S
5.20
ijft)
5i>.
Af=lh
PROB. 5.20
closes at
t = 0.
Find
Network
Q>
30
PROB.
lh<3 1hg>lh
analysis:
133
60^02(0
5.21
switch S opens at
5.21 For the transformer circuit shown, the
. Assume zero initial energy.
/
the voltage far) for
- < <
= 0.
Find
chapter
The
Laplace transform
6.1
In Chapters 4 and
ential equations.
discussed classical methods for solving differsolutions were obtained directly in the time domain
5,
The
we
There
is
the product
c=axb
(6.1)
so that
If a
we have
six-digit
(6.2)
(6.3)
into addition.
is the use of transform methods to solve integroConsider the linear differential equation
An analogous process
differential equations.
yW))
-/()
134
(6.4)
135
Algebraic
Differential
equation^
equation
Transform
FIG.
6.1.
X(s)
^
x(t)
Solve
Invert
where /(0
and
Eq.
is
When we
let s
6.4,
= W)]
(6.5)
7W*(0)1
Since frequency
Eq-6.5as
Y(X(s),s)
where X(s)
TftOL
*W -
= F(s)
T[f(t)\,
and
(6.6)
Y(X(s), s)
is
an algebraic
We
*(0
r-
(6.7)
[*(*)]
Differential
equation
EM
System
m
RM
function
6.2
= Hs) =
{"/(&-" dt
is
defined as
(6.8)
no-
where s
is
a +ja>
(6.9)
M.
New
E.
Jersey, 1964.
Network
136
/
= 0
instead of
that/(t)
and synthesis
= 0+. We
C[d(0]
analysis
for the
It is clear that
definition.
it
condition
\f{t)\e-"dt<
oo
(6.10)
Jo-
form,
it
f
/
|/(0I
dt<
oo
trans-
(6.11)
waveforms have Laplace transforms but also because the Laplace transform takes directly into account initial conditions at t = 0 because of
the lower limit of integration in the Laplace transform. In contrast, the
Fourier transform has limits of integration ( oo, oo), and, in order to take
into account initial conditions
as/(0 u(t)
+ /(0-)
(3(0,
at
= 0, the forcing
The
/(0
is
= -M
FWds
(6.12)
Alt} Jniaa
where at
is
York, 1959.
137
by
recognition,
we must remember
To
certain basic
transform pairs and also use a table of Laplace transforms. Two of the
most basic transform pairs are discussed here. Consider first the transform
u(t).
Next,
let
~ e<(01
F(j)-J
With
these
e#e-dt=s
(6.14)
two transform pairs, and with the use of the properties of Laplace
we discuss in Section 6.3, we can build up an extensive table
transforms, which
of transform
6.3
pairs.
Linearity
finite sum of time functions
forms of the individual functions, that is
The transform of a
t[| /,]=
is
the
2 OK')]
(615)
This property follows readily from the definition of the Laplace transform.
= sin <ot.
Expanding
fit)
For a
by Euler's
identity,
we have
= ^.(e** - r+*)
Transients, Prentice-Hall,
sin cat
(6.16)
Electrical
38
Network
anal/sis
and synthesis
is
the
sum of
1/1
=-
C[sin cat]
\
1
yto
:
2/\*
+.//
=
**
(6.17)
>*
Real differentiation
Given that [/(f)] = *(*), then
Eflwhere /(0-)
is
By
Proof.
sF(s)-/(0-)
(6.18)
= 0-.
definition,
[A0]=J*V'7'(0<*f
Integrating Eq. 6.19
by
parts,
[/'(')]
(6.19)
we have
= -7(0
/(Oe-'df
(6.20)
JoSince e~" -
[/"(')]
as
-*
oo,
= ^) we have
t[f'(t)]
Similarly,
we can show
|"7(0"l
n
s
f(s)
integral
on the right-hand
= sF(s)-f(0-)
side
is
(6.21)
-i
s
/(0 _)
_ S "-V(0-)
- u (0-)
(6.22)
where f
"-"
is
the (
that differentiating
by
l)st derivative
in the time
of /(f) at
domain
is
= 0-. We
thus see
equivalent to multiplying
by
s in the
Example
6.4a.
/(f)
= sin cot.
Let us find
By
j sin
cot)
(6.23)
cgsin<of]=,(^)
(6.24)
C[cos
cot]
we have
The
i(_^ = _I 5
c[cosa,,]=
sothat
Example
6.4b.
unit impulse.
Laplace transform
139
(6.25)
0.
is
=-
Ct(0]
(6.26)
C[*0]
Then
since
= *(j) =
(6.27)
= 0.
(0 )
Real integration
= F(s)> then the Laplace transform of the integral of /(f)
If [/(*)]
flfa)
divided by
that
s,
c[/(r)dr]=^
By
Proof.
c|"
by
LX
parts,
/(t)
-.C
[-
/(T) dT
]
d'
(6,29)
obtain
=-
/(t) dr
s Jo-
J
as
e_rt
/(t) dr]
we
rfrl
LJoSince e~" -*
(6.28)
definition,
c
Integrating
is
is,
oo,
and
"
+-
o-
e-'/(f) dt
(6.30)
s Jo-
since
=0
/(t)<*V
Jo-
(6.31)
It-o-
we then have
c[/(r)dr]=^
Example 6.5.
We know that
Since
h(t)
dr
C[(0]
(6.32)
" t u(t).
(6.33)
p(f)
=-
(6.34)
then
C[p</)]
-^ . _
[(')]
(6.35)
Network
140
and synthesis
analysis
Differentiation by
Differentiation
by
multiplication
by
s in the
in the
domain, that
W)] =
From the
Proof.
is,
-^
(6-36)
as
definition
Given /(O
-Z[t f(t)]
mrrz
let
us find
C[te~"']-
By
638>
we find that
"-i--s(rh)-<rb
Similarly,
we can show
that
(640)
"a-^-crrb* 5
where n
is
a positive
Complex
By
integer.
translation
where a
is
Proof.
(6.37)
Jo-
ds
Jo-
ds
if
- a) = C[e7(0]
< 6 - 41 )
a complex number.
By
definition,
1*7(0]
Example 6.7.
-J"
^/(Oe"" dt
Given /(t)
J"
e"
(M)
7(0 dt
qsinorfl-^^
= F(s -
a)
(6.42)
Since
643>
<
Similarly,
c^coso,.]-
^;^;^
(-45)
141
[/('
Proof.
By
(6.46)
= f V*7(r - a) dt
(6.47)
a)]
definition,
C[/(r
- a) u{t -
new dummy
Introducing a
C[/(t) m(t)]
= e-** F(s)
- a) u(t -
f
Jo-
a)]
variable t
= a, we have
t
= g ['f(r)e-
e-lT^f(r) dr
dr
= <f~ F(s)
(6.48)
Jo-
we
e~"
G(s),
C-i[r~G(,)]-
we can
moment,
find
ft (0,
E-HG(*)]-('X
and then take
by
setting
g(t-a)u(t-a)=g1(t)
(6.49)
Example 6.8a. Given the square pulse /(f) in Fig. 6.3, let us first find its transform F(s). Then let us determine the inverse transform of the square of F(s),
i.e., let
us find
Mt) =
Solution.
is
/(/)
Its
Laplace transform
is
C-i[F(5)]
(6.50)
- (/) -uit-a)
(6.51)
then
fit)
F(s)=\(l-e~)
Squaring F(s),
F\s)
we
(6.52)
obtain
- -5 (1 - 2e- + e"*")
(6.53)
FIG.
4.3.
Square pulse.
Network
142
analysis
and synthesis
is
&]-
C-4-J -fN(<)
Then
C-[F*(*)]
/(/)-
2(*
a)u(t
(6.54)
(6.55)
so that the resulting waveform is shown in Fig. 6.4. From this example, we see
that the square of a transformed function does not correspond to the square
of its inverse transform.
AW
f(t)
K6
K2
Ki
Xb
it
Kq,,
I
a
FIG.
FIG.
Triangular pulse.
6.4.
Example
6.8b. In Fig.
of a train of impulses
fit)
6.5, the
6.5.
Impulse
2T!
Ti
2a
"
is
7Ti
train.
shown.
It
consists
+ * *' - ri)
656>
train is
- *o + *!*-* + ^-2,Tl +
+ *e-* r
e ,r
is
often used.
(6.57)
Then
we can
Ai1
W)]-*. + T
A.
+^ +
A
+?
(6.58)
The transform
PERIODIC
is
WAVEFORMS
[/(*)]
-}
f(tyr"dt
+Jt
f{t)e-'*dt
+
(6.59)
flufUT
f(t)e-*dt
InT
is
[/]
T
dt
\ f{i)e-'*dt
Jo-
e- T [
+ e-'
e-nT f*f(t)e-* dt
= (1 +
e-T
l-e-
+ e-aT +
(6.60)
T
\ f{t)e-*dt
Jo-
T
~
J
T Jo- f(f)e-'*dt
\
6.8c.
determine
f(t)e-*dt
Joo-
Jo-
Example
143
Laplace transform.
its
dt
-1
(6.61)
1 1
- ~"
- e-' T
m
1
a
FIG.
Example
6.8d.
T+a
2T+a
2T
represented as
/(/)
Its
iKt)
Laplace transform
F(s)
(1
{pM)
is
.)
(6.63)
=s
[1
- e~" +
e~,T(l
- e-) +
*-2 T(l
- -") +
Network
144
which
analysis
and synthesis
F(s)
(1
- -^"Xl + e- T + e~UT +)
(6.64)
is
At
is
the entries are obtained through simple applications of the properties just
discussed. It is important to keep those properties in mind, because many
transform pairs that are not given in the table can be obtained by using
For example, let us find the inverse transform of
these properties.
F(s)
=,
j_
(s
ay
(6-66)
,
car
we
can write
"
F(s)-a
(s
we
ay
(6.67)
co'
obtain
dt
sin
mt at
(6.68)
= (r sin oat +
= 0 is zero.
to
cos oot)e~*
is
An obvious example
-/:c-**sin5rdr
If we replace tr %%
of sin
5/,
which
(6.69)
is
Z[S ux5t]
= -rj-s + 25
(6.70)
2,
we have
I
TTs
/
First,
we note
that
eriW
is
is
(6 - 71)
the evaluation of
+
J
,W dt
(6.72)
= 2jVC-2'df
From
145
we
(6.73)
see that
= f tV**Jo-
(6.75)
vm -^tv
(<76)
/(0
Later
we
consists
pr//A1
ELKO]
= 0.25 "
0.25
Now
t
1,
^i - j^
0.5
6.77,
= 0.25(1 - e- -
we
2te-**
.-_
(6.77)
obtain
- 2rtr-)
11(f)
(6.78)
is
equal to 2/(f) at
+ 0.5 = 0.162
(6.79)
is,
= 2/(1) =
-2.5<r*
6.9.
differential
equation
(6.80)
Network
146
Solution.
analysis
and synthesis
X(s)
3[s X(s)
(**
(s*
We then obtain
3s
+ 2) X(s)
and
simplifying,
we have
+5+2
(6.82)
X(s) explicitly as
-jr=mrm+%
To
(6.81)
= f'^W],
(6 - 83)
we expand
fractions
The
A^i,
and
Kt algebraically, we obtain
K_i = JTi = -1 Kt = |
a<0
transform of X(s) or
fe*
- e-' + fe-
(6.85)
Example
6.10.
Given the
2x'(f)
set of
is
solved classically.
4a<r)
(6.86)
+ x(t)
with the initial conditions x(0 -) = y(0 -) = 0, let us find x(j) and y(t).
x'(t)
Solution.
Transforming the
2(s
(s
set
of equations,
+ 2) X(s) + (* + 7)
+
1)
X(s)
+ (* +
X(s)
we
obtain
Y(s)
--
3) Y(s)
(6.87)
we have
+ 5^
(5/*)A u
A
(6.88)
Y(s)
(5/5)Au
+ 5A*
is
More explicitly,
cofactor of A.
X(s)
-5i
in partial fractions,
- 30s +
the
15
(689)
we have
Kx
+ Kz
K-s
^> = 7 + ?T27T5
Multiplying both sides of Eq. 6.90 by s and letting *
K1
is
is
~-*FT2r+sr
Expanding X(s)
A tf
147
<690)
= 0, we find
=sX(s)\,_ =3.
- 36
* "-7+5 + 5
(691)
is,
s*
As a
result of
+ 2s +
(s
1)*
+4
(6.92)
*<'>-"
8(i
fr
In similar fashion,
Wrt
y(s)
The
we
_
=
(3
+ 1) + 14(2)
+ iy+gy
(693)
is
8c-' cos 2/
(6.94)
obtain Y(s) as
"* + 17
1
+
~s 7T27T5 - -; +
*
+ l)+3(2)
+ i)* + (2)*
llfr
(,
(6,95)
y (j)
= ( _ i +1 i e-* cos 2f +
(6.96)
This example
We
= 0
In addition, we obtain both the complementary function and the particular integral in a single operation when we
use Laplace transforms. These are the reasons why the Laplace transform
method is so effective in the solution of differential equations.
the
initial
conditions at
Network
148
analysis
and synthesis
PARTIAL-FRACTION EXPANSIONS
&5
As we have seen, the ease with which we use transform methods depends
upon how quickly we are able to obtain the partial-fraction expansion of a
given transform function. In this section we will elaborate on some simple
and effective methods for partial-fraction expansions, and we discuss
procedures for (a) simple roots, (b) complex conjugate roots, and
(c)
multiple roots.
m=
D(s)
Since the degree of N(s)
is
s'
3s*
+ 3s + 2
2s
we
divide D(s)
F(s)
'
= 5 + 1s
+2s +
(6.98)
Here we see the remainder term can be easily expanded into partial
However, there is no real need at this point because the
denominator s2 +^2s + 2 can be written as
fractions.
s*
+ 2s +
= (5 +
1)*
(6.99)
so that the inverse transform can be obtained directly from the transform
tables,
namely,
C-W)] =
From this
form
example,
<5'(0
3(0
+ er*(sm t - cos i)
Example
6.11.
(6.101)
Find the
role.
partial-fraction expansion of
2*
+3
(JTWTT)
(6102)
The
If we see that F(s)
(s
Find the
+ 1) + (s + 2)
+ Dfr+g
(6103)
fr
6-12.
149
*W-
Exanpfe
Laplace transform
is trivially
partial-fraction expansion
of
s +5
j
+5
F(s) = , T -,
(6-105)
= _i_ T _i_^
mW = feai
+
+
+
2)
(s
(s
(6 106)
2)
(s
2)
Real roots
we examine a method
m=
- s tt^t:
)(s s0(s St)
(6-107)
(s
real roots,
< 3.
(6.108)
(s
s)F(s) = K
H
s
If we let 5
=*
in Eq. 6.109,
we
obtain
Ji-(#-J)J^)|^
Similarly,
we
(6.109)
s,
(6110)
general relation
A,
1
6.13.
(6.111)
Z)
(6.112)
K
150
Network
anal/sis
and synthesis
Ao
sF(s) |_>
+ 2s -2
+ 2X* - 3)
s*
(*
-o
+2s -2
-3) a
s*
Kl
+ 2s - 2
s2
*"
(6.113)
sis
5
13
~15
3
s(s+2)
Complex
roots
Equation 6.111 is also applicable to a function with complex roots in
its denominator. Suppose F(s) is given by
F(s)
JVC*)
D&Xs -
Ki
s-x-jfi
where A^/Di
is
- JPK' - * + JP)
+
K1 =
JVi(s)
s-x+jp
(6.114)
Djis)
ypDM+m
(6.115)
-irfD^-m
where we assume that j = a jP are not zeros of Z>i(s).
It can be shown that the constants Kx and K% associated with conjugate
roots are themselves conjugate. Therefore, if we denote K as
(6.116)
K2 = A-jB = K *
then
If
we denote the
we see that
fx(t),
A(0
i(o
(6.117)
_i
e"'(K ie "
2c**(A cos Bt
Kft-"*)
- B sin pi)
(6.118)
151
A more convenient way to express the inverse transform f^t) is to introduce the variables
M and
<X>
= 2A
McosO = -25
where
terms of
and
M and
obtain
<I>,
When
we note
lt
6.116.
In
(6. 120)
that
M cos O + jM sin O =
Mf =
6.14.
Kr in Eq.
is
Example
(6.119)
M and $ from K
Me'*
<1>
/i(0
To
sin
we
see
(6.121)
*<+Jfl
(6 . 122)
w-vr^rkr+ii
For the simple root s
2, the constant
(6123)
.
K is
K = (*+2)F(*)|,__8 =
(6.124)
**+2j+5=(* +
We see that
= 1,
Me**
+j2Xs
-y'2)
(6.125)
= 2, thus
*
+3
2(*+2) i1+*2
is
V5
c-i(tan-j+W)
(6.126)
then
- e-**
+ = <r* sin \^
~\~ tan-1 2j
(6.127)
= - e~u
5
2
-p e~*
V5
Multiple roots
Next let us consider the case
-1
tan 2)
cos (2/
We
in
will
Network
152
analysis
and synthesis
Method
W=
F(S)
(6.128)
- s )" ^(s)
The partial fraction expansion
with multiple roots of degree n at s = j
(s
of F(s)
is
Kt
Kx
(s-s ) n
(s-so)""
..... *-i
s-s
(s-s )-8
*i(?)
J>i()
(6.129)
Ko
However,
if
we were
is,
= (s-s rF(s)\
(6.130)
to use the
same formula
dition.
Ku
define
F1(s)s(s-s rF(s)
(6.131)
Thus
Fx(s) -
A,
+ Kt(s - s ) +
(6.132)
we
differentiate
Eq. 6.132 by
obtain
Fi(s) - *i + 2K^s 4
as
s )
*_!(/!
D(s
- so)""* +
(6.133)
Kx = Ft(s)
It is evident that
(6.134)
ds
Kt =
F^s)
(6.135)
2 ds
in general
K = -tWs)!
!->
.
j\ ds'
= 0,1, 2
-1
(6.136)
The
Example
6.15.
Laplace transform
153
'W-STTlJi
which we represent-in expanded form as
<'>
is
A=sF(s)\^=-2
To obtain
Fjis)
= (s +
l)
(6.139)
8 F(s)
-^
(6.140)
Using the general formula for the multiple root expansion, we obtain
^i=T7T
1 ds
!
fs
-2\
!_!
,_,.
=2
-3
|__!
I
<6142)
*-i^)L-(-5)L-;
IW __2 gl +
sothat
i+
-i
-5
(6.144)
Method B
for arriving at the partial-fraction expansion for
*(p
Dividing N(p
+s)
fiG*
+ tf-?T^
by
D%(p + s
we
obtain
),
(6145)
+ s ) = K + K p + Ktpt +-- +
1
X^j,""1
f"f
+ *o)
0l(l>
(6.146)
* I
am
Institute
me this method.
Network
154
The
anal/sis
and synthesis
F(P
o)-
is
F (p + s ) by the equation
related to
-,-%-!+
pn
Dl(p
+ So)
(6.147)
Substituting 5
F(s)
* +
g
/
\ Bl +
-^ + -\
(6.148)
We
F^s)
Setting p
=s+
1,
we
(s
fr
(6149)
ijUa
is
+ If F(s) =
-^
(6.150)
then have
- lp + 2/
l+/0"2
2 +2/>
-2^
-2p-2f
AT = 3, we terminate
We then have
is
1)
(d -1)
=-FJx7T- -
? ?
(6.152)
+~ -pTl
p
(6153)
= />, we have
m
Example
As a second example,
6.17.
F(,)
Since
we have two
sets
= 0;
(6155)
=^T1?
= 0, and s =
1) we have a
Let us arbitrarily choose to
s here, we do not have to make any substitutions.
since /j
(6154)
F&) is then
(at s
Ooi s = 1.
expand about *
155
2+s
=(7TT? = rT2FT7
Ft(j), we obtain
Expanding
F^s)
is
then
F(s)
+ 4)
^ + iy
3* + 4
+
+^
- 3j +
2
2
F(s)
-^ -
Fortunately
we
3s
answer
POLES
(6.158)
term
1)*
is
+4 = 3Qr +
1)
(*
1)
iy
1
3_ +
+ I (s +
(6159)
1)
then
"W-l-i+TTT + frTi?
6.6
(6.157)
(s
final
s\3s
+4
3*
(s
The
<6156>
(6160)
AND ZEROS
In this section
we
many
implications of a pole-zero
We
_ ^-i+;D(s-i-Jl)
+ l)\s+j2)(s-j2)
s =
1
(double)
s = y*2
(s
+y*2
156
Network
and synthesis
analysis
jo
|
plane
j2 i
.t
-<r
,1
i_W_
Double
pole.*.
12
-3 -2 -1
->2 x
-J<0
FIG.
diagram of F(s).
6.7. Pole-zero
at
00
-j\
For
frequency domain as
C[u(0]
=-
6162>
function cos
single pole at s
<a
i,
C[coso,
(6.163)
Oq
-o-* as indicated in
whose transform
The exponential
C[e']
which has a
Fig. 6.8a.
Fig. 6.86.
The
cosine
is
t]
= -r-1-~t
(6-164)
7<u
poles at s
has a zero at the origin and a pair of conjugate
diagram
pole-zero
the
shows
6.Sd
Figure
as depicted in Fig. 6.8c.
is
corresponding to a damped cosine wave, whose transform
Z[e-"* cos
co t ]
g
= s +r=-*
(s +
r+
<r
(6.165)
o>
The
Laplace transform
157
I
(a)
(b)
]W
m)'
""
M,
+,
*******
-a
x
-jao
id)
FIG.
6A
From these four pole-zero diagrams, we note that the poles corresponding to decaying exponential waves are on the a axis and have zero
imaginary parts. The poles and zeros corresponding to undamped
on the/o> axis, and have zero real parts. Consequently, the
and zeros for damped sinusoids must have real and imaginary
sinusoids are
poles
>
<r
functions are
*i()
=
s
ax
(6.166)
F&) =
s
a,
JU>
s plane
*2(K
Fi(K
0"1
V2
(a)
FIG.
*.*. Effect
(b)
Network
158
analysis
and synthesis
JU
1
X./C02
JO)
X
,
>i
X
1
a
X
si*
(
-jwi
X
2*
<
-j(i>2
FIG.
FIG. 6.10. Pole locations corresponding to sin to,/ and sin mj.
6.1
represents frequency of
that the distance from the origin on theyco axis
frequency.
oscUlation; the greater the distance, the higher the
responses fx {f) and
time
the
compare
let
us
thumb,
of
rules
these
Using
shown in Fig.
corresponding to the pole pairs {* lf s 1 *} and fo, s a *}
a (0
sinusoids.
damped
see that both pairs of poles corresponds to
6.11.
We
than/a(f)
sinusoid/i(0 has a smaller frequency of oscillation
s t Also,
of
part
because the imaginary part of * x is less than the imaginary
time
The
Res
because
2
1
(t) decays more rapidly than fx (t)
The damped
fz
responses /x(0
>Kcs
and/a(0
are
shown
in Fig. 6.12.
f2 (t)
- lme*
h) =
\^" ~~-
C-
lme'
___ __
-
=*-
S ^-^
(b)
in Fig. 6.11.
FIG. 6.12. Time responses for poles
159
ju
plane
m
at
-m
(b)
Let us examine more closely the effect of the positions of the poles in
upon transient response. We denote a
pole pt as a complex number (
at +jco i For a given function F(s)
p
with only first-order poles, consider the partial-fraction expansion
the complex frequency plane
F(s)
The
_*!L_
s-
+ _*!_ +
(6.167)
s-P
-Pi
Po
is
= K e + Kie* +
= K^"*e im,t + K1e'
+K
lt
iait
ne"*
+ + K^e*""*
K e"
i
*e
is,
at
(6.168)
> 0, then
iatt
We
unstable
is
axis,
such as for
the function
F(
"-(7f^'
is
shown
in Fig. 6.14a.
(6.169)
is
/(0
= - sin co
2io
(6.170)
160
Network
analysis
and synthesis
Vjao
*-.7o
f(t)
m ^-
sin (dot
(b)
(a)
they'd) axis
upon time
response.
N(s)lD(s).
Consider the system function H(s)
numerator and denominator polynomials, we obtain
H(s)
- Zq)(s - Zi)
- P )(S - Pi)
Hg(s
(S
If
- g)
(S
Pm)
(s
we
factor the
(6.171)
It is clear that
we can
whether the
behavior of the system. As we have seen, we can determine
the jm axis
and
poles
for
plane
right-half
the
checking
by
system is stable
its transient
concerning
information
obtain
We
poles.
for multiple
discuss
behavior from the positions of its poles and zeros; and, as we will
function
(j).
W"
and the
(6.172)
(i
- ftK " A)
excitation
s
(6.173)
pt
Then
H(s)
161
is
- Zq)(s - i)
- p )(s - pj(s - p)
gpEofr
(s
(6.174)
R(s) contains the poles and zeros of both H(s) and (*),
except in the case where a pole-zero cancellation occurs. As an example,
let us take the system function
It is clear that
H(s)
2(5
(s
2+j4)(s
1)
(6.175)
+ 2-j4)
whose pole-zero diagram is shown in Fig. 6.15c. It is readily seen from the
pole-zero diagrams of the excitation signals in Fig. 6.18a that the step
response of the system has the pole-zero diagram indicated in Fig. 6.156.
The response to an
tion of Fig.
we
zero plot,
plot
6. 1 5c.
numerator on the
itself.
know
we
an
inte-
ju
J(0
ju
J*
K-2
JT-2
JT-6
j2::
^r
-2 -1
-2 -1
-j2x
-;4-
(a)
= 3 cos 2i.
-74
-74-
(b)
(jb)
(0
excitation,
(c)
Response to
162
Network
course,
is
true.
differentiator
analysis
We
and synthesis
an integrator must
origin.
EVALUATION OF RESIDUES
6.7
What
role
where we
s
shall
Si
at
oo.
The
inverse transform
is
/(0
fV
(6-177)
we
discussed a
number of
different
diagram. Suppose
we
f(s)
are given
= 4>(8 ~ goX5 ~ ^
(s - p )(s - Pi)
where
m>
n and
all
F(s)
Our
task
is
(s
( 6 .i78)
-^ +
= -&- +
S Po
*
Pi
^ = (s-ft)^(s)
~ 2 )
- pj
(s
F(s)
as
+ -=* Pm
(6-179)
We know that
(Pi
'
'
(6.180)
JU
ozo
Pix
from a zero
where
/>),
163
t.
^ k,
P0
pole
is
t,
to the prod-
To
{.
A*
*1
zero plot of
F(s)
=
(S
The
F(s)
Ms - zp)(s - Zl)
- p )(s - p )(s - ft*)
- K
s
(6.181)
Kl
Pt,
pt
is
^1
(6.182)
-Pi-
where the asterisk denotes complex conjugate. Let us find the residues A"
and Kx by means of the graphical method described. First we evaluate x
by drawing vectors from the poles and zeros to/^, as shown in Fig. 6.17a.
The
residue
x is
then
K1 _
XqAB
(6.183)
CD
Network
164
and synthesis
analysis
to evaluate
ju
C-3
/>
We see that
6.176.
K - A RL
(6.184)
MN
-,
the
as indicated in Fig.
-1
we
the vectors so that the residues can be determined quickly and easily. Consider the fol-
-jl
55
F(s)
FIG. .I8. Pole-zero diagram
of-
2)(s
+ 1 - jl)( + 1 + jl)
F(s).
is
shown
(6.185)
The
in Fig. 6.18.
F(s)
at
1 +jl
are
is
Kt*
=
s
"*"
+ jl
(6.186)
s
+2
x.
shown
+ - jl
1
in Fig. 6.19a.
Kx =
X -p
We see that Kt is
^2/135
J2 1*5_ X
2/90
_3
2
JU
C-3
AIVr
-2/\&
-<r
180*
nJ
'
V90*
r
(a)
-a
(b)
the pole
The
From
Fig. 6.19*
^_
K,=!
Laplace transform
x 2_
= _3
V2 /-13S x 72 7+135
+ l-jl
THE
6.8
In
INITIAL
this section
transforms.
The
value of /(/) at
infinity,
that
AND
we
is
+ l+jl
(6.187)
+2
will discuss
two very
value theorem.
first is
the
= 0+
initial
is,
lim f(t)
The only
Kt to be
residue
F(s)
165
= lim s F(s)
(6.188)
at most,
step discontinuity at / == 0.
From
parts.
The
function /(f)
the relationship
Cf/'(0I
we
obtain
Therefore
continuous at
is
= 0,
that is,/(0 )
=/(0+).
hm C[/'(01 -
lim s F(s)
-oo
- /(0-) =
(6.189)
(6.190)
(6.191)
-oo
(b)
at
= 0.
Let us represent
D (/),
We
/(0-/i(0
where
D =/(0+) /(0-).
The
+ J>W
derivative of/(f)
f'{t)=f\(f)
Do{t)
(6^2)
is
(6.193)
Network
66
anal/sis
and synthesis
Du(t)
fi(t)
fit)
Since fx (t)
is
continuous at
we know from
0,
lim s F^s)
= A(0-) = /(0-)
which
simplifies to
of Eq. 6.193,
sides
(6.194)
we have
(6.195)
(6.196)
Now, if we take the limit ofEq. 6.196 as s-+ oo and let/(0+) -/(0-) =
D we have
(6.197)
lim s F(s) = lim s F x (s) + /(0+) - /(0-)
By Eq.
6.194,
we then
obtain
limsF(s)=/(0+)
Example
6.18.
(6.198)
m=
2(s
s*
1)
+2s +
(6.199)
us find/(0 -f ). Since the degree of the denominator is greater than the degree
of the numerator of F(s), the initial value theorem applies. Thus
let
lim s F(s)
Since
2(s
\)s
lim
C-H^Cj)]
=2
=2c-*cos2r
(6.200)
(6.201)
6.19.
theorem does
we see that/(0+)
3.
(5(0
+ 3r*
(6.202)
is
+
s
(6.203)
1
so that
= lim si 1 +
.-co
*-*,
oo
(6.204)
/
states that
= lim s F(s)
lim /(0
167
(6.205)
provided the poles of the denominator of F(s) have negative or zero real
parts, i.e., the poles of F(s) must not be in the right half of the complexfrequency plane. The proof is quite simple. First
f'(t)e-' dt
- /(0-)
s F(s)
i:
Taking the
limit as s
->
/'() dt
lim
j;
Evaluating the integral,
/(oo)
(6.206)
we have
s F(s)
- /(0-)
(6.207)
-o
we
obtain
Consequently,
/(oo)
= lim s F(s)
(6 .208)
(6.209)
-o
Example
6.20.
/(/)
3(/)
+ 2e~*
(6.210)
let
= iii
*(s + l)
=l +
us find the final value/( oo). Since the poles of F(s) are at s
the final value theorem applies. We find that
(6 . 211)
=0 and s = 1
we see
]imsF(s)=3
which
is
Example
6.21.
we see that
Given
from Eq.
f(t)
lim/(f)
But from
sF(s)
= le*
= oo
lim * F(s)
6.210.
(6.213)
(6.214)
7s
=
s
we have
(6.212)
(6.215)
(6.216)
-o
We see that the final value theorem does not apply in this case, because the pole
= 1 is in the right half of the complex-frequency plane.
Network
168
TABLE
6.1
Laplace Transforms
F(s)
fit)
1.
FW-JT/Oy**
fit)
2.
i/i(0
3.
|/W
4-
^/<
+ ./)
CxFxC^+fljF^)
fW
-/(0-)
dn
5.
a" Fis)
f/W
!*w
rfT
6.
f
f fir)drda
Jo- Jo-
7.
(-0"/W
8.
9.
10.
fit -a)u(t
- a)
>(,)
e-F(*)
**/X0
F(s
8(t)
- a)
dn
1
12.
(/)
13.
14.
-r
s
1
n\
1
15.
16.
/?- (
17.
sin cof
*
1
e-
- J f^'f^O-)
a
1
_ e-ft)
a>
TABLE
6.1 (cont.)
s
18.
cos
19.
sinh at
20.
cosh a/
21.
"'sin or
22.
*""' cos
cor
+ <0
a
i*-fl
0>
(*
(5
(* + )
+ )* +
mt
CO
co
e-**t*
23.
(j
24.
25.
/(/); n -0,1,2,3,...
*)
a)**"1
*
.
sin co/
2co
0*
<*)
(5*
a)W[(i
nth order)
26.
J-
(t/)-*
T(*
27.
r*
(A:
1)
Problems
6.1
(a)
/(0-sin(orf+|8)
(6)
/(/)
(*)
(/)
f)-td'(t)
(g)
/W - aW* - 4)]
(c)
(</)
6 J,
>1.
- cos (t -
*/4) u(f
(a)
/(/)
(6)
/(0-cos(/-W4)(0
w/4)
a)H
- 5]-
169
170
6.3
Network
analysis
and synthesis
fit)
-1
(a)
6.4
(b)
Prob. 6.3.
6.5
(a)
F(s)
(6)
{S)
2s
=
(s
"
+9
3X*
+ 4)
5s - 12
+ 4s + 13
4* + 13
s*
GO
all the inverse transforms may be obtained without resorting to
normal procedures for partial-fraction expansions.
Note that
The
Laplace transform
171
Find the Laplace transform of the waveforms shown. Use (a) the infinite
(6) the Laplace transform formula for periodic waveforms.
Both answers should be in closed form and agree.
6.6
series
method and
fit)
fit)
uv\\A
3T
2T
2T
3T
-A
lb)
la)
*t)
+1
r/
2
>
/
2T
-1
PROB. 6.6
(e)
6.7
Id)
()
sin It dt
6.8
(a)
x'{t)
(b)
x'(t)
(c)
x"{t)
id)
x"{f)
(e)
It is
sj-CO
given that
6.9
cos 3f dt
a<0 )
= as'fl) )
+ 2as'(0 = (0 + *~* (0
= for all the equations.
sets
equations:
()
2*'(0
x'(0
(b)
2x'{t)
+ jr*0) +
The initial conditions are all zero at t = 0
*'(0
8y(0
(5(0
of simultaneous
172
6.10
Network
anal/sis
and synthesis
H0-
()
+ 9X' +
+a
(s*
j
()
3)
s+P
(*+!)
(c)
S*
s*+2s +2
tf>
+ 3* + 1
s* +s
s + 5
~(s + W* + 2)
*
to
F(s)-
(/)
ns)
(g)
F(s)
(A)
F(s)~
s*
6.11
+ lX*+2)*
3j - J* - 3s + 2
(*
sKs-W
+-*'
()
sH.s+2)
_ r-s
+ 3s + 2
->*
(ft)
F(')
(c)
*"(*)
se
**
re-"
s
+0
write
6.12 Given the pole-zero diagrams shown in (a) and (b) of the figure,
What
can
polynomials.
of
quotients
Fjs)
as
and
(s)
the rational functions t
you say about the relationship between the poles in the right-half plane and the
signs of the coefficients of the denominator polynomials?
FiM
ju
-A
A
>ij0.6
Fzb)
*
>>0.6
-1
+1
H-/D.6
|->D.6
-A
-A
to
PROS. 6.12
173
6.13 For the pole-zero plots in Prob. 6.12, find the residues of the poles by
the vector method.
6.14
plots
shown
poles.
-2
j2
jl<
<
-1
-jln
-J2
(a)
(b)
ja
f-H n
-<r-3
-2
-x-
-1
-1
+1
PROB. . 14
6.15 Two response transforms, R^s) and R/s), have the pole-zero plots
shown in (a) and (b) of the figures, respectively. In addition, it is known that
i(0 +) - 4 and /-,(/) - 4 for very large t. Find r^t) and rjf).
Sd)
ja
X
I
1
-V -2
-l
6
-1?
i:
'
(a)
-n
(b)
PROB. 4.15
Network
174
6.16
It is
analysis
and synthesis
given that
_.
F(s)
as*
+ bs + c
+ 2j* +s +
s*
Find
a, b,
6.17
(6)
F(*)
3X*
(*
IX*
(c)
1)
s* + 3s + 2
FW - * + 3* + 3j +
2
5<s
<
+ 4)
+ 4X* +
8)
^-"cr+ixr+Q
chapter
Transform methods
network
7.1
in
analysis
by the
(0
is
= [t<01
and
I(s)
= Cfl(f)].
*(')
V(s)
The
v-i relationship
RI(s)
7.1,
71 )
or
(7.2)
w- !
1
(7.3)
(0
= 7p
/(5)
we
t<T)dT-M(0-)
obtain
= sLI(s) - Li(0-)
=
J-F(S)
sL
+ ^;
i(O-)
s
175
CM)
176
Network
IM
IM
f\HO-)
VM
)i(0-)
FIG.
7.1. Inductor.
The transformed
Fig. 7.1.
v(t)
=-
'
i(r)
C Jo-
dr
is
depicted in
+ v(0- )
(7.5)
*>- c
are then
F -J; / + *&=>
sC
I(s)
sC
V(s)
(7. 6)
- C o(0-)
From this
analysis
we
and admittances in
For
example,
series or parallel with energy sources.
from Eq. 7.4, we see
that the complex-frequency impedance representation of an inductor is sL,
and its associated admittance is XjsL. Similarly, the impedance of a
capacitor is 1/sC and its admittance is sC. This fact is very useful in circuit
analysis. Working from a transformed circuit diagram, we can write
mesh and node equations on an impedance or admittance basis directly.
-%
IM
IM
VM
+
=PC (l)
K0-)
V<
Transform methods
in
network
analysis
177
vs(t)
vi(t)
FIG. 7.3
(7.7)
Transforming
this set
of equations, we obtain
+ sM Is) - M ijp-)
M ^(0-) + sL, Us) - U i,(0-)
Ly i^O-)
(7.8)
This set of transform equations could also have been obtained by representing the circuit in Fig. 7.3 by its transformed equivalent given in Fig. 7.4.
In general, the use of transformed equivalent circuits is considered an
easier
way
v2W
FIG. 7.4
178
Network
analysis
and synthesis
-nnnp
fi,(0-)
+__
+
12
BC (0-)
FIG. 7.5
-.+.)
*->
-l
+ A'i,(0-) = -^/
1(5)
'm^Qi
^
I2
Li
(7.9)
^>
vuo'
L1M0-)
1*1140-)
QviW
FIG. 7.6
Example
f
= 0.
7.2.
amp, o(0 )
= 2 v.
is
is
thrown, the
initial
i(t)
Qv(t)
>'
FIG. 7.7
to 2 at time
lh
-x
=2
Transform methods
3
ivw
network
179
anal/sis
=0
nnnp-
in
/i\
FIG. 7.8
is
closed at
formed
circuit.
Solving for
The
I(s),
+2
now redrawn
in Fig. 7.8 as
(3
trans-
is
+s+ ?) 70)
C7.10)
we have
+3
1
(7.11)
+ IX* + 2) s + 1 s + 2
i(0 = C-HAO] = e~* + -
(7.12)
Consider the network in Fig. 7.9. At t = 0, the switch is opened.
2s
(j
Therefore,
Examine 73.
vjCf)
and
Z,=Jh
G = 1 mho
C=
V=
is
given that
lf
1
L
i<0-)
+
C=fcoc (0-)
FIG. 7.9
Before
we
NodeV1
*_
_ (l(0-0 +
Node
Vt
CPc(o-)
/
\
= (sc
+ -)ki(o
1
-- vM
^-4 ()
(7.13)
Network
180
V2
FIG. 7.10
If
(t(0 ) =
of node equations, we have
1
state,
1 v,
(*+^K (i)-?K/f)
(7.14)
we obtain
1
1
we have
J+l
s +1
+ 2s + 2 " (s + l)1 + 1
* +2
s +2
""
(s + 1) + 1
s* + 25 + 2
KlW ."
Krf,)
(7.15)
(7.16)
7.2
THEVENIN'S
= e- cos
/,
*(/)
- <r*(cos t + sin /)
(7.17)
single
known
as Thevenin's
Transform methods
in
network
181
anal/sis
-3X
Networks
n voltage
<S)
sources
Zi(s)
in current sources
FIG.
7.1
Thevenin's theorem
From the standpoint of determining the current I(s) through an element
can be
of impedance Z^s), shown in Fig. 7.1 1, the rest of the network
replaced by an equivalent impedance Ze(s) in series with an equivalent
voltage source Vs), as depicted in Fig. 7.12. The equivalent impedance
Zt(s) is the impedance "looking into" from the terminals of Zt(s) when
all voltage sources in JV are short circuited and all current sources are
open circuited. The equivalent voltage source V,(s) is the voltage which
appears between the terminals 1 and 2 in Fig. 7.1 1, when the element Z^s)
is
removed or open
theorem is
circuited.
for Thevenin's
element in N.
and
not magnetically coupled to the rest of the circuit, and whose impedance
1
According to the compensation theorem, we can replace Z^s)
x (s).
by a voltage source V(s), as shown in Fig. 7.13. Then by the superposition
principle, we can think of the current /(*) as the sum of two separate parts
is
is
Let the current It(s) be the current due to the n voltage and
<-
current
fl
VM
\2M
FIG.
1
7. IX
Thevenin's equivalent
Zi(e)
circuit.
Sons,
See H. H. Skilling, Electrical Engineering Circuits, 2nd Ed. John Wiley and
New York,
1965.
182
Network
analysis
and synthesis
Therefore I^s)
i.e.,
is
N
it
voltage sources
current sources
FIG.
7.
7aC0
is
is
we
see
passive so that
by the
relation
13
=-
V(s)
(7.19)
Zm(s)
the input impedance of the circuit at the terminals of the
where Zm(s)
is
source V(s).
= Ms) -
7.18 as
(7.20)
Zm(s)
all cases,
= Foc(s)
Z ln(s)
and
V(s)
(7.21)
Transform methods
in
network
analysis
183
FIG. 7.14b
so that
we can
Zla(s)I(s) -
KocCs)
V(s)
(7.22)
In order to obtain the current I(s) through Zx (s), the rest of the network
can be replaced by an equivalent source Ve(s)
Voc(s) in series,
with an equivalent impedance Ze (s)
Zm(s), as shown in Fig. 7.12.
Example 7.4.
LH0-)
'c<-)
FIG. 7.15
Z,(s)
Z,(s)
Next we find
Ve(s)
=R+sL
all
voltage sources.
(7.23)
184
Network
analysis
and synthesis
-rnnp-
>i
o2
FIG. 7.16
-/(*)*+ Z/(0-)-
M0-)
(7.24)
VJto
~ Z.{s) + Z(s) =
7(j)Jt
+ Lt(0
-) -Pq(0-)/s
(7.25)
A+sL+ 1/sC
FIG. 7.17
7.5. For the network in Fig. 7.18, let us determine the voltage vjt)
0, and we
across the resistor by Thevenin's theorem. The switch closes at t
0.
assume that all initial conditions are zero at t
Fvampk
Lj
Li
=c
Qv(t)
itftf
FIG. 7.IS
us redraw the
circuit in
VM
LiS
+ \jsC
(7.26)
Transform methods
Lis
La
in
network
analysis
185
* Vo
FIG. 7.19
left
of AT is
slM/sC)
(7.27)
V&)
We know that
Therefore
V&)
"
ZJis)
RVisYMsC)
(R + sL^sLi + lIsC)
Finally
o(0
(7.28)
+R
+ LJC
(7.29)
(7-30)
e-HKtCi)]
Norton's theorem
When
tance
is
required to find the voltage across an element whose admitYifs), the rest of the network can be represented as an equivalent
it is
From
Fig. 7.20,
Vx{s) =
The element whose admittance
is
/.(*)
(7.31)
Yt(s) + YM
Yt must not be magnetically coupled to
91
()>
Yds)
FIG. 7.20
Yito
Network
186
anal/sis
and synthesis
-nppp>
Pi
.1
c
IM(
<P
'
FIG. 7.22
FIG. 7.21
capacitor, as
shown
From
in Fig. 7.22.
The admittance
is
(7.32)
sL+R
'
Then
')
sL+R
V(s)
Example 7.6.
that v(t)
Yt(s) + Yds)
(sL
+ R)sC +
(7.34)
1
" O.le-6
current
Ie(s)
is
Us) =
0.1
V(s)
Rt +
J?!
circuit as
+sL
(s
5)
S)(s
+ sRxC +
R +sL
s*LC
- sC +
Us + RILXs +
sL
o-iTJoTP-wv
c=J=t
FIG. 7.23
0.5s*
(7.35)
10)
and 2
+ 5s +
* + 10
is
10
(7.36)
Transform methods
in
network
analysis
187
FIG. 7.24
I^s)
is
IM
then
+ GJ
(7.37)
+ 6)
(7.38)
/<(')
BJLYJto
1
(S
sy(s
+ 6) - (s + 5)
(s + 5)V +6)
(*
/(*)
as
(s
+ 5)*
(*
+ 6X* +
(7.39)
5)
//*)
(*+5)
j+5 + j+6
finally
(7.40)
obtain
(7.41)
As we
e(t) is related to
the response
tit)
by a
is
When
the Laplace transform is used in describing the system, the relation between
the excitation E(s) and the response R(s) is an algebraic one. In particular,
when we
related
We
how
= E(s)H(s)
relation
(7.42)
and how
this function
188
Network
analysis
and synthesis
T
1
'R
'w(D
a?-r
FIG. 7.25
Impedance
a current source and the response is a voltage,
excitation and
then the system function is an impedance. When both
we have a
terminals,
pair
of
response are measured between the same
is
impedance
driving-point
impedance. An example of a
When
the excitation
is
driving-point
m = m = R+
HKS)
I,(s)
I?22L
(7.43)
sL+l/sC
Admittance
a voltage source and the response is a current,
admittance IJV, is
H(s) is an admittance. In Fig. 7.26, the transfer
obtained from the network as
When
the excitation
is
H(s)=
m=
l_
(7.44)
innnp
*t
wvfit
1(
Si-
J2W
^=r
FIG. 7.26
Transform methods
in
ZiM
)yt
Z*8)
network
anal/sis
189
T
1
FIG. 7.27
current
Va(s)
I{s)sa
Z {s)+Zt(s)
VJs) = Z&)I(s)
(7.45)
Since
V (s)
Vg(s)
then
(7.46)
ZJs)
ZM + Z
(7.47)
t(s)
When
I(s)
= IM + /(s),
(7.48, 7.49)
sC
we find
(7.50)
I(s)
1/sC
R + sL+1/sC
h(s)
7'
W Q)
is
4=ic
FIG. 7.28
(7.51)
Network
190
anal/sis
and synthesis
FIG. 7.29
From the preceding examples, we have seen that the system function is a
the
function of the elements of the network alone, and is obtained from
let
us
network by a straightforward application of Kirchhoff's laws. Now
system
the
obtain the response transform R(s), given the excitation and
the
function. Consider the network in Fig. 7.29, where the excitation is
that
assume
We
current source i(t) and the response is the voltage v(t).
0. Let us find
the network is initially inert when the switch is closed at t
i(t)
= (sin a>ot)u(i).
2.
i(0
is
3.
First,
we
waveform
in Fig. 7.31.
H(S)
sC
1.
it (t)
- (sin ou
1/sL
f) u(t).
+G
C[s
(7.52)
s(G/C)
1/LC]
is
(7.53)
so that
V(s)
<Oo
= /(s)H(s) =
s*
a>
C[s*
s(GIC)
(7.54)
1/LC]
ig(t)
ig(t)
A
a
<i
FIG. 7.30
FIG. 7.31
Transform methods
2.
in Fig. 7.30,
i(0
transform
Its
is
in
network
anal/sis
can be written as
ig {t)
= u(t) - u(t - a)
= i (i _
j,(s)
191
(7.55)
e-)
(7.56)
The response
V(s)
+
C[s*
p-0*
-
s(G/C)
(7
57);
l
1/LC]
Note that in obtaining the inverse transform C-1 [F(j)], the factor e-*
must be regarded as only a delay factor in the time domain. Suppose we
rewrite V(s) in Eq. 7.57 as
K(s)-^(l-0
(7.58)
Then
if
we denote
we
r{^]
*<0
(7.59)
is
Vl (t
- a) u{t - a)
(7.60)
excitation.
The waveform
3.
tf
and
V(s)
its
is
transform
then V(s)
we denote by
v,(t)
It (s)
(l
as
If
^^
=
+
+ 1 _ Cl)
+
(0
- (0
u(t
is
be represented as
a)
(7.61)
1 1
s \
as
as
3
as J C[s
(7.62)
/
s(GIQ
(7
K
63)
excitation,
we
1/LC]
ramp
'
see that
C-1 [F(s)]
= Vl (t) + - Vi(t) - - v (t a
where vt(t)
is
a) u(t
a)
(7.64)
192
Network
Let us
now
response R(s)
discuss
some
= H(s) E(s).
R(s):
R(s)
, v
!s s
+
By
A<
Si
(7.65)
t
rW^ZA^ + ZB^"*
(7.66)
The terms Af?** are associated with the system H(s) and are called free
response terms. The terms B^'* are due to the excitation and are known
as forced response terms. The frequencies st are the natural frequencies of
the system and j, arc the forced frequencies. It is seen from our discussion
of system stability in Chapter 5 that the natural frequencies of a passive
network have real parts which are zero or negative. In other words, if we
jo> { , then <x< <, 0.
<rt
denote st as st
Example
For the
7.7.
va(t) =
Since V,{s)
K (*)
1/sC
VJis)
R +MsC
V,(s)
is
the response
is
2\s*
(7.67)
2
(7.68)
1/
then
W- VM =
0.4
(st
+ Ws + 2)
J+2 +
JK>
-2Q
-WW
C-Jf=i= uoW
FIG. 7.32
0.4*
*
+ 0.2
+l
(7.69)
(7.70)
Transform methods
It is
0.4 cos
As a
in
network
analysis
193
is
is
+0.2 sin/.
variable.
Consider
our discussion,
first
let
7.33c.
VJjw)
_ jaRC
jmRC +
_
R+
V,(J(o)
1/jcoC
(7.71)
1
coRC
(7.72)
We have, approximately,
*W =! jcoRC
Then
the response
(7.73)
V^ijco),
we
obtain
nMsiRC-vJtl)
(7.75)
at
Note that the derivation of Eq. 7.75 depends upon the assumption that the
R-C time constant is much less than unity. This is a necessary condition.
Next, for the
function
R-C
is
V/J*)
VJijm)
l/ja)C
1/jfoC
+R
jwCR
(7.76)
1
hgW
VO<t)
(a)
(b)
R-C differentiator.
(6)
R-C integrator.
Network
94
If
we assume
analysis
that
and synthesis
o>RC )J>
1,
then
VM^T^zVjUa)
jcoRC
Under
so that the
7.4
(7.77)
is
(t)~--r Va(r)dT
(7.78)
RC Jo
7.336
is
approximately an integrating
circuit.
In this section we will show that the impulse response h(t) and the system
R(s)
We
(7.79)
thus see that the impulse response h(t) and the system function H(s)
is,
= H(S)
Z-^[H(s)] = h(t)
Z[h(t)]
(7.80)
Example
7.8.
/(/) in
the
R-C
~
Simplifying H(s) further,
we have
^^O-JTw)
The impulse response
Ht)
which
is
shown
is
(7 ' 81)
(7 - 82>
then
C-i[/f(5)]
in Fig. 7.35.
= i ^r) - -L e-wj
(,)
(7.83)
Transform methods
network
in
analysis
195
i(t)
)9(t)
/Z-t/RC
_
R*C
FIG. 7.34
response of
Fig. 7.34.
Since the step response is the integral of the impulse response, we can
use the integral property of Laplace transforms to obtain the step response
as
W-rf?]
where
oc(f)
Similarly,
(7.84)
we
ramp
^-m
(7.85)
where y(t) denotes the ramp response. From this discussion, it is clear
that a knowledge of the system function provides sufficient information to
obtain
all
the system.
Example
7.36.
7.9.
Let us find the current step response of the R-L circuit in Fig.
is the response and V (s) is the excitation, the system function
g
Since I(s)
is
V{s)
Therefore H(s)/s
is
h{s)
s
The
(7.86)
R+sL
n_ i \
=i
s(R+sL) R\s
s + RjLJ
now
(7.87)
obtained as
1
a(/)=-(l -e-W*X)ii(/)
To
check
(7.88)
R-L
circuit,
Network
196
which we found
analysis
in
and synthesis
Chapter
5.
1
(7.89)
Kt)
The
step response
is
a(0
</t=-(i -*-w*>0(i)
(7.90)
-J>
It is readily
seen that
R
-VSMAK)
V,M{
FIG. 7.37
FIG. 7.36
system
7.10. In Fig. 7.37, the only information we possess about the
in the black box is: (1) it is an initially inert linear system; (2) when v&) = 6(t),
Example
then
With this information, let us determine what the excitation vjf) must be in order
to produce a response vf)
- te~" u(f).
First,
to be
H(s)
V^s)
Vjs)
7T2
2*4-5
t j+3
VAs)
Simplifying
2(a
(7.92)
+3)
(7.93)
+2)*
JW "
H(s)
K<W =
The system
+ 2X
<
The unknown
(j
(s
+ 2Ks + 3)
2.5X* + 2)
(7.94)
20 +
1
(s + 3)
+ 2.5X + 2) " s + 2
we have
0.5
*
+ 2.5
(7.95)
excitation is then
vtf)-{e-*t -0.5e-*
M)u(t)
(7.96)
Transform methods
7.5
in
network
analysis
197
x (f) is
<
Ft(s),
and
if
convolution of/i(f)
WTO,
that
and/t(r)
is
is,
t[ j*Mt
where the
is
is
integral
ft(t
(7.97)
t)/,(t) dr
and
is
denoted operationally
as
fm -
Proof.
C|/i(0V^0]
From
t)/^t) dr
dt
(7.99)
- t) =
=
we have the
Then Eq.
- t)/,(t) dr =
r>t
(7.100)
f 7i(
" t) (* - t)A(t) dr
(7.101)
Jo
C[/i(0Vi(0]
(7.102)
Jo
Jo
let
rt
identity
f/i('
Jo
we
(798)
u(t
If
- fflVMt)
= r so that
t
'
-.t
e-^f,>
(7 103)
Network
198
analysis
and synthesis
f 7i()
Jo
u(x)e-" dx f 7(r)e- w dr
Jo
= F (s)F2(s)
(7.104)
The separation of the double integral in Eq. 7.104 into a product of two
integrals is based upon a property of integrals known as the separability
property. 2
-2
Example 7.11. Let us evaluate the convolution of the functions/i(f) = e ' u(t)
and fift) = t u(t), and then compare the result with the inverse transform of
F^s) FJs), where
(7.105)
W
The convolution
variable
of_/i(f)
t for
and/2(f)
p
obtained by
is
first
substituting the
dummy
in/i(/), so that
fiff
T)
c-* (
'-T)
u(t
(7.106)
t)
Then/iWVaOis
f *fi(f
T)/*( T)
rfT
Integrating
let
^T = ~a*
T 2r
dr
(7.107)
Jo
by parts, we obtain
fi(0* fiff)
Next
_2
r " _T)
I
Jo
Jo
({-\+\ *-*) W
(7-108)
From Eq.
7.105,
we
have
so that
An
equation
.
t
flit
New
York, 1940.
Jo
Treatise
is
71 10)
expressed by the
- r)/a(r) dr =
Jo
Sons,
"(')
MMA -
r)
dr
(7.111)
Transform methods
This
is
readily seen
network
in
= *i(')
CWOViC)] - Wi()
CC/i(0V*]
and
199
analysis
(7-112)
(7113)
we
JA(t)
Mr)
= t (t)
shown
as
- ,
and
in Figs. 7.38a
7.39a.
m(t)
(7.114)
.,
A('-t)/2(t)V
Jo
= ( t).
product
A(|
is
- u(t - t)
t)
/x (t)
.
t)/
(t)
The
(7.115)
by a variable amount
t.
_ t)t u(t)
<f
function
Next, the
(?
n6)
AV.-/(0 =
we must
t
(</).
Since the
considered a variable in
(7.117)
!J(0
In Fig. 7.39,
same
we
result as in
c- 1 [F1 (s)fW]
-1
l3j
m(o
(7118)
analysis.
R(s)
we
= E(s) H(s)
(7.119)
where e(r)
is
"*[() H(s)]
the excitation
and
('e(r) h(t
- t) dr
(7.120)
200
Network
analysis
and synthesis
Mr)
Mr)
1
/
/
(a)
(a)
\M-t)
M-r)
1
T
(b)
M-t)
Mt-r)
1
1
(e)
hM M-t}
M-r) Mr)
id)
ffh
M
FIG. 7.39
FIG. 7.38
Using Eq. 7.120, we obtain the response of a system directly in the time
domain. The only information we need about the system is its impulse
response.
EVro-pW. 7.12a.
Let us find the response /(f) of the R-L network in Fig. 7.40 due
to the excitation
c(t)
- 2<r* y(f)
is
.AW-|*-raBW
Therefore, for the
R-L
circuit
(7.121)
(7.122)
under discussion
Kt) *-"(*)
(7-123)
Transform methods
in
network
V~
t)
Kr) dr
i(i)
201
analysis
as
- 2j V"-rtr* </r
(7.124)
VvWvV-
0*"
FIG. 7.41
PIG. 7.40
Kt)-Kd(t)
(7.125)
Let us show by means of the convolution integral that the response r{t)
to the excitation e(t) by the equation
K0-JT4)
Using the convolution
K0
integral,
is
related
(7.126)
we have
(t)*('-t)/t-a:
r)dr-Ke(f)
(7.127)
r-
a variable in the expression for r(t), we see that the ideal amplifier in
is an impulse-scanning device which scans the input (/) from
domain
the time
/
to f oo. Thus, the response of an ideal amplifier is a replica of the
input e(f) multiplied by ibcgain IT of the amplifier.
Since
7.
/ is
In the Section 7.5, we discussed the role that the impulse response plays
in determining the response of a system to an arbitrary excitation. In
this section we will study the Duhamel superposition integral, which also
describes
an input-output
The superposition
integral requires the step response a(f) to characterize the system behavior.
plan to derive the superposition integral in two different ways.
We
Network
202
The
simplest
is
analysis
and s/nthesis
examined
We
first.
relationship
R(s)
= E(s) H(s)
(7.128)
E(s)
(7.129)
C-1
[^- -s(s)]
(7.130)
yields
K0 =
a(0*[e'(0
+ *(0-)d(0]
= e(0-) a(f) +
where
and
e'(t) is
<x.(t)
is
the derivative of
e'(r) a(f
e(0 )
7.13.
Duhamel superposition
- t) dr
(7.131)
is
Equation 7.131
referred to as the
Example
e(t),
'
J
/
is
0,
usually
integral.
i(t) in
the
R-C circuit
in Fig. 7.42
when
is
(7.132)
as
shown
in Fig. 7.43,
H(s)
Us)
V(s)
R(s
M)
s
is
(7.133)
IjRC)
is
R(s
(7.134)
l/RC)
vg (t)
*WI
Slope
a A2
zizc
FIG. 7.42
FIG. 7.43
Transform methods
in
network
analysis
203
~-tlRC
<t)
(7.135)
) =0 and
= A 1 d(f) + A t u{t)
v'^i)
The response
we
7.132,
(0
(7.136)
then
i(0
-jTj'V,(r)a(t -r)dr
A x Hr)(t
<5(t) o(/ -
t) dr
r)dr
At <x(f -
r)dr
Jo-
Jo-
e-Ur-r)IRC fc
ic
/{
(7.137)
Jo_
)] u(t)
As we
see
transient problems.
staircase approxi-
mation of e(0 as
e(t)
+ An u(t - /iAt)
'11
Signal
}a3
s
/
-A2
hABi
<P+>
0h-AT-|
2At
_L
3At
4At
signal.
(7.138)
Network
204
analysis
and synthesis
the system to be linear and time invariant, we know that if the response to a
t). Therer) is X< a(f
unit step is <x(f), the response to a step K+ u(t
fore we can write the response to the step approximation in Fig. 7.44 as
K0 =
If
At is
K0 -
e(0+) a(0
small, r(t)
+ Aa(f - nAf)
(7.139)
can be given as
e(0+) o(0
At
AE
* oc(*
- At) At
(7.140)
+ *h (, - 2At)At +
+ ^aO -
At)At
At
At
AT
n
r(t)
(0+) (0
lim 2
+ Ar-0<-+
~
tV
AT
'
At) At
1-CO
(7.141)
v(t)
.- 2(f)
and /(0-)
-2
amps.
Find and
100
r
VWfe|
vwQ
1000
5h
PROB.7.1
after having been at
7.2 The switch is thrown from position 1 to 2 at t
t
1e~' sin fit.
for a long time. The source voltage is vjif)
(a)
(b)
Find the
initial
vtf).
plane
Sketch one possible set of locations for the critical frequencies in the *
transform.)
and write tint form of the response v&). (Do not take the inverse
(c)
Transform methods
in
network
anal/sis
205
MtOB. 7.2
13
/(f)
for
In the circuit shown, all initial currents and voltages are zero.
using Thevenin's theorem.
t >
Find
51(0
PROS. 7.3
7.4 In the circuit shown in (a), the excitation is the voltage source e{t)
described in (6). Determine the response i(t) assuming zero initial conditions.
e(t)
10
*t)
'S*
4=i*
1/2
-1
M
PROS. 7.4
13
conditions.
Network
206
anal/sis
and synthesis
PROB. 7.5
7.6
The
opened.
0.5 sin
circuit
Vlt
n(0.
The excitation
initial
resistor
is i(t)
energy. At t =
the switch S
that the response is v(t)
X such
te~ **
is
u(t).
i
v(t)
=r*
PROB. 7.6
7.7 Use transform methods to determine the expressions for i^t) and /j(/)
-10 '.
Assume zero initial
lOOe
in the circuit shown. The excitation is v(f)
energy.
1000
K=\
PROB. 7.7
circuit shown, the switch S is opened at / = 0. Use Thevenin's
to determine the output voltage t^f). Assume zero initial
theorem
or Norton's
7.8
energy.
For the
Transform methods
in
network
207
analysis
PROB. 7.8
7.9
6u(0>
For the transformer shown, find i^t) and ,(/). It is given that e(t) =
and that prior to the switching action all initial energy was zero; also
M=1A.
PROB. 7.9
7.10
For the circuit shown, find /,(/), given that the circuit had been in steady
10Q
WWW
0.
Kml
100
WW
400
vl=
2hojlo8h
j=z=_40v
PROB. 7.10
7.11
cos 20
Find
u(t).
/,(/)
Assume zero
initial
energy.
The
excitation is e(t)
100
Network
208
TVTP-
y/W
2h
10O
300
7.50
|6h
PROB. 7.11
7.12
(/),
find v t{i).
Assume zero
initial
= V^IV^s). When
v^t)
conditions.
10
r-nAAA-
if
=j=
T
v
*
G) 5V
=r
3f
fOS v3
PROB. 7.12
7.13
find v(t)
when
/(/)
(b) the
convolution integral,
mo
m
lf=t
io:
ih<
PROB. 7.13
7.14
excitation were
<*/) * 3e"*'(0,
PROB. 7.14
Transform methods
in
network
analysts
209
7.15
were
c(/)
7.16
= 3S'(t),
(a)
(0
(*)
sinr
f(f
t)x(t)/t
x(t).
a<T)e-<-f >/T
J
(/)
7.17
+j
x(jt
r) e-*
dr
=* t
K
()
(s
+ aXs + b)
2fr+2)
(s*
+ 4)
(s*
F(s)
<<0
7.18
"
By
1)
shown
in the
T
PROB.7.18
7.19
Using
and
(b) the
Duhamel
superposition
PROB.7.19
210
7.20
Network
Using
integral, find
analysis
(a) the
<0 for
and synthesis
e(t)
PRO B. 7JO
(b)
C-0.5 ju
-r
-2
-1
to
PROS. 7.21
7.22 The excitation of a linear system is x(r), shown in (a). The system
impulse response is Hf), shown in (ft). Sketch the system response to x(t).
neat, carefully dimensioned sketch will
(No equations need be written.
suffice.)
Transform methods
in
network
anal/sis
21
(a)
h(t)
(b)
PROS.
7.23
is /(/)
7.22
A unit step of voltage is applied to the network and the resulting current
(a)
this network.
(b)
admittance function.
PROS. 7.23
PROB.
7.24
c(r)
(a)
(b)
=6e-'
12v.
chapter
8.1
In this section we will study the relationship between the poles and
zeros of a system function and its steady-state sinusoidal response. In
other words, we will investigate the effect of pole and zero positions upon
the behavior of H(s) along the jw axis. The steady-state response of a
system function
is
where
=M {my*
(8.1)
an
or magnitude response function, and
odd
and
is
an
^(o>) represents the phase response,
is
and
8.1 A.
The
is
equal to 0.707
amplitude \H(j(Omu)\. In terms of decibels, the halfpower point is that frequency at which 20 log \H(j(o c)\ is down 3 db from
20 log |#(/a)max)|. The system described by the amplitude and phase
characteristics in Fig. 8.1 shows that the system will not "pass" frequencies
of the
maximum
212
213
Amplitude
response
0707
-b
fc
(a)
Phase response
FIG.
response
8.1.
then
<f>{w) is linear,
filter.
<f>(a>)
is
We see
approxi-
mately linear over the range (o c <,w<, +co c If all the significant
harmonic terms are less than coc then the system will produce
minimum phase distortion. With this
example, we see the importance of an
.
-VWVAr
we
concentrate
will
on methods to
ViW
=r
V2 (s)
ically.
To
curves,
obtain
we
let
amplitude
s
=ju>
and
phase
FIG.
84
in the system
and express H(ja>) in polar form. For example, for the amplitude
and phase response of the voltage ratio VJVX of the R-C network shown
function,
H(s)
is
V^s)
Letting
ljRC
F^s)
l/RC
(8.2)
is
l/RC
jm
l/RC
(8.3)
214
Network
analysis
and synthesis
M(w)
FIG.
8.3.
HO) = (to
l/RC
2
1/K
--/tan-i
C2)*
a>BC_
=
M(ft>)e'*
(0,)
(8.4)
The amplitude and phase curves are plotted in Fig. 8.3. At the point
(o = 0, the amplitude is unity and the phase is zero degrees. As a> increases,
amplitude
is
is
Ms - Zq)(s -
=
(s
)(s
Pi)(
Zi)
- Ps)
HUco)
*,)0
*P
Zq)0'<>gp
0>
Pi)(J> - P*)
Po)0>
jco z or jco p corresponds
(8.5)
ApO'cu
(g6)
to a vector
Each one of the factors
t
}
from the zero zi or pole/*, directed to any point jco on the imaginary axis.
Therefore,
if
we
jco
- z, =
rV",
jco- p,=
M,e'*'
(8.7)
i(y><>+v>i-a-i-ti)
(g 8 )
jj(jca)
as
shown
in Fig. 8.4,
j4
NNl
MoMiMj
0! is negative.
FIG.
8.4.
In general,
215
following equation.
n
H
M(.) = ^-
is
on
thejco axis
given as
^(eu)
=J
2
It is
F(s)
=5*
4s
4s
+ 2s + 2
(s
+ H-jl)(s+l-jl)
(8.9)
Let us find the amplitude and phase for FQ2). From the poles and zeros
2, as shown in Fig. 8.5. From
of F(s), we draw vectors to the point to
216
Network
anal/sis
and synthesis
m
J2
V5/
/\45*
x
>
Vio7
>
-l
/\71.8*
x
FIG.
8.5.
-n
it is
MQ2)
and
= 4(
,_
,- )
= 90 - 45 - 71.8 =
<f>(j2)
clear that
<f>Q2)
1.78
-26.8
we have enough
at three or
JU
X
\M5'
yi
XX90*
-s.90*
1.0
'
/S
1A5
-yi
(a)
FIG.
8.6.
Br.
-j
(6)
217
=
=
lim F(ja>)
4Q0)
(8.10)
(l+Jlxl-Jl)
o>>0
this equation, we see that the zero at the origin still contributes a
90 phase shift even though the vector magnitude is zero. From Fig.
45
45
90.
90
is #0)
8.6a we see that the net phase at <w
From
approximately equal to
coJfii,0,
where
Extending
y>
all
1,
Then
~ 4eoh _
ca k
#a>)
toh
M(w)
and
mh
<
~ 90 - 90 - 90 =
-90
Table
8.1,
we
obtain
F(s)
= s* +
s* +
1.03
1.23
_ (s + jl.015)(s - jlMS)
~ (s+ ;1.109Xs - jl.109)
(8.11)
Amplitude
10
+90
0123456789
-90
Frequency,
in
Eq.
8.9.
10
218
Network
TABLE
Frequency,
8.1
Amplitude
CO
Phase, degrees
1.0
1.8
25.8
1.5
2.0
-5.3
-50.0
-66.0
-78.5
3.0
1.3
5.0
0.8
10.0
0.4
whose pole-zero diagram is shown in Fig. 8.8. At co <= 1.015, the vector
from zero to that frequency is of zero magnitude. Therefore at a zero
on the /o> axis, the amplitude response is zero. At co = 1.109 the vector
from the pole to that frequency is of zero magnitude. The amplitude
response is therefore infinite at a pole as seen from Eq. 8.11. Next,
consider
<
the
JU
co
y'1.109
<J./1.015
>
1.015,
1.015
it is
and o><
at
co
is
When
zero.
1.015
is
now
in the
same
direction as for co
-y'1.109
<
1.015.
axis, the
1.015
FIG.
8.9.
and
8.126.
219
1.109
8.8.
while the zeros in (b) are the mirror images of the zeros in
(a),
]1,
and are
5.
<E
NX"
FIG.
the/co axis.
the/o> axis.
8.1 1. Effect
220
Network
analysis
and synthesis
JW
J2
>2
yi
-n
X
-y'2
-J2
(a)
(a)
<r
o-;i -
FIG. 8.12.
-2
-l
fl
-2
(b)
(b)
Nonminimum phase
function.
phase of (b) is greater than, the phase of (a) for all frequencies. This is
because the zeros in the right-half plane contribute more phase shift
(on an absolute magnitude basis) than their counterparts in the left-half
system
plane. From this reasoning, we have the following definitions.
function with zeros in the left-half plane, or on the jco axis only, is called a
minimum phase function. If the function has one or more zeros in the
in Figs. 8.12a
it is
and
8.126.
Let us next consider the pole-zero diagram in Fig. 8.14. Observe that
the zeros in the right-half plane are mirror images of the poles in the
100
y- Minim urn phase
l-ioo
1
^^/" Nonminimum phase
YY)
-360
10
12
co x
JO>
on
>2
<!.
221
Kml
~
It is
-2 -i
x-yi
A system function whose poles are only in the left-half plane and whose
zeros are mirror images of the poles about the jm axis
function.
is
called
an
all-pass
BODE PLOTS
Ms)
(8.12)
D(s)
1.0
10
*-
+ 180
l -I*
-360
10
12
8.14.
Network
222
analysis
and synthesis
Ka
>l
3
S
togu-
K2
l*KI
Magnitude
(a)
K positive
Phase
K negative
e-
IT
logw(b)
FIG.
8. 16.
amplitude response
M()
is
|HC/)I
- j^}
(8.13)
\D(J>)\
If
we
20 log M(w)
= 20 log
\N(jco)\
- 20 log
D(jco)\
(8.14)
(a)
a constant,
(b)
(c)
(d) a
we have
+ 2<w +
a*
/?*
K = Kt
is
(8.15)
223
The constant
The factor
(b)
The
s.
20 log
to.
20
db/decade or
is
constant for
For convenience,
a.
2Qlog
shown
shown
+^=
The phase
in Fig. 8.18a.
Arg
as
is
Bode
at to
1,
we
is
all co.
as
a pole
in decibels
O+
let
us set a
2Qlog(o)S
1.
Then the
, )M
(g
lfi)
is
l)*1
tan~x
a>
(8.17)
in Fig. 8.186.
20 log
For
to
1,
\j(o
a<1 s* 20 log
201og|7o
0.1
1|
02
03
0.5
1.0
Frequency,
FIG.
8. 17.
db
(8.18)
is
20 log
l|
0.7
m db
2.0
3.0
(8.19)
5.0
= 0,
7.0
10.0
224
Network
anal/sis
and synthesis
12
(+!)->.
2
8"
025
0.5
(ii
1
fc-
(b)
TABLE
a> =-
Approximation,
Error,
db
db
db
2 octaves below
octave below
break frequency
octave above
2 octaves above
=i
o) = 1
m =2
o)
o-4
8.2
Actual
Magnitude,
Frequency
225
Straight-Line
0.3
0.3
1
3
12.3
12
1
0.3
maximum
error
is
co
1,
or in un-
a.
normalized form: m
For quick estimates of magnitude response, the straight-line approximation is an invaluable visual aid. An important example of the use of
these straight-line approximations is in the design of linear control
systems.
roots. For complex conjugate roots, it is consymbols so that we can use the universal curves
adopt
standard
venient to
(d)
Complex conjugate
of a magnitude
as
shown
a>
and an angle
in Fig. 8.19.
Explicitly, the
real axis,
and
= cos 0,
If
factor.
the
is
a and
ft
- - P
are related to
(8-20)
juo
following:
-J"o
g>
cos
to
P = <u
sin
o),
Vw*
and co by the
(8.21)
-/woVW 2
damping factor,
VI
*
and
<ot .
Network
226
is
is
to
7r/2,
Bode
the
The magnitude
for convenience.
20 log
|1
is
a)
the smaller
To examine
>
and synthesis
analysis
+]2t,a>\
<f>(w)
then
is
= 20 log [(1 =
a> 8) 8
+ 4 <u 8]*
8
(8.22)
2gn>
-i
tan"
(8.23)
l-a>8
We
for
Universal curves for magnitude and phase are plotted in Figs. 8.21 and
8.22 for the frequency normalized function
G(*)
(sK)8
+ 2(s/a) ) +
(8.24)
1
We see that the phase response, as viewed from a semilog scale, is an odd
function about
eo/a>
The phase at co
1
<w
is
90 or n\2 radians.
1
+ 10
+5
^x^^-r-o.6y
-b
~~
Asymptote
-10
-15
0.1
0.5
1.0
Frequency,
^\
vft.
A.
0.2
^\
r-io-^N.
o>
2.0
5.0
10.0
qp'IfflOl-
227
228
Network
anal/sis
and synthesis
woaiv
8
CM
229
For a conjugate pair of zeros, we need only reverse the signs on the scales
of the magnitude and phase curves.
KVraMpig 8.1. Using Bode plot asymptotes, let us construct the magnitude
versus frequency curve for
0.1s
GW
(8.25)
'(Ht
+
16X10*
10
We see
there are
-12db/oetave
8.3
SINGLE-TUNED CIRCUITS
then
= {LC)-
l/i
.
W=
H(S)
An
example of a single-tuned
whose
!o(i>
Vs)
circuit is the
R-L-C
1/LC
1/sC
R+
sL
ljsC
s*
+ (K/L)s +
(8.26)
1/LC
230
Network
anal/sis
and synthesis
FIG.
231
FIG. 8.26
are
'
(8.27)
2\LC
2L
J} J
< (4/LC).
In terms of a and
/J
in Eq. 8.27,
//(5) is
ff(s)
+^
(s + a+j/f)(s + a-7/8)
'
(8.28)
From the pole-zero diagram of H(s) shown in Fig. 8.26, we will determine
the amplitude response \H(Jm)\. Let us denote the vectors from the poles
to they'w axis as |MX and |M,| as seen in Fig. 8.26.
can then write
We
|HO)l
where
K = a* +
j8*
(8.29)
|MJ
|M,|
and
iMii
|M,|
- [ + ( + m*
- [a + (a, - W\"
aspects.
(8.30)
eu
a>msx, at
which
is
Network
232
analysis
|H0>)|
and synthesis
+ F?
+ (a, + /?)*][* +(,-j8)8
(' +
8
4
> + 2tt>\%* - f) + (a +
('
[a*
(8.31)
/?*)'
/?*)
dJH(U>)\*
__
<fo>
eoSuz
(8.32)
/**)
8
jS
(8.33)
(8.34)
damping factor
and
*
)l
f
dm
we determine
a>*
<8*)]
* |H
with respect to
+ /8*)W + 2(q' + (* + /W
+ 2a "
(a'
*
\
and the
wj^ is
- (o) Vl
-*)*-
(C^o)
= <(\ ~ 2*)
(8-35)
Since comax must always be real, the condition for a>max to exist, i.e., the
condition for \H(jw)\ to possess a maximum, is given by the equation
21* <,
(8.36)
<
45.
cos 0, o)max does not exist for
so that t <, 0.707. Since
In
this
When 8 45, we have the limiting case for which max exists.
the
same
poles
have
0.707 and the real and imaginary parts of the
case,
magnitude,
i.e.,
/?,
or
>.
>
Vl
(8-37)
comax
is
imaginary;
it
> <Vl - P
<8 - 38 >
as seen in Fig. 8.27. Moreover, the point at which the circle interfrom the triangle
sects the positive jm axis is comax- This is readily seen
to
exist,
233
JU
max
>
<r
H-->
i
-"max
-70 x
with sides a,
/S,
By
circle.
we find
that
(oJLx
Fig. 8.27
is
- j8* - a*
called the peaking circle.
When a
>
/J,
(8.39)
o>
= 0,
When
of the
a>
=w
/?,
When a
is
A figure
(8.40)
2 cos
high-g systems, as given in Fig. 8.28c, and poles far removed from the jo)
axis represent low-Q circuits (Fig. 8.28a). Although the Q of the circuit
given by the pole-zero plot of Fig. 8.286 is theoretically defined, it has no
practical significance because the circuit does not possess a maximum
point in
its
amplitude.
By means of
point,
which
\H(jw c)\
is
- 0.707 |#(./aw)|.
co c
we can
is
234
Network
and synthesis
analysis
ju
jco
JP
J0
Umax
j- *Jmm
a\
max
-jftk-
(a)
(6)
(c)
circles,
a, eomax ~ p.
(a)
= P, ma
0.
(6)
>
fi,
= fa
Area (piPi*aid
In terms of the vectors
be expressed as
|MJ and
(8.41)
\MX
Area (&PiPi*<ot) =-
co it
|M,| sin y
>
(8.42)
where
y> is
the angle at
it
we
Ju
From Eqs.
8.41
|M 2
is
equal
to
|MX
|M,|
= -^
sin
\H(ja>i)\
where
(8.43)
y>
is
(8.44)
K is a constant, then
.Ksin
FIG. 129
and 8.42,
|
|H(M)I
y>
(8.45)
20*
235
For a given pole pair {puPi*} the parameters /J, a, and ^Tare prespecified.
we have derived \H(Jm)\ in terms of a single variable parameter,
Therefore,
the angle
y>.
|H(/>max)|
When v
y>
\H(ja>J\
= coc
so that a>t
o) a
Let us
.
= Wmax, and,
(8.46)
= 0.707 and
= 0.707 I^C/o^nu)!
Let us consider
first
1, to,
circle as
shown
in Fig. 8.30.
obtain
We will
y>x
= w/4 because it
When
when
(Umax
= 0.
For a high-g
highly peaked at
\HQ0)\
m=
where comw
<*>< the amplitude is
shown in Fig. 8.316. In this case, if
are two half-power points m Ci and co
0i
circuit,
<Umx, as
Peaking
'circle
FIG.
8.30.
Network
236
anal/sis
and synthesis
0.707
Iff,
(a)
FIG.
8.31. (a)
(b)
Low-G
circuit response,
(b)
High-Q
circuit response.
(8.47)
As a
result, the
>Oi
is
= a>mx
(8.48)
0>Ct
The bandwidth of the system for a high-g circuit of this type is described by
BW = m Ct
(o
(8.49)
0i
we
is 0, as seen
of obtaining the phase response. The phase shift at eo
it rad, as shown in part (c)
oo, the phase shift is
from Fig. 8.32a. At <a
a> =i >max, the phase
of the figure. Finally, in the neighborhood of a
plane
is approximately
lower
half
in
the
shift resulting from the pole
is controlled
region
in
this
in
phase
change
The
(Fig.
8.32&).
tt/2
0,
in the
response
the
phase
seen
that
readily
It
is
the
pole
in large by
px
region of a> has the greatest negative slope, as seen from a typical phase
1953.
237
J<*
Hi
b)0
k
(*)
(b)
(e)
circuit,
() o><
a.
Finally, as
circuits, let
#(s)
34
=
s*
Now we
determine the
maximum and
|/f(/*comax)|
(8.50)
+ 6s + 34
and
half-power points
awx
and
<o c ,
34
H{s)
(8.51)
+ 3-;5)
and the poles of H(s) are shown in Fig. 8.34. We next draw the peaking
circle with the center at j = 3 and the radius equal to 5. At the point
where the circle intersects thejfa> axis, we see that eomx 4. To check this
a* gives
result, the equation <*, =
cBmx = (5* - 3*)* = 4
(8.52)
(s
3+;5)(s
/J*
-180*-
circuit.
Network
238
analysis
and synthesis
The amplitude
6.78
= <or
|H(/4)|
|f(/a>max)| is
then
34
(3+j9)(3-jl)
==
1.133
(8.53)
30
The
point
at
at s
= 2.0.
draw a
circle
AB
at A,
we
(equal to
this
to c
~ 6.78
(8.54)
circle construc-
is
_.
Then
co c is
given as
co c
Finally,
we
sl(AC)1
|H(j'6.782)|
n
8.4
is
V46
6.782
(8.56)
(8.57)
which
- (AO)* =
34
/
V(34
46)*
(6V46)
.
S
0.802
DOUBLE-TUNED CIRCUITS
In Section
poles.
8.3,
circuit.
pairs
We will consider
V2 (s)
two
239
the special case when the R, L, and C elements in the primary circuit are
equal in value to their counterparts in the secondary. Since the primary
and secondary inductances are equal, the mutual inductance is
M=KL
(8.58)
V^s)
=^R +
L+
/.(s)
J^j
- sM /2(s)
(8.59)
= -sM h(s) +
{r
sL
+ -M
/,(*)
we readily determine
Vito
K*
l
'
set
=-
2<o
(8.61)
^RM'L
(1
K*)(s*
\
If
we
*M
A=
set
I?(l
then
we can
^
K*)
= ^2*
1 - K*
(8.63)
K
'
write
H(s)
=
(s
where
{ Sl , Sl *}
11
SiXs
sfXs
s)(s
(8.64)
s,*)
(1
-_]
K)*J
(8.65)
-k
Li
-x
ci
jc)*J
Network
240
analysis
these assumptions,
and synthesis
we can approximate
(8.66)
Similarly, {s *,*}
can be given as
{s s,*}
The
~ -o> jo) ^l +
J
poles
<u
is
(8.67)
is greatly enlarged in
circuit.
clarity purposes.
origin.
^4|Mo|'
241
In terms
(8.68)
]H(J(0}1
is
a>
= a>
we have
(8.69)
<u
Aco
|HO)|S'
4IMJIM.I
It is evident that
(8.70)
IMJ and |M,|. The double-tuned problem has thus been reduced to a singletuned problem in the neighborhood of
Consequently, we can use all the
on the peaking circle that were
derived in Section 8.3. Let us draw a
o)
results
peaking
circle
- +/>,
(8.71)
y> then determines the location of the maxima and half-power points
of the response. Without going into the derivation, the amplitude response
can be expressed as a function of y according to the equation
angle
|HO)l
A<o sin
4a)
sin
K(ta>
2(1
let
(8.72)
K*)
situations:
> WiKfl:
a>
iW =
*-=
2(1
-K*)
be undercoupled.
(8.73)
Network
242
anal/sis
and synthesis
jo>
8lM
2
\
i
-fo
T
7
WO
y
A
ly\ w
x^^
(a)
FIG. 8.38.
2. o>,
single point
(b)
(a)
m=
(Fig. 8.38ft).
At
lmax
(c)
co^Kjl:
(c)
Overcoupling.
circle intersects
the amplitude
the/eo axis at a
equal to #max
in Eq. 8.73. In this case
Kj2, and we have critical coupling.
3. <u
cogA/2 The peaking circle intersects theytu axis at two points.
a>
a>
is
<
o> x
and
to,,
The
by the
equation
<*>!,*
max
tt>
>
(8.74)
.[(!H*
(
Note that
have their
maximum
points at
me
"a mix
FIG. 8.39.
(a)
Undercoupled
coupled case,
(c)
Overcoupled case.
uCl
uCt
<<
critical
coupling,
half-power points
Fig. 8.40.
circuit.
243
BW =
The
Example &2.
a>
Ct
m Ci
(8.75)
a double-tuned
circuit is
given as
As?
=
(*
+2
-HylOOX*
+2
+2
-ylOOX*
+yl06X>
+2
-/106)
(8/76)
From H(s), let us determine the following: (a) the maximum points <ot ma and
<o t ; (b) the 3 db bandwidth BW; (c) the damping factor ; (d) the coefficient
of coupling*; (e) the gain constant A; and (/) the maximum of the amplitude
response //max.
Solution, (a) The natural frequency of oscillation o> is taken to be approximately halfway between the two poles, that is, <o = 103 radians. In the neighborhood of , we draw the poles s 2 +/100 and s = 2 +/106, as
shown in Fig. 8.41 . From the peaking circle centered at the point s = 2 + /<.,
shown
in Fig. 8.41,
we
obtain
mt max
so that
(b)
mo =
<i
max
*o
m*r
Next we draw a
2* = 2.236 radians
+ 2.236 =
2.236 =
circle centered at
<o
V(3^2)
(8.77)
105.236 radians
(8.78)
100.764 radians
we have
">c,
"^3*
Ct so that
1
- 4.123 radians
(8.79)
Network
244
and synthesis
anal/sis
The
db bandwidth
is
The
coefficient
which
oj)
{ is obtained
8.2.
103
of coupling
8.246 radians
t
4
(d)
Example
then
BW - 20Ol -
circle,
circle for
u0.0194
'" 15^
(8.80)
a>
2,
(8.81)
is
o>oA:
(8.82)
K = = 0.0582
We thus have
(e)
is
equal to
2{.A:
*
(f) Finally, the
- K*
maximum
(8.83)
2(2X0.0582)
1
- (0.0582)*
amplitude
flmiz
Hmmx
= ""**
(8.84)
is
=
2(1
-**)
-0.5009
(8.85)
ON
83
POLES
245
TIME DELAY
What is
will
is
8.86,
T),
(8-87)
Let us examine the amplitude and phase response of the pure delay.
From the equation
*-***
< 8 - 88)
HO) -
we
We
response, that
<f>(a))
T is
(8.89)
coT
(8.90)
is,
T= -
d<f>(w)
(8.91)
dco
A()
*te)
(a)
(h)
(e)
(a)
Amplitude,
Network
246
anal/sis
and synthesis
For pulse
an
tion.
n (-)
H(s)=*tf
(8.92)
IT(-Pi)
= a
with zeros at z
at/>,
<r
real frequencies is
lr
<K<)
? tan-1 W
=
=2
.
ot
*-i
Envelope delay
J-l
1 to
.
tan"
+ CO,
x
'
(8.93)
a,
h o' + (c (of + A o* + (
(894)
co,)*
see that the shapes of the delay versus frequency characteristic are
all
The zeros
However,
Now
let
pole atp
do not have
Ls is the only
is #a>)
ir/2; thus the delay is zero.
us consider the delay due to one singularity, for example, a
a +jcoQ The delay due to the one pole is
.IL
-J
is
dm
We
fflj
*
pertinent:
to this pole
is
Am = and occurs at
about
co
<w
to
.
co
The delay
(8.96)
is
symmetric
247
FIG.
2.
8.43.
The frequency
at
o>x
3.
The
"effective delay
simply 2a
graphically
The
co
bandwidth"
is
is
half the
maximum, or
l/2or
is
(8.97)
ff
then
eo
<r
<m<
to
cr
or
intersections of the circle with the jco axis are the half-bandwidth
points, as
4.
2.
Thus,
if
we wish
The delay of an
is
always
alone.
We
is
shown
in Fig. 8.44.
rough
and then
differentiate numerically.
Finally,
when
it
the phase response goes through the origin. If it does not, there
is
is
hard to
Network
248
wo -
4<?o
">o
analysis
and synthesis
3o wo - 2<ro wo ^o
FIG.
o +
8.44. Plot
ffo
o +
2<ro
wo +
3<ro
wo + 4o
<*
Problems
8.1
and
Find the poles and zeros of the impedances of the following networks
on a scaled s plane.
plot
nnnr*
|f-
4h
io
Z(s>-
la)
o
M/*
1(-
20
iM
ZM-
m
20
4h
ZW-
PROB. 8.1
&2 The
circuit
shown
in the figure
is
a shunt peaking
video amplifiers,
(a)
Y(s) is of the
K(s
form
- s^s s>
(s
-*,)
circuit often
used in
249
=c
Y(,>
:r
PROS.
Find the amplitude and phase response for the following functions and
83
sketch.
() F(s)
(*) F(s)
+K
+K
s
(c) F(s)
Note
8.4
id) F(s)
that
7T v
quantities.
<**">
C() +71*.)
Show that
85 By means of the vector method, sketch the amplitude and phase response
for
s-l
+0.5
j
(a) F(s)
sis
(C) F(*)
10)
s
(b)
F(s)
's*
s
(e)
F(s)
'
(/)
+2s +2
Plot
F(5)
s*
+1
-1
^)F(j) - (,/2xV+ 9)
8.6
on semilog paper
F(*)
the
F(5)
-2s +
5*
F(,)
- 2j + 5
-(TTW+T)
=
s*
(j
+2s +5
2X*
Bode
plots of magnitude
100(1
+ 0.5*)
s(s
F(s)
<*>
50(1
(1
+2)
+ 0.025jX1 + O-IJ)
+ 0.05*X1 + 0.01s)
1)
250
8.7
Network
Plot
analysis
on semilog paper
()
W
8.8
and synthesis
F(*)
the
Bode
plots of magnitude
10005
5 10~ 5*
(1
+ O.OOZsXl +
(1
200(1 + 0.05s)
+ 0.02jX1 + 4 10-** + 10-V)
+ 10~V)
*W-3 + 2s + 5
s*
a>
and
\F(ja>
)\.
Sketch the
8.9 For the circuit shown, determine the current ratio IJIg and find: (a)
the point <>,, where its amplitude is maximum; (b) the half-power point <oc ;
1. Use geometric construction.
(c) the point to where |/i(a) M)//B(<o u)|
2.5 h
PROS.
8.10
ltt
Af\to)
8.9
is
maximum
at
o>
Show
m* =
/<*
r^*^' -9
'
= at
|cos 20|.
PROB.
8.10
8.11 In connection with Prob. 8.2 plot the poles and zeros of the impedance
Find, approximately, the maximum point of the
l/l%s).
function Z(s)
amplitude response. In addition, find the bandwidth at the half-power points
and the circuit Q.
251
8.12 The pole configuration for a system function H(s) is given in the figure.
From the plot, calculate:
(a)
(6)
a>
yioo
-5
<T
-y'100
PROB.
8.12
20
oWW
lh
^SW"
YM-
=rif
PROB.
8.14
fri v\
(s
+ 4 +y50X* + 4 -y50X* + 4
+j6Q)(.s
+4
-y60)
252
Network
analysis
and synthesis
and
circle to determine the maximum and half-power points
coupling K.
the circuit Q. Find the gain constant A and the coefficient of
and halfa x i mu
8.17 For the double-tuned circuit shown, determine the
of
power points and the circuit Q. Find the gain constant A and the coefficient
coupling K.
10
"9
M=5xl<r 6 h
PROB.
8.18
a =
8.17
0, 1,
and 2 for
()
F(s)
(b)
F(s)
(c)
F(s)
(<*)
F(s)
(e)
s+2
= s-3
s
+3
3*
=
(s
s*
s + 1
+ Is + 5
IX*
+ 2)
(s+2Ks*+2s+ 2)
chapter
Network
analysis
9
II
NETWORK FUNCTIONS
9.1
electric
port
may be
Zi
= = sa
(9.1)
Whether the one-port is actually a single 5-Q resistor, two 2.5-Q resistors
in series, or two 10-Q resistors in parallel, is of little importance because
the primary concern
is
Yto =
2s
V=
^ = 2s + 3
3 and
(9.2)
parallel with
a $-Q
resistor in
=!=2f
One-port
network
V
I'
FIG.
PIG. 9.2
9.1
253
io
254
Network
Two-port
network
Vi
v2
2'
r
FIG. 9.3
Two-port parameters
A general two-port network, shown in Fig. 9.3, has two pairs of voltagecurrent relationships. The variables are Vx , V2 , Ju lz Two of these are
dependent variables ; the other two are independent variables. The number
of possible combinations generated by four variables taken two at a time
describing a two-port
is six. Thus there are six possible sets of equations
.
network.
The z parameters
Vx =
z^ + zj
(9 3)
V2 =
21 l
wJ%
z ll
J,=0
7,=0
(9.4)
v,
21
'M
/!=0
7,-0
Vi
zb
FIG. 9.4
Vi
Network
anal/sis
255
II
parameters. Note that z u relates the current and voltage in the 1-1' port
only; whereas z M gives the current-voltage relationship for the 2-2' port.
impedances.
As an example,
in Fig. 9.4.
let
T circuit
+z
= zb + z
7,=0
z ti
7,-0
(9.5)
*1*
=Z
=Z
7,-0
_^2
*l
7=0
Observe that z ia
z 21
When the open-circuit transfer impedances of a
two-port network are equal, the network is reciprocal. It will be shown
later that most passive time-invariant networks are reciprocal. 1
Most two-port networks, whether passive or active, can be characterized
by a set of open-circuited parameters. Usually, the network is sufficiently
complicated so that we cannot obtain the z parameters by inspection, as
.
did for the T circuit in Fig. 9.4. The question is now, "How do we
obtain the z parameters for any circuit in general?" The procedure is as
we
We write
Vx
and
Vt
h = nn V + n^Vi H
- n.1^1 +
+
1
- "*iFi +
1
is
r*
+ nu
lk V
rk
+ nSk V
rt
+ ns
Sk V
+ "**n
(9.6)
Network
256
analysis
and synthesis
If the circuit is
n it
=n
ft
equations,
As a
AM
= Gu +
made up of R-L-C
is,
sCu
ith
(9.7)
sL,il
it is
clear that
must be equal to
they'ith cofactor,
ti ,
that
is,
A w = AH
= z^,
as
we
shall see.
Returning to the
and Vt We obtain
set
9.6, let
us solve for
Vx
Vl
~ A 1+ A
(9.8)
In relating this
parameters,
it is
last set
clear that
u=
ii
An
A
ztt
= A8a
(9.9)
A^ = A w
it
follows that za
= z12
the
Ia
CD
=-Yc V
(9.10)
+ (YB + Yc)Vt
Vi
v2
YB
FIG.
9S
d>
Network
The determinant
anal/sis
II
257
is
Ay = YA YB + YA YC + YB YC
(9.11)
+
^~ YBAy
Yc
Zai
~Ay
(9.12)
*
Now
Zu
Ay
YA + YC
Ay
We readily obtain
Zl
-z
- z + z - ^
*11
- za + z -
-s-
r,
-T
YB + Yc
z
Zo
We then find
(9.13)
Ay
Ay
-&
Ay
(9.14)
Zc
Ay
The y parameters
Suppose we were to write a
Fig. 9.3.
set
- uA + iA +
V% mA + "hJ* H
= msi/i+
Vi
= m uJ +
l
"
+ i
r-
m^/j
+ W.A
(9.15)
+ m kkTk
where
the
it
represents the
258
that for
Network
anal/sis
and synthesis
m mH
an R-L-C network,
it
for all
and / Thus
reciprocity
holds.
we
equations.
h~ A
(9.16)
h = VuVi + yit v*
h = y i^i + ynVt
(9
17 )
'.
=A+
(-
we
A=
A
\s
(9.18)
obtain
2(2s
-A
+ ~h
s
In straightforward fashion
l)h
1)
ls%
+ 4s +
(9.19)
-A - - ls + 2s +
*
FIG. 9.6
Network
The
analysis
259
II
+ 4s + 1
2(2s + 1)
8
2s + 25+1
yi = Vi* 2(2s + 1)
When ylx = yn or zu = za2 the network is symmetrical. 2
yu
sfa
2s
(9.20)
911
we
_h
Vy Fi-0
/,
3/m
v* Fi-0
(9.21)
h
K
h
Vn
Vm
Ft-0
v* Fi-0
The reason
parameters
Fig. 9.5.
that the
is
To
now
obtain y a and y M ,
we
We then
have
Vu
yi
yn, *2i
(9.22)
= -Yc
,Jz
m yn Z3.
h>i
(b)
(a)
FIG. 9.7
*A
and
symmetrical network
is
by interchanging the
1-1'
Network
260
analysis
and synthesis
1-1' to obtain
= YB + Yc
^12 = -Yc
^22
,g
23)
The h parameters
set
Vi
I
The
= *uA + hi*V*
"tl-M. "22'
by the
relationships
-h
Fj=0
h
"21
7l-0
1*
=~
h
"89
see that h lt
and h 91 are
Jl-0
seen that h u
is
(9.25)
=h
F,-0
We
(9.24)
"I"
It is easily
hu
(9.26)
yu
The parameter h 22
is
to z M by
ha
is
related
(9.27)
Both the remaining h parameters are transfer functions; h 21 is a shortand h 12 is an open-circuit voltage ratio. Their
relationships to the z and y parameters is discussed later in this chapter.
For the Pi circuit in Fig. 9.5, the h parameters are
fcll
=
Ya
h 12
h ai
+ YC
Yc
YA + YC
= Yc
YA + YC
YA YC
v
YA +YC
fc_
(9.28)
Network
analysis
II
261
h
+
NIC
FIG.
9.8.
let
= -Z,
(9.29)
= *
(9.30)
NIC.
Vx - kVt
(9.31)
A = kh
If
we
conditions.
h vt
(9.32)
21
j&
h tl
the
NIC is nonreciprocal.
NIC is
ii
(9.33)
i
.k
The NIC
is
is
circuits.
For a
Chapter
Matrices,
4.
It
Practical
Network
262
realizations of
and synthesis
anal/sis
transistors.
Some of
1,
LA.
(9.34)
ara
known as the transmission matrix because it relates the voltage and current
at the input port to their corresponding quantities at the output. The
reason the current It carries a negative sign
is
that
most transmission
engineers like to regard their output current as coming out of the output
A=
B= - Yi
h
lt=0
Fa=0
(9.35)
C=
r=o
From
these relations
we
see that
represents
an open-circuit voltage
a short-circuit transfer impedance; C is an opencircuit transfer admittance; and D is a short-circuit current ratio. Note
that all four parameters are transfer functions so that the term transmission matrix is a very appropriate one. Let us describe the short-circuit
transfer functions B and D in terms of y parameters, and the open-circuit
transfer functions A and C in terms of z parameters. Using straightforward algebraic operations, we obtain
transfer function; J?
is
A=^
B = _J_
J/21
i-
D = _2u
z 21
Vtl
(9.36)
c=
For the
ABCD
is
expressed by the
equation
YA
det
B~]
C D
= AD-BC=l
Trans.
(9.37)
IRE on
Circuit Theory,
Network
263
II
ABCD
former in Fig.
analysis
whose denning
9.9,
Vi
equations are
Vl - nVt
/1
<*SD
= i(_ j8)
FIG.
we have
n
(9.39)
B"
is
~n
^s
C D
(9.40)
I
L
Note, incidentally, that the ideal transformer does not possess an impedance or admittance matrix because the self- and mutual inductances
are infinite. 5
For the
impedance shown in
(9.41)
nZ,
Taking the
ratio
of
Zl
Thus we
see that
an
==
h
n*ZL
ideal transformer is
L (Fig.
to be
(9.42)
an impedance transformer. If
1 we would see
9.106), at port
As a second example
*
For a
Introduction ta
Modem
Network
Synthesis,
M.
E.
Van Valkenburg,
New
York, 1960.
Network
264
h
m:li
v2
()
c
n?i,
c=
network
ABCD
Fig. 9.5.
Ys
+Yc
Yc
B=
Yc
YA YB + YB YC + YA YC
C=
Yc
Y
+Y
A
C
(9.43)
If
we check for
reciprocity
AD-BC = (YA +
from Eq.
YC)(YB
we
9.43,
+ Yc) - (YA YB + YB YC + YA YC
a
Y
In
-&*!
9.2
see that
(9.44)
when port
1 is
open
circuited:
= 0.
*\ = z
V = z22/2
/x
i2-'a
(9.45)
Network
Therefore
we have
= -y1
h lt
analysis
II
= 2H
265
(9.46)
V.
Similarly, since h
is
we can
h a ==
**
(9.47)
parameters alone.
what the
by
circuit.
The
and y
relationships
easily
by using matrix
[Z]
(9.48)
L*si
yu yn\
(9.49)
J/ti
In simplified notation
we can write
=
[/] =
[V]
and
Replacing
[7]
in Eq. 9.50
Vt]
[Z][7]
(9.50)
[Y][V]
(9.51)
by [Y][V], we obtain
[V]
[Z][Y)IV]
(9.52)
so that the product [Z][Y] must yield the unit matrix [U]. The matrices
[Y] and [Z] must therefore be inverses of each other, that is,
1
[ZTr
From
z
the relationship,
and
[Y]
we can
[IT-^IZ]
(9.53)
and y parameters.
Zjg
Zji
A*
A.
(9.54)
A.
Network
266
TABLE
9.1
[*]
z 12
z 22
*22
z ia
[A]
[71
s/
3/12
A12
AT
a,
A,
Am
Agg
3/a
3/u
Agi
>
A,
A*
Agg
Ay
3/n
J/12
Au
Au
/<
3/22
An
An
A*
2/21
An
J?
J?
3/12
B
An
AT
A12
Ah.
Agg
D
C
D
ft
w.
A,
A,
M
A,
ii
A,
z 12
A,
Z 22
*2
3/ii
3/u
2/21
A,
[A]
Z 22
3/ll
ii
A.
3/22
An
21
Z21
3/21
3/21
An
Agi
z22
3/u
AM
3/21
A21
Agi
*m
where A
Vii
3/21
= zuz g ZigZax
g
and
ii
= T~
Zia
J/22
=
(9.55)
where
Av =
^n2/ 22
Vi^/n-
2/12
Zg!=
Using these
?21
identities
we can
derive the h
tMe
9.3
b. T
= AD- BC
(9.56)
In this section
we
Network
analysis
II
267
means of z or y parameters
Vt =
alone.
z ai/j
(9.57)
Vx - z^
If we take the ratio of V, to
Vx we obtain
,
Yl
Vt
By
letting
%L
(9.58)
*ii
we
derive
(9.59)
V\
In similar manner
two-port as
we can
Via
(9.60)
Is
and
_ _M
(9.61)
The open- and short-circuit transfer functions are not those we usually
deal with in practice, since there are frequently source and load impedances to account for. The second category of two-port transfer
functions are those including source or load impedances. These transfer
functions are functions of the two-port parameters z, h, or y and the
source and/or load impedance. For example, let us derive the transfer
*?
+
Vi
Two-port
network
FIG. 9.11
V2
268
Network
analysis
and synthesis
"~l
l-
Zll
^-^
+1*12*2
1+
s~~-
v2
Vi
l'o-
2'
FIG.
By
Two-port equivalent.
9.12.
Vt we obtain
,
yJR
V1 yu + 1/R
the same. YS1 is
rsl
Note that
resistor
(9.63)
R, and y n
We must be
is
the transfer
make
admittance when port 2 is short circuited.
nature.
similar
this distinction in other cases of a
In order to solve for transfer functions of two-ports terminated at
either port by an impedance ZL , it is convenient to use the equivalent
circuit of the two-port network given in terms of its z parameters (Fig.
parameters (Fig. 9.13). The equivalent voltage sources z lg/a
9.12) or
careful to
y
and zIx in Fig. 9.12 are called controlled sources because they depend
6
upon a current or voltage somewhere in the network. Similarly, the
current sources y lt Vt and y^Yx are controlled sources. For the circuit
in Fig. 9.12, let us find the transfer impedance Zm = VJIU with port 2
"1
a.
1
1
y2iVil
Y2
y22
yi
'
1
I
FIG.
v2
9.13.
2'
Two-port equivalent.
New York,
1964.
Network
terminated in a load impedance
the /* mesh,
ZL
If
write the
269
II
mesh equation
for
we have
-tA-(*ii
Since
we
anal/sis
+ Zx )li
(9.64)
z -
r = ^rr
two-port network
< 9 - 65 >
is
-*
h
In similar fashion,
we
"*
+ zz,
(9.66)
Z22
Via.
(9.67)
^8 + yM
Next, suppose
we
shown
VJV
for
in Fig. 9.14.
Next,
we
/t
+ 2ll)(^2 + m) - *1**21
Vt = RtI%, we may now arrive
(9.69)
(1
From
the equation
at the following
solution.
Y*=_Ml =
V,
?t
{R1
z^)(R t
+ z,,) - znZlt
*u
r^aro >*
FIG. 9.14
(9.70)
Network
270
WW^O1-^h
MAA"
Z22-Z12
211-Z12
(Z21-12Ml
v2
Z12
Vl
circuit
circuits
9.13 are not unique. Two other examples are given in Figs. 9.15 and 9.16.
Observe that the controlled sources are nonzero in these equivalent
circuits
only
the circuit
if
nonreciprocal.
is
shown in
Fig. 9.17.
Observe the voltage-controlled source A gl F2 at port 1 and the currentcontrolled source Ag^ at port 2. Let us find the input impedance Z ln
.
The
Vi
+ hi*V*
h lxh
(9.71)
and
(9.72)
Vt we find
,
V =
htiZiJi
1
t^
(9.73)
h&ZL
we have
hiah z &
( i.
\
(9.74)
h^LjJ
(9.75)
so that
fc
MZ
h
yi2
o-ii-
Vi
-WW.yu + yn
-<
,73
y22
+ yi2<
Q)
Vi
circuit
Network
anal/sis
271
II
circuit.
We can easily check to see that Eq. 9.75 is dimensionally correct since Au
has the dimensions of impedance, h 22 is an admittance, and h 12 h tl are
,
9.4
INTERCONNECTION OF TWO-PORTS
In this section
we
We will see that when a pair of two-ports are cascaded, the overall transmission matrix is equal to the product of the individual transmission
matrices of the two-ports. When two two-ports are connected in series,
their z matrices add ; when they are connected in parallel, their y matrices
add.
ports
Na and N
in cascade or in tandem, as
shown
in Fig. 9.18.
We see
that
ra-ra
12a
+
Vi
for
tt
a is
ara
(9.77)
hb
+'
Vsa
Vn
(9.76)
*i
Nb
v2
-
Network
272
h r"
h'
9-*-
DC
Vi
L.
<
Za
V2
V2
'
*b
Vi'
Gyrator__
FIG. 9.19. Gyrator in tandem with
Correspondingly, for N
T circuit.
we have
(9.78)
first,
we
obtain
Ma-eara
We
<network
is
As an example,
let
is
related to
Zm =
a*YL
21
(9.80)
Zi
The constant a
in Eq. 9.80
is
its
Vx =
If
we
a(-/)
h - -a vt
(9.81)
Network
so that the transmission matrix of the gyrator
"0
analysis
in
II
is
.a
AD-BC<=-1
(9.82)
is
za
C D
+z
V* + V. + V.
i
La
(9.83)
V* + Vc + V.
+ *
ah
az h
we check
If
we
AD-BC=-\
We
(9.84)
situation in
is
nonreciprocal.
shown
in Fig. 9.20.
Na
and
ferrites.
274
Network
anal/sis
and synthesis
lc
FIG. 9.21
V
V
L/J
and
From
UJ
Fig. 9.20,
we
J/llo
Vila'
-Vila
2/*2a-
~2/ll6
2/l2&
-2/216
2/226-
>1"
(9.85)
>"
(9.86)
UJ
must hold.
~ V\a = V\h
h = Ao + Aft
V\
V*
^2o
'26
(9.87)
A = Ao + I
we must be
careful
is
does not interfere with the internal affairs of the other, ideal transformers
are used to provide the necessary isolation. In matrix notation, the sum
of the individual [/J matrices of the two-ports in parallel must equal the
[/] matrix of the overall two-port network. Thus we have
(9.88)
Vila
2/
2/21a
Vtia
J/116
2/126
2/226-
y**
Network
*
II
275
/&,
/la
anal/sis
JV
e>
V2a
Ideal
Vi
Vz
In
lib
C
Nb
Vu
Vu,
'-'
If
we connect
|yii
y1
Ly
yJ
two-ports in
+
lytia +
[vna
series,
as
yita
Vtu
Vin
(9.89)
ytu>-
shown
+
y*ta +
Vub
pil
^t]
LZi
2tJ
+ 2m
z Mo + z*i J
Z
*11&
_ pUa + U
Zflb
Lzaia + Z
Zl*.
"l
(9.90)
222&-1
21ft
first,
is
As a
final
in Fig. 9.6.
example,
let
first
= y*u, = i
= y*n = s
Viiii
ym>
a is
T circuit so
(9.91)
by
inspection.
276
Network
analysis
and synthesis
These are
zllo
2Mo S +
(9.92)
Z]
ZlSn
*12o
*Mo
2s
(9.93)
yi2a = yio = t
2 12a
now
obtained as
3/ii
tflt
+ Viu
= s(s + 1)
2s + 1
Vila
2s*
+ 4s + 1
+ 1)
2(2s
(9.94)
Vl2a
Vltb
sf_
2s
9.5
2s
2s
2(2s
l)
2s*
INCIDENTAL DISSIPATION
As we have
dissipationless;
i.e.,
Network
analysis
II
277
somehow.
An
"load
down"
in a
manner such
that
Lt
(9.95)
Ct
if
<=
R +
4
sL{
(9.96)
m'a
= JR, + sL +
t
= R + L&i +
{
r,
a)
(9.97)
C (s +
4
a)
Si
c,
L-vw-J
HM
Hi(s)
(a)
FIG. 9.24.
* H(s + a)
(b)
(a) Original
network.
(A)
Loaded network.
Network
278
Jf
-It-
=*=**
2Q
1Q
<2h
FIG. 9.25
Similarly,
on node
basis, if the
original (unloaded)
network
n if
is
=G +
sCt
(9.98)
=G +
sCt
=G +
Q(s
g(
is
(9.99)
t
+ a) +
+ a)
L,(s
H(s
H(s
+
+
it is
) after the
is
/^(s)
a).
Vu
By
s
find the
y parameters for
we have
7s
12
= 1 + + - - 1 + 7
4
2s
Via
first
inspection
(9.100)
4s
= i=-4
= *C1 = $mho
(9.101)
Ct =
is
f,
*Ca = imho
(9.102)
Network
anal/sis
II
279
80
-vw
L-\H
if
J
20-
6Q<
< 1Q
=*f
10
1 2h 1
3
= J.
L=
r,
h,
we have a
olL
resistor
=1Q
(9.103)
>
= i + Ks + i) +
- 1 + A(* + t)
y'
K*
i)
=i+
+1 4
+
D* + 2
=i+
4(s + |)
2a-
(a
'-- (H~*H)
(9.104)
We
see that the y parameters of the loaded network could have been
obtained from the y parameters of the unloaded network by the relation-
ship
t (s)
H(s
a).
realizability in
Chapter
10.
280
Network
analysis
and synthesis
z,
>
'i
*2
*
I
^*4
*J*i
currents
all
ladder.
voltages
and
apply.
vi+1
il+i
Zi+s +
vi+>
- yJA v^s)
VJL) - '*(*) U') + v&) = [1 + Z^*) Yt (s)] v&)
h(s) - Y2 (s) VJLs) + Us)
= {Y&)[1 + ZJs) Us)] + Y,{s)} VAs)
V (s) = I (s)Z (s)+Va(s)
= {zjlyai + z,yj + rj + (i+ z,n)} v&)
/(*)
(9.106)
Upon examining the set of equations given in Eq. 9.106, we see that
each equation depends upon the two previous equations only. The first
equation, Vt
V% is, of course, unnecessary. But, as we shall see later,
it is helpful as a starting point. In writing these equations, we begin at
+
Vi
Zi
\-^H
z3
Y2
Y*
T
v2
j.
FIG. 9.28
Network
analysis
II
281
and work towards the 1-1' port. Each succeeding equation takes into account one new immittance. We see,
further, that with the exception of the first two equations, each subsequent
is obtained by multiplying the equation just preceding it by the
immittance that is next down the line, and then adding, to this product
the equation twice preceding it. For example, we see that /,( is obtained
by multiplying the preceding equation V&) by the admittance Yt(s).
The next immittance is ZJis). We obtain KB(i) by multiplying the previous
equation IJs) by ZJs) to obtain /gZjj then we add to this product the
equation twice preceding it, V^s), to obtain Va
It Z,
Vt The process
equation
(as
to the other),
and adding to
this
product the
Using
of equations,
this set
we
P*C0 by the
last
equation
as transfer immittances
if
we
let
VJs) term is
K2(j) =
V-fe).
and current
ratios in similar
Vt(s) as a factor.
1.
first
in which the
equation
R=
first
1 2
of the branch
is
z(s) =
-^L
= _i+R
+
llsC
C=
is
Similarly, if a shunt
an inductor L^
(9 . 107'
v
)
= 2Q and
y(s)
7T~
2 + s
( 9 - 108 >
282
Network
Example 9.1. Let us find the voltage ratio VJVlt the current ratio Ijlu the
input impedance Zx = VJIlt and the transfer impedance Zn = VJI-i for the
network in Fig. 9.29. First, we must represent the series branches as impedances
and the shunt branches as admittances, as shown in Fig. 9.30. The branch
current and node voltage equations for the network are
h*-
-/fflnp-*
3s
3h
Vi
2i
Vi
V2
2f=t=
*h
FIG. 9.30
FIG. 9.29
V^s)
Us)
= 2s
Va(s) =
V,
3s(2s)
-(6s*
s
1)
6s*
(9.109)
It(s)
VM
+-
14s
14g+
+ l=6s* + 57+-l
6s2
\)
+ 2s =
S*
(a)
6^+57^+8
3
14s
s*
Yi
<*)
+2s
6s
57s8
8
(9.110)
2s*
(c)
li
** ~
id)
Example
network
9.2.
Find the
14s*
/x
+2
"
14s2
y n and
j/2 i
for the
is
given as
is
Network
rVW*-i
V.
analysis
II
283
I.
2a
Vi
2f
lh =j=2f
FIG. 9.31
The
Ja
=1
Va =2
= K2) + 1 = * +
(9.111)
2$*
l/5\
5/2
12
7J + 5 +
h
yn
-,'
la
h
~^~~5
(9.112)
2
's.aa
Vi
2
+l
FIG. 9.32
number of other
contributions
Bashkow,10 O'Meara,11
To name
10
1963), 286.
284
Network
analysis
>
and synthesis
computer is an
functions.
Problems
9.1
Find the
z,
y, h,
and
circuit configurations.
-OQ
PS
Y
'
PROB.
9.2
Find the
(e)
(b)
(a)
z,
y, h,
lo
and
9.1
iA/W
\AAA/
<
i
o2
Network
9.4
For the
lattice
and bridge
Ya =
analysis
and
Yb
MZb
285
1/Z
II
y parameters
in
93 For the circuit shown, find the voltage-ratio transfer function VJVX and
the input impedance x = V^lx in terms of the 2 parameters of the two-port
network
and the load resistor RL .
Z
1
I2
+
Vi
Rl<
v2
_
2'
1'
PROB. 9S
Network
286
anal/sis
and synthesis
9.6 For the cascade connection of two-ports depicted in the figure, show that
the transfer impedance Zj 8 of the overall circuit is given in terms of the a parameters of the individual two-ports by the equation
Vu.b
Vita
Na
Vi
Nb
PROB.
9.7
figure.
of
given by
VnJti
= -
is
Lx
v2
PROB.
9.6
9.7
eh
*ii
gl*
<f22j
9.9
AD - BC =
-VA
common
Wv-^O
-o
mh
J2
V2
Vi
PROB. 9.10
Network
9.11
anal/sis
II
287
input circuit
and voltages
in (a).
R,
9.12
of the
Find the
T parameters
figure.
9.13 Find the y parameters of the twin-T circuit in Prob. 9.2c by considering
the circuit to be made up of two T circuits in parallel.
9.14
circuits
shown.
Network
288
1
anal/sis
and synthesis
>
-o
V2
Vi
V
1
2*
Ideal
transformer
(a)
h<
V2
Vi
Ideal
<b)
transformer
2'
PROB.
9.14
PROB.
9.15
Ideal
transformer
PROB. 9.16
Network
analysis
II
289
9.15
Find the
9.16
9.17
For the circuit in Prob. 9.26 determine the y parameters of the uniformly
shown.
loaded circuit derived from the original circuit with the dissipation a =0.1.
Plot the poles and zeros of both cases.
9.18 Find the transfer impedance VJIX for the circuit in part (a) and the
voltage ratio VJV1 for the circuit in part (b). Plot the poles and zeros for the
transfer functions obtained.
o>- /Tnnn^-r-^voTP
}h
in
:4f
Vi
10
V2
(a)
WW-
Vi
2h
lf=:
if;*:
:ia
v2
PROB.
(6)
9.19
method
Find the
network
9.18
utilizing the
in Section 9.6.
10
HWv
20<
20
_lh
Hwvi-'Tnnp-^
=t=if
if
PROB.
9.1*
30
|VvV-|
Yf-rifippJfL-
20-
if
Vi
if==
1Q<
V2
4f=r
PROB. 9.20
9.20 Determine the voltage ratio VJVlt the current ratio IJIlt the transfer
impedance VJIlt and the driving point impedance Fi//a for the network shown.
chapter 10
Elements
<|>f
realizability
theory
10.1
CAUSALITY
AND
STABILITY
The
(10.1)
Our task is
The first
to synthesize a network
As an example,
is
causal, whereas
<
0, that is,
<0
(10.2)
e- (0
(10.3a)
h(t)
= e- "
290
(10.36)
Elements of
realizability
theory
291
W-T)
(a)
FIG.
not causal.
is
(b)
(b) Realizable
10.1. (a)
impulse response.
realizable (causal)
we
not realizable.
is
If
we
T)
is
realizable (Fig.
10.16).
is
that
dco
<
|log \H(ja>)\
+o>a
oo
(10.4)
r.
<
oo
(10.5)
fall off
For example, the ideal low-pass filter in Fig. 10.2 is not realizable
because beyond m c the amplitude is zero. The Gaussian shaped curve
order.
|HO)l
shown
in Fig. 10.3
is
(10.6)
e~
o*
(10.7)
Am. Math.
Network
292
analysis
and synthesis
\H(ja>)\
\H(ju)\
FIG.
FIG.
f"
J-*
is
not
finite.
10.3.
Gaussian
filter
character-
istic.
acteristic
>
(10.8)
dea
eof
(10.9)
\H(jo>)\
vrr
co-
h{t)
=A
sin <o c t
(10.10)
nt
A is a constant. From the sin xfx curve in Fig. 3J4, we see that
nonzero for t less than zero. In fact, in order to make h(t) causal,
it must be delayed by an infinite amount. In practice, however, if we
delay h(i) by a large but finite amount t d such that for t < the magnitude
of h(t t^ is less than a very small quantity e, that is,
where
h(t) is
\Kt
Q\
<
*<0
wwv
1Q
Vi(s)
lr.^z
FIG,. 10.4
Elements of
< 0.
h(t
(For a more
realizability
an
excellent treatment
where
stable,
Cx and
C, are
then from
<
<
the response
0<.t<co
C*
IKOI
by Wallman.*)
e(t)
0^r<oo
KOKCi
KOI
is
then
293
theory
we
CiJ"|*(t)I dr
If a linear system
is
obtain
<
C,
(10.11)
dr
<
(10.12)
oo
/;
approaches
is
that the
lim/i(r)-"0
<-00
it
|A(0I<C
all*
(10.13)
system
Network
294
anal/sis
and synthesis
function possess poles in the left-half piane or on theyw axis only. More-
on theyVo axis must be sini]pie. 3 As a result of the requirement of simple poles on they'w axis, if H{,s) is given as
over, the poles
H(s)
as
bm
m
s
+
+
a n_ 1 s
n~ 1
m~ l
b m _ lS
+
+! +
+a
bis + b
fl x s
(10.14)
then the order of the numerator n cannot exceed the order of the denominator m by more than unity, that is, n m < 1. If n exceeded m
by more than unity, this would imply that at s =jo> = oo, and there
would be a multiple pole. To summarise, in order for a network to be
stable, the following three conditions oii its system function H(s) must
be satisfied:
1.
2.
3.
The degree of
the denominator by
Finally,
it
tjhey'w axis.
more than
unity.
< 0.*
In
is
zero for
10.2
HURWITZ POLYNOMIALS
In Section 10.1
its
we saw
the poles
on
theyco axis
*
i tt
In Chapter 6
it
of]
polynomials
known
they'co axis
gave
as Hurwitz
rise to
terms as
sin wot.
4
G. Raisbeck, "A Definition of Passive Linear Networks in Terms of Time and
Energy," /. Appl. Phys., 25 (Dec., 1954), 15ip-1514. The proof follows straightforwardly from the properties of the Laplace transform.
Elements of
polynomials.
polynomial P(s)
is
theory
realizability
295
2.
P(s)
As a
result
of these conditions,
P(s)
then
all
then
ac,
must be
= (s+
Hurwitz because
hand,
is
all
real
if s {
<x
(10.15)
+ j@ is a root of P(s),
The polynomial
negative.
P(s)
the coefficients a {
must be
if P(s) is
1)(j
+js/2)(s
-js/2)
G(s)
(s
l)(.y
2)(j
(10.16)
On the
other
(10.17)
3)
not Hurwitz because of the root s = 1, which has a positive real part.
Hurwitz polynomials have the following properties:
is
= y<
s = .jm
s cti.jPi
s
yt
a-i
Since P(s)
is
= (s +
y< Xji
and
positive
and
positive
(Of real
P(s)
real
real
<*)[(*
oO*
(10.18)
it
follows
may be
is
the polynomial
is neither even
nor odd. This is readily seen because the absence of a term a{ implies
cancellation brought about by a root s y< with a positive real part.
2. Both the odd and even parts of a Hurwitz polynomial P(s) have
roots on the ja> axis only. If we denote the odd part of P(s) as n(s) and
the even part as m(s), so that
fln-i. a-2>
fl <*i
P(s)
if
= n(s) + m(j)
(10.19)
Network
296
then m(s) and n(s) both have roots on tne jw axis only. The reader is
8
referred to a proof of this property by Guillemin.
3. As a result of property 2, if P(s) is either even or odd, all its roots are
on
they'co axis.
The continued
4.
quotient terms.
y(j)
= m(s)ln(s),
odd
odd
to even
n(s)lm(s) or
Suppose we denote the ratios as y>(s)
then the continued fraction expansion of y>(s) can be
written as
y(s)
= q xs +
(10.20)
+
ss
6
P(s) = (j) + m(s) is Hurwitz.
sion,
we must perform a
series
To
if
the polynomial
= 2
y>(s) is
(10.21)
n(s)
is of one higher degree than n(s). Then
obtain a single quotient and a remainder
where m(s)
m(s),
we
if
we
n(s)
The degree of the term R^s) is one lower than the degree of (*). Therefore
if we invert the remainder term and divide, we have
(L
Inverting
=q . + M2l
(io.23)
RM^qj + M)
*,(*)
(10.24)
*.()
E. Guillemin, The Mathematics of Circuit Analysis, John Wiley and Sons, New York,
An excellent treatment of Hurwitz polynomials is given here.
*
proof can be undertaken in connection with L-C driving-point functions; see
1949.
M.
Elements of
We
realteability
theory
297
of
ip(s)
We
R^s)
odd or odd
to even parts
7
finite in length.
fTO-irO^M
(10.25)
then F(s) is Hurwitz, if W(s) and F^s) are Hurwitz. For example,
test whether the polynomial
F(s)
is
- i* + s* +
5s*
3s
+4
let
us
(10.26)
n(s)
We now
= s* +
3s
= "W/C0 by
and then inverting and dividing again, as given by
dividing n(s)
by
m(s),
the operation
s*
+ 5s* + 4(s
s* + 3s*
2s* + 4)j + 3j(*/2
j + 2*
3s)s*
s)2s*
+ 4(2s
2s*
4)j(j/4
*,)
= 2& - 5 +
()
is
(10.28)
s/4
'
n(*).
Network
298
Since
all
analysis
and synthesis
Example
10.1.
Let us
test
= 5s + 2s9 +
GO)
is
+6
3s
(10.29)
is
division
2s2
6)5*
s*
+ 3* (s/2
+ 3s
by a
common
factor
5s
G(s)
We know
+ 3s is
s*
(s*
2/s
10.2.
is
+
>H)
3s)
which we referred to
= 57 + 2s* +
2s5
The term
(*)
wiO)
s*
3s
is
the mul-
earlier.
now
non-Hurwitz.
is
+ s* + 4s3 +
(10.30)
Hurwitz.
The continued
Hurwitz.
is
Example
85*
8*
+4
(10.31)
obtained.
1
'
fr
j^t)
(1 - 32)
(S4-+-4)
fV(s)
s*
W(s)
It is clear
that F(s)
Example 103.
is
(10.33)
not Hurwitz.
Let us consider a
F(s)
= 5* + 5s +
2t*
35
+2
(10.34)
Elements of
The continued
realizability
theory
299
fraction expansion is
3
3s)S*
s*
+ 2s* + 2 (s
+ 3s*
-j*+2)j +
3*(-,j
5s)-5*+2(-,s/5
-4*
2}5iXfs
then F(s)
is
then F'(s)
F'(s)
is
if
F(s)
is
given as
= s7 + 34s + 2s3 + *
= 7s* + 15** + 6s +
(10.35)
1
(10.36)
Without going into the details, it can be shown that the continued fraction
expansion of F(s)IF'(s) does not yield all positive quotients. Therefore F(s) is
not Hurwitz.
POSITIVE REAL
10.3
In
this section
known
we
FUNCTIONS
of a class of functions
These functions are important because
1.
F(s)
is
2.
The
real part
part of s
is
Re
[F(s)]
for
Re s
map
<r
on
F(s) plane.
*
onto, a point
would correspond
to,
or
cit.
Network
300
analysis
and synthesis
j\mF
JU
F(t) plane
plane
o
F(sO
(TO
F(ao)
FIG.
10.5.
Ref
map onto a point Ffo) in the right half of the F(s) plane.
In other words,
maps onto the
right half of the F(s) plane. The real axis of the s plane maps onto the
real axis of the F(s) plane.
A further restriction we will impose is that F(s) be rational. Consider
the following examples of p.r. functions:
= Ls (where L is
F(s)
is
a>0.
Re
Then
Therefore, F(s)
is p.r.
Ka
(f)"
If F(s) is
o*
>0
(10.37)
a>
an impedance
a capacitor of I IK farads.
We thus see that the basic passive impedances are
Similarly, it is clear that the admittances
sponding element
is
Y(s)
=K
7(5)
Y(s)
= K
p.r.
functions.
Ks
(10.38)
is
real
and
positive.
p.r.
upon
Elements of
in Fig. 10.6, the average
power
Average power
by the network
dissipated
Re
realizability
[ZJtjco)]
|/|
theory
301
is
(10.39)
[Zjytt))]
(10.40)
Consider
point impedance
is ZJ^s).
Let us load
the network with incidental dissipation
Passive
network
ZiW
ZxCO, then
Zfc)
= ZJis +
a)
is
real
Re Z^jco)
Re ZJa
so that
no.
(10.41)
and
positive.
10.6
Since Z^s)
is
the
(10.42)
+ jm) ^
(10.43)
driving-point impedance
admittance,
is
is
is
p.r.,
then
its
also p.r.
2.
we
see that if
p.r.
3.
=s
1/s is
not
p.r.
parts,
4.
Network
302
The
6.
may
differ at
zeros at 5
most by
oo.
0.
be
p.r.
are
(a) F(s)
(b) F(s)
may have only simple poles on theyco axis with real and positive
right-half plane.
residues.
(c)
Re F(jm)
for
a.
all
new
F(s)
2.
Re F(s)
is
real
when s is real.
0, when Re s ^
0.
jm
axis.
It is
Condition
i.e.,
F(s)
The
jCOi
+ jco^
K = Ki* so that
JO)
+ JCOx
s*
(O*
Elements of
If
x is
found to be
reaiizability
theory
303
conditions.
In order to
let
we must
To do this,
from the
F(s)
(10.45)
6(s)
We
can separate the even parts from the odd parts of P(s) and Q(s) so
that F(s)
is
8( S )
JV 2(s)
where Mj(s) is an even function and Ns) is an odd function. F(s) is now
decomposed into its even and odd parts by multiplying both P(s) and
Q(s)
by A/g
- Ns so that
r(s)
_ M + NiM*- N*
Mt + JVa Ma - Nt
= M M8t - N * Nt
M 8 - Nt
i
We
x
MM
MN
and
MN
s
are
odd
If
we
let s
= joy, we
Nx - Mj Nt
M* - Nt
(10.47)
JV,
Nt
^-^
is
(10.48)
is
Odd
while the
and
functions.
Ev[F(s) ]
and the odd part of F(s)
[F(s)]
^VY'
is
(10.49)
any polynomial
is real,
is
written as
F(ja>)
it is
clear that
and
Re
[F(jo>)]
Re
[F(ja>)]
j Im
[F(ja>)]
+ j Im [F(jo>)]
(10.50)
= Ev [F(s)] \_ja
= Odd [F(s)] \^, a
(10.51)
(10.52)
we
determine the real part of F(ja>) by finding the even part of F(s) and then
joy. We then check to see whether Re F{joy)
for all <w.
letting s
Network
304
A(u)
Single
root
FIG. 10.7
FIG. 10.8
a(ja>)*
That
is
- JVaO)2 =
M^to)*
+ N ((o)* ^
(10.53)
there
is
squared, gives
A(a>)
4 M O) MJjai) -
Ntito) NJJm)
(10.54)
F(s)
s*
to be p.r.
First,
we know
that, in
on they'd)
+
bs
(10.55)
must be greater
S*
+C
'
5*
(10.56)
C
We will show later that the coefficient a must also be zero when b = 0.
Let us proceed with the third requirement, namely,
the equation
Re F(ja>)
^ 0. From
- JVxO) NAjo>)
a(-(o* + c) + 1x0*^0
A(co) = (b a)afi + ac ;>
M^o)) Mfim)
we have
which
It is
of
simplifies to
(10.57)
(10.58a)
(10.586)
evident that in order to prevent A(m) from having positive real roots
eu,
a, that
is,
b^a. As
result,
when b
= 0,
fulfilled in
1. a, b,
2.
then a
0. To summarize, the conditions that must be
order for F(s) to be positive real are
^ 0.
c
a.
Fx(s) =
We see that
sr
is p.r.,
305
3s
\+
(10.59)
,
2
F<S) =
(ia60)
TT1
+2
s
As a second example,
let
biquadratic function
F(s)
TT^TT
+ b^ + b
<
ia62>
s"
We
definition.
F(s),
which
is
(s
_
The
s*
+
(* +
[(a
ft)
ti
bof
fliftjs'
ft x
a &.
(10.63)
Re
[F o1
0)
~ :y 18+
- W< - "
t
!
)" +
to
CO
O
+
(
flA
dO.64)
ti
We
Re
[F(jca)] is truly
always positive
m, t =
(a.
a 1b1
I^
[(flo
feo)
Re
_ aAf _ 4flA
[F(jco)]
(1Q65)
Network
306
Re
[F(jco)]
root.
10.65
is
zero (double,
(10.66)
Ko
or
( 10 - 67)
[(fl
If
then
*A ^ (V^ - yfb
or
(10.68)
(10.69)
)*
(10.70)
+ ^-aA<0
aA - (a + b ) <, 2VaA
< aA - (a +
fa, + * ) - aA <
If
(a,
then
but
aA ^ (\Za V*o)
so again
The second
when
2.
real root is
situation in
eo*
[F(jco)]
and
when
(10.75)
(10.76)
we have
Thus
We
(10.73)
(10.74)
[0*o
10.75
(10.72)
From Eq.
which Re
(10.71)
is
a necessary and
(10.77)
(10.78)
aA = (V^o~-V6o>
we will have double zeros for
Consider the following example:
then
F(s) =
s
s
We see that
aA = 2
so that F(s)
is p.r.
(s/a
Re
g
+
T
+
00.79)
[F(jeo)]
2s
5s
+
T
+
~
25
(10-80)
16
- V^ )2 = (V25 - Vl6 )*
(10.81)
Elements of
realizability
theory
307
The examples
illustrate the
and
real.
For example,
F(S)
+5
TTTT,
+ 1)
3s 8
<
ia82>
s(s*
-2s
5
+& = -rr7
+
= j is negative.
not
(io.83)
Therefore F(s)
p.r.
we can make
Z(s)
i77^
= Zi(s) +
two impedances in
we know
that Z(s)
p.r. functions,
<
Z^s)
must
also be p.r.
p.r.
(1085)
and
where a and
We see
then
m = JMM.
must also be
1084>
K are
real
and
F(s)
= -*s + a
(10.86)
F(s)
=
s + a
(10.87)
positive quantities,
must be
p.r.
We
then
+a
a
fi
must be
p.r. also.
+a
(10.88)
Network
308
analysis
and synthesis
whether
= _*L_
s + a
jr( s)
F(s)
V
we conclude
(10.90)
+ a/Ks
Therefore, the
that F(s)
sum of
is
the
also p.r.,
is p.r.
The
is
s/K
s/K and
10.4
(10.89)
If we write F(s) as
is p.r.
we
o,K0
to break
Ziis), Z4s), .... ZJs), and then to synthesize these individual Z,(j) as
elements of the overall network whose driving-point impedance is Z(j).
Zis)
= Us) + Zs) +
---
+ Zn(s)
(10.91)
of functions Zs).
general as
""
W=
7( *
fl
-s "
fe
is,
divide P(s)
by
we can denote
Z1(^)
and
Za(s).
Z(s)
= - + R(s)
(1(>
92)
and a remainder
R(s),
Let us
which
Z^s)
+ Z&)
From
(10.93)
D/s
3.
Re
[ZJJa>)]
for all m.
Elements of
realizabiiity
1 is satisfied
same argument.
theory
is satisfied
309
because
by
this
Re ZAjco)
From the
a- partial
foregoing discussion,
= Re Zjja>) ^
it is
seen that
if
made such
(10.94)
at s
still
remain
= 0,
is
of
p.r.
quotient Ls and a remainder term R(s), again denoted as Z^s) and Zt(s).
Us)
Here ZJs)
is
also p.r. If
ZXjs)
expanded into
(10.95)
(10.96)
Re
Here
so that
=Re (_i^L_\ =0
(_2*M
Zj(.y) is p.r.
will
Z(s)
ReZfjwi
74s)
ZJs)
(10.98)
Network
310
anal/sis
and synthesis
Zi(s)
ZzM
Z(s)
^
>
FIG. 10.10
where both 2^ and Zj are p.r. Now let us "remove" ZjCs) from Z(,s) to
give us a remainder Z^s). This removal process is illustrated in Fig.
10.10 and shows that removal corresponds to synthesis of ZjO).
Example
10.5.
function
p.r.
= s* +2s +6
s(s + 3)
(10.99)
We see that Z(s) has a pole at s = 0. A partial fraction expansion of Z(s) yields
*(*)
2
*
+3
-2&)+2fr)
If
(10.100)
we remove Zt(s) from Z(s), we obtain Z2(s), which can be shown by a resistor
1(
1Q
Z<s)
i"
Za
FIG.
10.11
Example 10.6
7s
+2
n>- 2j +4
(10.101)
Y(jco)].
The
real
14o>*
(10.102)
We
minimum of Re [Y(ja>)]
occurs at
to
Yx =\ mho
= 0,
and
is
from
Y{s)
and denote
equal to
Elements of
the remainder as YJs), as
p.r.
is
because
shown
we have removed
theory
in Fig. 10.12.
only the
realizability
minimum
Y^s)
obtained as
3s
(10.103)
It is readily
Thus the
capacitor.
final
is
f-farad
network
:jo
^20
>20
Y*'}
"
Y(s)
>-
YM
"
=f
>
J
FIG. 10.12
Example
10.7.
FIG. 10.13
Consider the
p.r.
impedance
fo8
3**
Z{s)
3s
(10.104)
Z1(s)=Z(s)-l =
The
reciprocal of
+3s
6s*
unity.
3^ +
6s*
Removing a constant
+ 3s
(10.105)
Zt(s) is an admittance
no) -
This pole
<x>.
fit
is
+3s
3s*
r1(i)=25+
5?4n
WWW
10
Z(s)
(10.106)
1
Zi(s)
FIG. 10.14
partial fraction
(10107)
Network
312
WWW
in
Z(3)
FIG. 10.15
Y&) The
reciprocal of YJs)
Tito
25
at *
10.15).
oo
YJs)
to give a capacitor of
is
now
obtained as
(10.108)
^--j-j
is
Zjfc)
3*
(10.109)
series with
a capacitor of
farad.
The
final
www10
ZM
3h
Z\z!=.
FIG. 10.16
These examples are, of course, special cases of the driving-point synproblem. However, they do illustrate the basic techniques involved.
In the next chapter, we will discuss the problem of synthesizing a network
with two kinds of elements, either L-C, R-C, or R-L networks. The synthesis techniques involved, however, will be the same.
thesis
Problems
10.1
()
s8
(ft)
5*
to
S1
+5* +2s + 2
+5* +5 + 1
+ 5s + 5* + 5
+ 45* + 55 + 2
to
5s
+25
V)
+5
+ 25* + 2/ + 5
31
10.2 Determine whether the following functions are p.r. For the functions
with the denominator already factored, perform a partial fraction expansion
first.
(a)
F()
**
F(s)
(&)
IX*8
(*
+
+
2X
1X5
3*
(s
F(s)
(c)
F(s)
()
F(s)
+2s +4
2s*
(s
+4*
.*
5**
+ 4)
+ 3)
+4
+
3*
103
F(s)
+s
+
**
+ 2)
10.4 Show that the product of two p.r. functions need not be p.r. Also show
that the ratio of one p.r. function to another may not be p.r. (Give one example
of each.)
10.5
GivenZfr)-
*+
*\
(a)
What
(A)
Find A* for
(c)
10.6
Re [ZQot)]
Prove that
if
on
10.7
Z()
Z(s) by
10.8
WW
positive real.
-j
first
= 0.
must also be
co
is
p.r.
W=
S8
+ 2j* + 3j + 1
+ 5s + 2* + 1
What does the continued expansion imply if Y(s) is the driving-point admittance
of a passive network ? Draw the network from the continued fraction.
3 4
1
Network
analysis
and synthesis
the
10.9 The following functions are impedance functions. Synthesize
impedances by successive removals of/ axis poles or by removing min [Re (/)].
()
+ 4s
7*T2
j + i
(*)
s(*
s8
2?
<c>
s*
2)
+4
27+1
+ 3s +
s + l
chapter
PROPERTIES OF
I.I
L-C
and then we
L-C
will
driving-point
IMMITTANCE FUNCTIONS
Let us
represent Z(s) as
Z(S)
where
N N
x,
M M
lt
are
odd
M^s)
i^
a (s)
+ JVj(s)
^i
+ N&)
/J J j-j
(U1)
dissipated
= J Re [Z(jd)] \I\*
Average power
(1 \ .2)
is
Re Z(
where
Ev Z(s)
Ev 2ija>)
= tf'('W)-^')^)
Mt\s)-Nt\s)
315
(11.3)
,,
n
(1L4)
,
Network
316
In order for
anal/sis
Ev Kjio)
and synthesis
= 0, that
MJlJm)
is,
80'o>)
~ W) fUja>) =
must hold:
M
M
= - Nt
N
t =
(a)
(11.5)
(b)
In case
(a), Z(i) is
and in case
We
(11.6)
see
(A)
from
this
Z(s)
Z(s)
(H.7)
(H-8)
functions:
'"
Zjj^s) or YxcC*) is tne ratio of even to odd or odd to even P013
nomials.
have only imaginary
2. Since both Mis) and Ns) are Hurwitz, they
or Y^s) are on
Z^s)
of
zeros
and
poles
the
that
follows
roots, and it
1.
L-C immittance
a,*'
a*1
function given by
a,
(u
9)
know,
Let us examine the constraints on the coefficients a { and *,. We
coeffireal,
the
positive
to
be
first of all, that in order for the impedance
impedance
an
that
know
also
We
cients must be real and positive.
qo is
function cannot have multiple poles or zeros on theyw axis. Since
the
and
numerator
the
of
powers
highest
the
axis,
defined to be on theyw
if
example,
For
unity.
most,
at
differ
by,
can
polynomials
denominator
of
the
order
highest
the
then
2n,
is
numerator
the
of
order
highest
the
= 0.
317
i.e.,
V+
so that the poles will be at *
(^V
*-
*i*
and
(11.10)
at
"*"
= 0,1,2,3
dm)
Z(S)
Jft'
we
partial fractions,
< 1112>
obtain
= S + -2p-j + -*&- +
KkS
(1L13)
where the t are the residues of the poles. Since these poles are
all on
theyco axis, the residues must be real and positive in order for
Z* to be
positive real. Letting s =jm, we see that Z(;o>) has zero
real part, and
can thus be written as a pure reactance jX(o>). Thus we have
ft),
ft)
ft)
= J x(<)
(11.14)
+K +
dco
Since
all
<o*
the residues
<
are positive,
dX(io)
a>,
(,,-cV
it is
IT*
A
is
similar development
is,
we have
(1U5)
L-C function,
<
1116>
= B(m)
Network
318
and synthesis
anal/sis
is
given as
+
_
^'(s'
+ co^ +
***
Letting s
= jco, we obtain
(11.18)
to*)
X{a>) as
= ; x(co) =
zo)
<o a*)
^^^ <}
Kco(-coa
+j
cog )
(11.19)
0,
Let us draw a curve of X(co) versus o>. Beginning with the zero at co
<o
as
encountered
let us examine the sequence of critical frequencies
next
the
positive,
always
increases. Since the slope of the X(io) curve is
large or
critical frequency we encounter is when X(co) becomes infinitely
from
goes
and
sign
changes
X(co)
a>
As we pass 2 ,
the pole is at a> 2
whenever we
at ,.
and zeros of the function must alternate. The particular X(a>) under
powers
discussion takes the form shown in Fig. 11.1. Since the highest
the
of the numerator and the denominator always differ by unity, and
oo,
s
=
and at
lowest powers also differ by one, we observe that at s =
frequency, whether a zero or a pole.
and a zero at
is a zero at s
external
called
oo
are
and s
s
are
frequencies
critical
critical frequencies, whereas the remaining finite
to t
and
a>
cog,
example,
previous
3,
referred to as internal. Thus, in the
there
is
always a
critical
ZiC(s)
or
YLC&s)
is
the ratio of
odd
nomials.
IM
FIG.
I.I
to even or even to
odd poly-
X(w)
+2
+1
+3
FIG.
I.Z
The poles and zeros are simple and lie on theyeo axis.
The poles and zeros interlace on they'eo axis.
The highest powers of numerator and denominator must
2.
3.
4.
by
5.
Z(5)
by
infinity.
left.
4)
+ l)(s + 3)
3
s + 4s + 5s
Z(s) =
3s* + 6s
2
K(s + l)(s* + 9)
z( =
2
(s + 2)(s + 10)
(s
differ
unity.
2.
1.
(11.20)
On
diagram
is
shown
in Fig. 11.2,
is
_ 2(s +
1)(s
s(s*
11.2
SYNTHESIS OF
L-C
whose pole-zero
an L-C immittance.
9)
(11.21)
+ 4)
DRIVING-POINT IMMITTANCES
L-C function
is
+ Kxs
(11.22)
cog
is
an impedance
Kx s
is
Network
320
and synthesis
anal/sis
2X,
tf.h
Wi
o^HTHf
l0Q0>
'-vOOOv-'
2X1
'
2! h
7T5" n
Z(s).
FIG.
2^5/(5
o) 42) is
11.3
_,.
Z(S)
+ l)(s* +
*- + 4)
9)
(11.23)
2s
+ * + -72-
(11.24)
11.4.
The
s( S
(s
+ 2)(s + 4)
+ l)(s* + 3)
HI
2h
if
Z<8)
FIG. 11.4
321
Klziz
Y(s),
FIG. 11.5
The
r(s)
is
is
=s+
s
s*
(11.26)
1
from which we synthesize the network shown in Fig. 11.6. The L-C
networks synthesized by partial fraction expansions are sometimes called
Foster-type networks. 1 The impedance form is sometimes called a Foster
series network and the admittance form is a Foster parallel network.
A useful property of L-C immittances is that the numerator and the
denominator always differ in degree by unity. Therefore, there is always
a zero or a pole at $ = oo. Suppose we consider the case of an L-C
impedance Z{s), whose numerator is of degree In and denominator
is of degree 2n I, giving Z(j) a pole at s = oo. We can remove this
pole by removing an impedance L^ so that the remainder function Z^s)
is still L-C:
ZjCO
degree by
1.
degree
ZCO
is
LiS
2n
(11.27)
1,
is
of
2,
Zt(s)
to give
Yt(s) =
l/Z^s),
Ys(s)
will
(11.28)
2h<
lf=t
YM.
if:
FIG.
a. M. Foster,
11.6
"A Reactance Theorem," Bell System Tech. J., No. 3 (1924), 259-267.
322
Network
Ll
c*3=
Cizk
FIG.
FIG. 11.8
1.7
We readily see that F3(s) has a zero at s oo, which we can invert and
remove. This process continues until the remainder is zero. Each time
we remove a pole, we remove an inductor or a capacitor depending upon
whether the function is an impedance or an admittance. Note that the
final structure
We
+ 4s1 + 3
= oo, which
we can remove by
remainder
Z,(j), as
shown
Then we have
in Fig. 11.8.
Zt(s) -
Z(s)
s
= -44s +
2s
y3(s),
as
may
first
and a
10s
+ 4s +
a
(11.30)
3
Inverting Z^s),
oo.
we
again remove
+
Y3(s)=Ys)-s= *f
s
4
4s + 10s
3
Removing
of | h and
the pole at s
oo of
Z^j)
llYs(s),
2h
o
Y3 (s)
if-
FIG.
1.9
FTT1
fs + 3
Ginr*
(11.31)
(11.32)
323
/toot^
2h
Remainder
Z4>
FIG. 11.10
as
shown
oo,
expansion
s*
+ 4s2 +
is
3)25
2s
+
+
+
fo +
4s3 +
12s3
16s(2s<-->
6s
I0s)s*
s*
+
+
4s 2
3(ts <->
3)4s3
5.2
2->
is'
4s3
+
+
I0s(is<->Z
is
3(|$ <->
3.2
2s
3)2y(fj*->
2s
We see that the quotients of the continued fraction expansion give the
elements of the ladder network. Because the continued fraction expansion
-'innp
2h
fh
**4=
FIG.
f"
ffi
11.11
324
Network
anal/sis an
synthesis
always inverts each remainder and divides, the successive quotients alternate between Z and Y and then Z again, as shown in the preceding
expansion. If the
necessarily be
(s*
i*
z( S)
The continued
+ l)(s* +
3)
j
z
;r
s(s* + 2)
(H.33)
is
2s
|j
i
The final synthesized network is shown in Fig.
5h
*p
*p
FIG.
1.12
previously as being
all
the poles
and zeros
325
can be said that both the Foster and the Cauer forms give
elements for a specified L-C driving-point
function. These realizations are sometimes known as canonical forms.
statement,
11.3
The
properties of
known
properties of
onto the
we
it
minimum number of
the
will
a axis. 8 We
assume that
all
will
two kinds of elements can be realized in a Foster form. Based upon this
assumption we can derive all the pertinent properties of R-C or R-L
driving-point functions.
Let us consider
first
the properties of
R-C
Fig. 11.3,
we can
Z(s)
- ^ + K + -&_ + -&- +
s
s + a
s +
g
R a = K, C = \\K Rx = Kja
x
(11.34)
cr
where C
l/A^,
x
X,
lf and so on. In
order for Eq. 11.34 to represent an R-C driving-point impedance, the
constants K\ and at must be positive and real. From this development,
two major properties of R-C impedances are obtained, and are listed in
the following.
Ci
Hf-
IH
Co
-AMr-
R2
*i
FIG. 11.13
Sons,
E.
Van Valkenburg,
New York,
Introduction to
Modern Network
Synthesis,
Network
326
and synthesis
anal/sis
The
poles of an
It
we
first let
a.
Z(<r)
= + K. + -&- + _&_ +
a
a+
a + a
o-j
da =
and
It is clear
Let us
? H
*
a*
(a
now look
(11.35)
(a
atf
1-
o%?
^^^0
da
that
derivative of
(n
^^
36}
(11-37)
If
is
not in the
circuit,
then Z(0)
is
simply the
sum of all
the resistances
in the circuit.
= R + R2 +
+ R
(11.38)
because all of the capacitors are open circuits at a = 0.
At a = oo, all the capacitors are short circuits. Thus, if R^ is in the
circuit, Z(oo) = R^. If R x is missing, then Z(oo) = 0. To summarize
Z(0)
these last
Z(0)
oo,
present
=
C missing
Z(oo)
If
we examine
the
o,
Rn
missing
Rm
present
Z(oo)
see that
(11.39)
Next,
let
327
function alternate.
nearest the origin
must be a
zero.
Therefore, if Z(s)
Z( S)
(5!
(s
critical
frequency nearest a
oo
given as
is
+
+
ga)(s
ffOO
+
+
g4)
(11.40)
or,)
Then, if Z(y) is R-C, the singularity nearest the origin must be a pole
which we will assume to be at s = gx the singularity furthest from
the origin must be a zero, which we will take to be s = g
Let us plot
4
;
z , ff) =
(g
(g
versus
g,
Z(0)
is
beginning at a
+
+
+
a^ia +
ga)(ff
g4)
(11.41)
a3)
and extending to a
Z(0)
= ^*
= oo.
At a
= 0,
(11.42)
= -a
= <V
We
= au
Z()
a*
FIG. 11.14
Network
328
anal/sis
and synthesis
In addition,
we see
>
<r4
>
>
<r8
or
>
Oi
(11.43)
0.
>1
that
(11.44)
a, ex.
2.
The
lie
on the negative
The
An
example of an
= oo must be a zero.
must be
R-C impedance
Z(s)
= (s +
real
and
positive.
is:
+ 4)(s +
+ 2)(s + 6)
l)(s
s(s
real axis,
R-C
8)
(11.45)
= (s + l)(s + 8)
(s + 2)(s + 4)
(s + 2)(s + 4)
Z(s) =
(s + D
(s + l)(s + 2)
Z(s) =
s(s + 3)
Z(s)
(11.46)
= ^ + K + -^i- +
S + Ot
s
a
R-C
(11.47)
we
FIG. 11.15
329
Fig. 11.15.
as an
R-L
R-C impedances.
is
to be realized as
It is therefore
important
an R-C impedance or an
admittance.
11.4
OR
R-L
ADMITTANCES
We
for
F(s)
The
3(a
2)(s
s(s
+ 4)
remainder function
(11.48)
3)
is
obtained as
(11.49)
where F(oo) = 3. If F(s) is an impedance Z(s), it must be an R-C impedance and it is realized in the series Foster form in Fig. 11.16. On the
other hand, if F(s) represents an admittance, we realize Y(s) as an R-L
network in the parallel Foster form (Fig. 11.17).
-\[
f
f
AV
30
10
If
Z(s)
FIG. 11.16
4
Van Valkenburg,
loc
cil.
330
Network
analysis
and synthesis
30.
*k:
Jo:
lh.
Y(s)
FIG. 11.17
l&y
35*+
9s
3s)3s*
9s
24(3
24>*
11.48
is
+ 3s($s
s* + f*
is)9s
24(27
9s
24)is(s/72
4*
If F(s) is
11.18. If F(s) is
-Wv
30
Z(s)
270
jinpr^-
o:
If:
Y(s)_
FIG. 11.18
FIG. 11.19
33
II
The immittance
parallel Foster
R-C admittance
F(s)
is
R-L impedance or a
given as
= KK s + K + -&- +
s + a
(11.50)
The
significant difference
pedance
between an
is
circuit is Kj(s + at); whereas, for the R-L impedance, the corresponding
term must be multiplied by an s in order to give an R-L tank circuit
consisting of a resistor in parallel with an inductor.
The properties of R-L impedance or R-C admittance functions can be
much
functions.
significant
derived in
on
Ki
s
(11.51)
This does not present any trouble, however, because the term above does
not represent an R-L impedance at all. To obtain the Foster form of an
R-L impedance, we will resort to the following artifice. Let us first expand
into partial fractions. If Z(s) is an R-L impedance, we will state
without proof here that the partial fraction expansion of 7^s)js yields
positive residues. 6 Thus, we have
2i_s)js
Z>=*2 + K +
a>
he.
Actually,
cit.
^-+-s
(11,52)
a{
see
Van Valkenburg
Network
332
and synthesis
analysis
-VW
ifi
40.
Z(s)
&={=
to:
FIG.
FIG. 11.20
where *o,
Z(s) in
Ku
K ^
we
If
0.
11.21
for synthesis.
m-
2(s
(s
+ l)(s + 3)
+ 2)(s + 6)
(11.53)
F(s) represents
satisfies
F{s)
is
F(s)
so
we see
F(s)js,
The
(11.54)
is
_ 2(5 +
s(s +
14
- 3)
2)(s + 6)
by
s,
we
4
t
+
2
(11.55)
obtain
is
i.
IX*
Is
+
s
(11.56)
remainder function
a zero at s
FIG. 11.22
= 0.
333
YM
FIG. 11.23
the pole at s
0.
is
obtained
power
8*
is
+ s*)6 + is + 2s*(l
6 + 4s + js*
4s + fs )12 + Ss + s\3ls
12 + J
ls + s*)4s +
4s +
2
f**(f
iV)& +
**(49/5s
aS
OAAA
S*
As*
Y(s),
11.6
SYNTHESIS OF CERTAIN
If F(s) is
shown
in Fig. 11.22.
Under
certain conditions,
R-L-C
R-L-C
FUNCTIONS
driving-point functions
may be
syn-
z(')~
4- 2s
VL+ it
+
S'
( 1L58)
334
is
Network
analysis
and synthesis
neither L-C, R-C, nor R-L. Nevertheless, the function can be synthesized
by continued
s*
fractions as shown.
+ s+
+ 2s + 2(l-Z
+ s+1
s + iy + s +
S2 + s
l>*
s*
i(s
+ l(s + \*-Z
s+l
1>
Y(s)
The
An
Y(s)
is
(11.59)
FIG. 11.24
= (s + 2)(s + 3)
(s + l)(s + 4)
negative,
=1+
s+1
we cannot
+4
all
is
(11.60)
When we
alternate
multiply by
y(a)
s,
we
i
6
(11.61)
+4
obtain,
ito-*
2
+
V
(11.62)
+4
Note that
y(j)
3_(l__U + J!
2
\3
1/
h
6
+4
(11.63)
335
60-
lofa:
Y(s)
in
At 4=
FIG. 11.25
The network
that
realized
is
is
given in Fig.
11.25.
we
try to
quotients result.
order of division to
make
5s
the quotients
all positive.
The continued
is
+ s*)4 + 5s + s%i
4 + * + !**
|s + is*)6 +
6 +
5s
+ s%n/5s
is
Hi
i**
+ Hs
Asy^s
+ **(
*
s*)&s(6I15s
As we
see,
network
is
WV
AAAr
-K-
:&
Z(t)
FIG. 11.26
The
resulting ladder
Network
336
and synthesis
anal/sis
R-L-C
it
was
under special
use of a ladder form or the Foster forms. These conditions are not
given here because they are rather involved. Instead, when a positive
Problems
11.1
(a)
Why?
4X* +
A_,, " jQ ++ 9X*
+ 25)
8
16)
(**
w=
There
is
113
no need to
fr*
8)
*fc+4)
L-C impedance.
+ 9Xi* + 25)
+ 4X* + 16)
1X5*
jCi"
Synthesize the
Z(s)
in the
1X*
7(
11.2
(j
Z*S)
'
i.e.,
6s*
+ 425* + 48
18j + 48s
J+
in henrys
=FC8
PROB.
II
11.4 There exists an L-C network with the same driving-point impedance as
the network shown in the figure. This alternate network should contain only
two elements. Find this network.
-nnnp
2h
337
lh
4=if
2W
If:
PROB.
11.5
11.4
Zta=
shown
+2
"i+2s*+25 +
If
is
(6) Synthesize
in
2b
PROB.
II
&
11.6 Indicate which of the following functions are either R-C, R-L, or
impedance functions.
s*
+2s
(a)
z(,)
" FT4?~+1
(b)
ZW
(c)
Z(5)=
Z(4)
(e)
Z(5)
s*
+ 4* + 3
+ 45 + 3
?T6F+8
5*
+ 5s + 6
+,
j
+ Ss* + 6
** +3
L-C
338
Network
analysis
and synthesis
If
ju
-5
-3
-1
PROB.
11.7
11.8 From the following functions, pick out the ones which are
tances and synthesize in one Foster and one Cauer form.
+ l)Qr + 3)
+ 2X* + 4)
sjs + 4X* + 8)
is + lXs + 6)
= 4js +
2js
Y(s)
(s
sis
= (s +
Y(s)
l)fr
3)
+2)
l)fr
sis
R-C admit-
+ 4)
+2)
11.9 Find the networks for the following functions. Both Foster and ladder
forms are required.
Z(s)
().
is
(b)
3is
Z(s)
find
1
2+Y
2s
Synthesize
+ 4)
sjs* +3)
+s* +6s +
Y as an L-C admittance.
PROB.
11.11
Y when
K9
=
4)
+2)
1)(*
s
11.10
1)(*
s(s
11.10
*w- 5
s*
+ 2s + 3j + 1
+ s* +2s + 1
339
11.12 Find the networks for the following functions in one Foster and one
Cauer form.
(s
Z(s)
11.13
2X*
20 +
+ 4)
+ 4)
0.5)(*
+s + 1
+ J8 +s
s* + s*+2s + l
Z(s)
V+4 + 3*+a + l
4^ + 3^+4^+2
Z(s)
2s +5
(s + 2X* + 4)
Synthesize Z(s) = i
i\/ v ^ mto tne form
(s + IX* + 5)
s*
Z(s)
+2s*
5
11.14
r*Tnnp
in the figure.
6
If
shown
Zfr).
PROB.
11.15
sents
Of the
11.14
series Foster
JO)
-4
-3 -2
-l
ju
-4
-3
-2
-1
JO)
-4 -3 -2 -1
PROB.
11.15
10
form.
Network
340
analysis
and synthesis
JU
*i
LU
PROB.
11.17
3s
(a)
s*+3
+ 105*
+ 455* + 105
IPs*
(c)
s*
11.16
s.
+ 15s
+ 15
s
s* + 1055 + 945*
W) 15** + 4205s + 945
s*
(*)
6s*
Synthesize networks for the functions above whose input impedances approxi-
mate tanh*.
chapter 12
Elements of transfer
function synthesis
12.1
h l/i-o
(12.1)
*1 lli-0
voltage-ratio transfer
In terms of the open-circuit transfer impedances, the
function
is
given as
T,
Yl
(12.2)
ratio
In terms of the short-circuit parameters, the voltage
Yl
is
= _ 2
z"
of the overall
= _?S*_
h
341
to be
(12.3)
shown
+*
network
as
shown
in
is
(12.4)
Network
342
and synthesis
analysis
Two-port
network
Vi
V2
FIG.
The
FIG. 12.2
12.1
= -' =
Vi
where
G=
l/R.
y2
is
(12.5)
+G
shown in
z ai2
*a
V.
as
V^Vg is
(z
RMtn.
*)
(12.6)
- 2gii8
us
discuss
certain
First, let
properties
which
FIG. 12.3
satisfied
is
when
T(s)
is
a rational
they'to axis.
a Hurwitz polynomial.
3.
that
Suppose P(s) and Q(s) are given in terms of even and odd
is,
ns)
where Ms)
is
m = MMIM
m +
parts,
(12 7)
.
w,(5)
e(S)
8( S )
even and Ns) is odd. Then T(ja>) is
MiC/oO
MJja>)
+
+
A^
NMa>)
T(ja>) is
=
UfA/co)
+ Nt*(ja>)]
(12.9)
and
is
Arg TQm)
is
mm
= arctan
343
'
If arg T(jO)
Now
let
1. The poles of zgl (,s) are also the poles of z (j) and z {s). However,
u
it
not all the poles of z u (s) and z M (y) are the poles ofzn (s). Recall that in
Chapter 9 we defined the z parameters in terms of a set of node equations
as
Z^s)
Zl ^12
2.o
If there
zlt ,
= z ii + Zj
Z 22 Z 22 + Zj
Z 12 = Z 12
z ii
poles of z u
and z. The
h
+
l_
*\l> 2*22
z'tt
reverse
is
not necessarily
true.
I,
v2
r
Vi
<*'ll. *'22
'l2
(a)
FIG. 12.4
3j
-o
v2
Network
344
anal/sis
and synthesis
A-i
r^
Vi
Y!
4"
y'n, y'22
y'12
"
3f
/2
>
Y2
V2
^T
*h
Vi
V*
FIG. 12.6
FIG. 1X5
poles of yis(s) are also the poles of y u(s) and y M(s). However,
not all of the poles of yu(s) and y M(s) are the poles of yM(s). This property
is readily seen when we examine the two-port network in Fig. 12.5.
2.
The
= y'* +
yit = v'm
y**
y%
2
ViM = -s +
y(s)
Yt
and
Yt
3s
= - + 3s
s
yis(s)
-3s
Jfc
k ak M
(12.11)
This equation
L-C
- klt ^
2x4
is
known
= >
to
Modem
345
/2
FIG. 12.7
-T7>
Vi
h
<
+
Vj
FIG. 12.8
111
ZEROS OF TRANSMISSION
zero of transmission
transmission, there
is
h
k
v2
FIG. IX*
FIG. 12.1*
346
Network
z3
Zx
y2
Vi
ZB
Y<1
FIG.
^6
12.11
will also
Zi
yn
yuVn.
and
(12.12)
Vi^ti
Vn-
(12.13)
Yl
5ll
h Usk
(12.14)
In Chapter 8 we saw that transfer functions that have zeros of transmission only on the jco axis or in the left-half plane are called minimum
phase functions. If the function has one or more zeros in the right-half
plane, then the function is nonmimimum phase. It will be shown now that
FIG. 12.12
347
FIG. 12.13
oo
zero of transmission at s
on
In Section 12.4 it may be seen that lattice and bridge circuits can easily
be nonminimum phase. It can also be demonstrated that when two ladder
networks are connected in parallel, the resulting structure may have
right-half-plane zeros. 2
12.3
SYNTHESIS OF
AND Zn WITH A
I-S2
TERMINATION
parameters of the
Zm
and
Ytl(s) is
'21
I"*-'
Before
and
(12.15)
1
Vn
y
as depicted in Figs. 12.13
expressed as
(12.16)
i
12.14.
it is
necessary
two important points. The first deals with the ratio of the odd
to even or even to odd parts of a Hurwitz polynomial Q(s). Suppose
to discuss
FIG. 12.14
Van Valkenburg,
op.
cit.,
Chapter
11.
Network
348
Q(s)
given as
is
Q(s)
where M(s)
is
M{s)
+ N(s)
and N{s)
(12.17)
is
the
odd
part.
We know
the
odd
is
an L-C driving-point
function.
The second point to be discussed is the fact that the open-circuit transfer
impedance ztl or the short-circuit transfer admittance yn of an L-C circuit
that in an L-C
is an odd function. To show this, we must remember
90 out of phase with the
circuit with steady-state input, the currents are
voltages.
or
and
Arctan^=Jrad
(12.18)
Arctan^=?rad
(12.19)
Y11 =
where P(s)
circuit
is
either
^ = ^+
M{s)
Q(s)
even or odd.
N(s)
Now, how do we
ya
Y" =
i+y
it
into the
form
(12.21)
The answer
is
quite simple.
ytl
P(s)/N(s)
In
1
[M(s)IN(s)]
(12>22)
From this we
yn
obtain
349
(12.23)
_M(s)
y"
On the
other hand,
if P(s) is
JV(s)
odd,
we
divide
by M(s) so that
Z(!VMW_
+ [tf(s)/M(s)]
y
1
(12 24)
.
M(s)
(12.25)
yM
M(5)
We assume that P(s), Mis), and N(s) do not possess common roots. For
our purposes, we will consider only the synthesis of Yn or Ztl with zeros
transmission at s
Zn(s)
= fW =
*('"
-i"-
"i
>
(12 26)
.
oo.
of transmission at s
We can now proceed with the matter of synthesis.
following example.
sr
We
3s
+ 4s + 2
numerator P(s)
is
a constant,
it
Consider the
oo.
Since the
350
Network
analysis
and synthesis
+ 4s
g
3ss + 2
+ 4s
s
z as
(12.28)
same poles. Our task is thus simpliz 22 so that the resulting network
synthesize
must
where
we
fied to the point
requires
that we first examine the
z
This
of
zeros
transmission
has the
21
possible structures of the networks which have the required zeros of
transmission and see if we can synthesize z 22 in one of those forms. For
the example that we are considering, a network which gives us three zeros
of transmission at s = oo is shown in Fig. 12.15. We can synthesize z22
to give us this structure by the following continued fraction expansion
.
of
l/z 22 .
3j
2)i
4s(ls+-Y
s+
|s
^)3j*
2(&s +-Z
3*
2Ys(is^Y
is
synthesized
oo
is
of a low-pass
As a second example, consider the transfer impedance
filter.
Zgl
(,s)
is
,+
<12S)
4.
-nRPP-
-ffflHT1
<D T
FIG. 12.15
+2
FIG. 12.16
351
FIG. 12.17
*n
-*
=
3s*
*
""
+2
s*
+ 4s
T
+2
(12.30)
3s*
at s =
is a high-pass
by a continued fraction expansion of M .
The
which
is realized
final realization is
shown
in Fig. 12.17.
Yn(s)
+ 4s + 2
which has two zeros of transmission at s = 0, and one zero
Since the numerator is even, we divide by s* + 4s so that
3s' + 2
y = -rrr
s + 4s
s*
(12.31)
3s*
at s
oo.
~x
(12.32)
Vi
i/
Vu
2s
5s/2
= ~
s* + 4
(12.33)
If we invert ylt we see that we have a series L-C combination, which gives
us another transmission zero at s
0, as represented by the f-farad
capacitor, and we have the zero of transmission at s
oo also when we
remove the inductor of f h. The final realization is shown in Fig. 12.18.
H(2h<
)Vt
FIG. 12.18
352
Network
analysis
and synthesis
1Q<
Network
Zn
\)
I
(a)
is
denoted as
Zn
In this section we will consider the synthesis of constant-resistance twoport networks. They derive their name from the fact that the impedance
looking in at either port is a constantresistance R when the other port is terminated in the same resistance R, as depicted
Two-port
network
in
Fig.
works
12.20.
Constant-resistance
net-
FIG.
12.20. Constant-resistance
synthesis,
constant-resistance
Y*Y
(12.34)
*.
t.
FIG.
Na
V2
353
R<
Nb
12.21. Constant-resistance
networks in tandem.
could be
be decomposed into a product of simpler voltage ratios, which
tandem.
in
connected
then
realized as constant-resistance networks, and
objective is to realize
K_
K(s
Va
(s
We
can
(12.35)
first
synthesize the
_ K (s - 2 )
s Po
V.
\\
v*
= K^s
s
Y*
VhjVa
p,
in
as constant-resistance networks
realize
(12.36)
Pi
-zj
= Xa(s
Vt
tandem to
-*i)
structures as
shown
in Figs. 12.22o
the input and output ports do not possess comclose examination, we see that the bridge and
a
Upon
terminals.
mon
are identical circuits. The bridge circuit is
12.22
Figs.
in
circuits
lattice
merely the unfolded version of the lattice. Consider the open-circuit
balanced structures;
i.e.,
impedance
First
we determine
the
as
ii
transfer
(12.37)
= V* ~ V
h
*'
z ai
(12.38)
Network
354
'.
(a)
+"
(b)
FIG.
and
is
obtained as follows.
r_
Next we
Za + Z b
Vt - Vt =
find that
_h
Vi
(Zb
(12.39)
- Z )I'
tt
= (Z -Za)|
6
FIG.
12.23. Analysis
of bridge
circuit.
(12.40)
^ = Z> ~ Z
so that
From
Now
(12.41)
us consider the
355
we
note that z M
= zu
terminated in a resistance
R, as shown in Fig. 12.226. What are the conditions on the open-circuit
parameters such that the lattice is a constant-resistance network? In
other words, what are the conditions upon z u and z n such that the input
impedance of the lattice terminated in the resistor R is also equal to R7
let
In Chapter 9
we found
Zu
Since z u
= *ii - -SjJ-r
Zn + R
In order for
Zu
z 11
lattice
network,
- zia =
(12.44)
we then have
iKZ,
which
(1243)
+*
For the
(12.42)
(12.45)
R*
(12.46)
Z^Lb
simplifies to give
z^R
V9
(zu
+ K)(zM + R) - ztfr*
lattice.
(12.47)
Network
356
which
anal/sis
simplifies to
and synthesis
IS-
5*&ii
(zn
V.
+ Kf - **i
= __Zt__2
2Rz11 + 2R
In terms of the element values of the
V,
From
(12.48)
lattice,
we have
- Za)R
R(Z b + Za) + 2R*
l(Z b
(12.49)
UZ b - (RVZh)]R
Vt
we obtain
R[Z b
(R*/Z)]
- R*)
+ R8) + 2RZ
a
a
_ j(Zt - R )
(Z + R?
2R a
KZ
(Z 6a
* (Z >
R)
(12.50)
Zb + R
In Eq. 12.50, the constant multiplier J comes about from the fact that
the source resistance
we can
express
Z6 in Eq.
12.50 in terms of
HP +
If
we
let
(12.51)
G as
G(s)]
1-G(5)
"
V2
Va
1R-Z
Example
12.1.
The voltage
(12.53)
2R + Za
let
be normalized to unity.
ratio is given as
2l=lJlI
2s + 1
V
(12.54)
we
which, as
recall, is
an
By
we have
(12.55)
Z =>s,
Since
Z^Za =
1,
we then
Za --s
obtain
(12.56)
shown
357
in Fig. 12.25.
If
FIG. 12.25
Example
12.2.
V ~2s +
'
is
shown
l s
2s
V
let
First,
we
~
+
l
1
Vt
j*
-2s +2
s*
+2s+2
V~
separate the numerator
(12.57)
+ 2s+2
in Fig. 12.26.
Yl
has already been synthesized,
s*
(12.58)
into
odd and
we have
V*
Va
(f
(5*
+ 2) - Is
+ 2) + 2s
(12.59)
jut
-<T
yi
-1
+1
-fl
FIG. 123*
Network
358
If
we
analysis
and synthesis
Vt
[(*
[(5*
+ 2)/2t] + 2)/2*] +
a* + 2
s
2s,
we
(12.60)
1
=2 +I
which consists of a $-h inductor in
pedance Z is then
series with
*-JT5
and
is
obtain
(12.61)
1 -farad
capacitor.
The im-
(1262)
voltage ratio
FIG. 12.27
FIG. I2JS
Next,
12.29.
let
if
ZA - *
(12.63)
359
ED
AMr
R
-vwR
Vi
.*.
Zb
Vt
Under
circuit.
(12.64)
Zt + R
R + Za
Vx
Exanple 123.
z>
s*
j*+25 +
(12.65)
l
us write
VJV1 as
V1
Vt
[2*/(*
(12.66)
1)]
Is
so that
Zb
and
We
recognize
circuit.
(12.67)
+l
~ s* + l
*
Za
as a parallel
is
L-C tank
shown
circuit
(12.68)
and Z as a
in Fig. 12.30.
2h
r-CH
-a/v\
10
i-vwk
10
10!
:2f
FIG. I2J0
series
L-C
tank
Network
360
Example
analysis
and synthesis
12.4.
(s
Vs
Fx "
in terms of
At
first,
(5
+ 2X* + 4)
+ 3X3* + 4)
we break up
circuits
(12.69)
connected in tandem.
ratios
S+1
5+3
Va
Vi
+4
and
va
ratio
(12.70)
(12.71)
35+4
VJVi, we have
s
5+3
zbl +
(12.72)
1
so that
Zn
+ 2 and
ratio
zal ~5+2
vjva we have
5+4
35+4
AnH
(12.73)
1
1
+zot
(12.74)
25
jr_.
FIG.
5+4
12.31
(12.75)
and
Zha
The final
synthesized network
is
shown
361
+4
(12.76)
'
2*
in Fig. 12.31.
Problems
12.1 Give an example of a network where: (a) a transfer function has
multiple zeros on they'to axis; (b) the residue of a pole of a transfer function on
thsjo> axis is negative.
12.2
Show
shown
in the
figure.
lo-
lh
'TflHP
3h
10
lo-^VVW-
-o2
af
-o2
Hf:io
=fc*f
iq:
-o2'
l'o-
-o2'
l'o(b)
(a)
PROB.
find
12.2
nnnp
ff
lf^=
^=2f
12.4
ln
.20
f:
lh
S10
PROB.
in.
12.3
show
362
Network
anal/sis
and synthesis
s
R
JC
Vi Zfci-*
zb
* *
(a)
%
Vi
Zta -ii
*^
PROS.
123 For the networks in Prob.
v2
12.4
yjvv
12.6
Yl.
V
(b) Synthesize
2+Y
+ 2)
+ 2s + 2
O.S(*
s*
10
PROB.
(a)
that
Ywhen
l\
12.7
show
<
V2
12.4
Zh 1,
tt
then
L_
Vt
+za
Za and Zh if
_
_
V ~
l
12.8
363
s*
s*
+3s +
+45* +5j
+2
V,
s+2
Vl
Vt
Vx " 2** + 2* + 6
3(jr + 0.5)
Vt
Vx " 4j + 1.5
(e)
12
2(s*+3)
=4fi,i?1
~into
give
12j*
Yl
"
V,
15i
+ 7* + 2
PROB. IL9
12.10
Yl
(a) Synthesize AT. as
_L_
+3
2 2s
Na to give
a constant-resistance
lattice. (R =- 1 n.)
a constant-resistance ladder as in Prob. 12.4.
as a constant-resistance bridged-T circuit. (R -
12.11
()
s*
(*)
(c)
id)
Z- *
+3s* +3s
+2
3s*
3s
3s
s*
'a
s*
s*
+ 3s* + 3s +2
(*
+ 2)
3s*
+
s*
Y-
s*
(e)
'
shown
(Jt
1
Q.)
Q.)
in the figure
364
Network
anal/sis
and synthesis
PROB.I2.II
12.12
()
(b)
12.13
circuits.
-s+1
+s + 1
h ~ s* - 3s + 2
vt s* + 3s +2
via s* - 20s* + 5s Vx(s) ^ s* + 20s* + 5s +
V,
s*
Vx
s*
20
20
chapter 13
Topics
13.1
THE
FILTER DESIGN
design
in filter
PROBLEM
13.2
The
IN
tion/(rr)
365
Network
366
by the
/at*;
<*i
1.
When we
and w(x)
let
= f{x)
most common:
aj is minimized where
is
intervals.
2.
at
first
derivatives of e are
made
to vanish
= *o.
x
3.
where
/*
is
|lnx-
The value of
<, xt
4. Interpolation.
e is
made
/(*;
!,..,*)=
(13.1)
fa(x; 04,
where x
t.
Since an
...
oc)
**'
oe,
e"*
(13.2)
numbers.
The keystone of any approximation problem
lies in
the choice of a
can be simplified
oc's
The problem
error criteria. All the error criteria cited, except the Chebyshev, can then
be reduced to a set of linear algebraic equations for the unknowns
ii
>
Time-domain approximation
The principal problem of time-domain approximation
consists of
Topics
in filter
367
design
= ["[fc(0-**(')f *
minimum.
is
**(0-Z.0
k=l
(13.3)
Jo
*&(<)] dt
*-i
minimized when
is
.=r
h(t)<f>k (t)dt
l,2,...,n
(13.4)
,ri
sum
(13.4)
*=1
#*(*)=!
has a transform
*=1 s
Realizability is insured if in the
1, 2,
n.
(13.5)
sk
orthonormal
set {a te***},
Re ^
^ 0;
Frequency-domain approximation
In frequency-domain approximation
the
principal
problem is to find a
whose magnitude
-1
+1
FIG . I3X
Ideal
fihcrchar_
acteristic.
Trans, on Circuit
Network
368
different
analysis
and synthesis
ideal low-pass:
whose phase
is
approximately
is
is
here.
13.3
In Chapter 10
we saw
FILTER
filter
in Fig. 13.1
is
not
its
is
mations.
In low-p.ass filter design, if we assume that all the zeros of the system
function are at infinity, the magnitude function takes the general form
where
is
is
For example,
a)
if
2"
(13.7)
M()
,.
(1
K,,,
+ o>\2nn)^
is
M(l)
= -^
< 13 - 8 )
is
alln
at to
for
all n,
that
is,
(13.9)
is thus seen to be <w = 1. The parameter n conof approximation in both the pass band and the stop
band. Curves of M(m) for different n are shown in Fig. 13.2. Observe
The
cutoff frequency
Topics
Radian frequency,
0.3
0.2
0.1
0.6
0.4
369
design
In filter
a
0.8
5^ ,v
-2
-4
\
-6
n=3
-8
n=5
n
\VV
=7
-10
I*
-12
\\\
-14
^ \
-16
\\
-18
-20
\\
-22
-24
-26
FIG.
13.2.
filters.
that the higher n is, the better the approximation. The amplitude approximation of the type in Eq. 13.8 is called a Butterworth or maximally flat
response. The reason for the term "maximally fiat" is that when we
expand M(to) in a power series about <o = 0, we have
Jlf(eo)
We
(0
- Ao(l -
l<o
= 0.
For
>
1,
tn
+ |co 1
A<o"
i^eo""
(13.10)
M(w) =!
<wl
(13.11)
to"
wrn
20 log M(co)
is
for a Butterworth
obtained as
(13.12)
Network
370
analysis
and synthesis
One
How
question remains.
falls
asymptotically at a rate of
do we obtain a
characteristics M(co)l
The procedure
is
as
We first note that the amplitude response M{m) and the complex
M\w) =
If
we
new
define a
(13.13)
we
H(j<o) H(-jco)
see that
H(-s)
H(s)
(<o)
(13.14)
^(-ot) 2)
(13.15)
to give (s 2).
the poles
and
zeros of H(s) are the mirror images of the poles and zeros of H(s), we
simply choose the Hurwitz factors of A(js) as H(s). An example will serve
3)
Butterworth
response given by
JW
\OJ)
+ to
(13.16)
(s )
is
=
1
Factoring
(**),
(13.17)
(13.18)
--(s8)8
we obtain
1
(**)
2s
2s
s'
2s
2s
s*
= H(s)H(-s)
(13.19)
We then have
if(s)
=
s
2s
2s
1
1
l)(s
+ i + A/3/2X* + i - A/3/2)
(13-20)
The poles of #(y) and fT(-*) are shown in Fig. 13.3. Observe that the
poles of H(s) are mirror images of the poles of H(s), as given by the
theorem on Hurwitz polynomials in the Chapter 2.
Topics
13.3.
Poles of H(s)
design
371
-^ f->
HM^.-
FIG.
in filter
H(-s)
for
an n
= 3 Butterworth filter.
For a Butterworth response, the poles of H(s) H(s) are the roots of
(-1)V B = -1
=
The poles
**-"
or simply by
sk
Expressing st as s k
=
=
e*<-u/ft*
e
(13.21)
1
(13.22)
n odd
fc
ak
In
even
i(i/n) '
<+'-i>/an] I
0, 1,
by
+ ja> k
= o, 1, 2,
,2n
(13.23)
by
ow
1
ir
sin
2n
/2fc
1\tt
/2
1
(13.24)
ct)v
2k
sin
+ n-
l
IT
cos
In
/2fc-
1\tt
12
It is seen from Eqs. 13.22 and 13.23 that all the poles of H(s) H(s) are
located on the unit circle in the s plane, and are symmetrical about both
the a and the jco axes. To satisfy readability conditions, we associate the
poles in the right-half plane with H(s), and the poles in the left-half
st
^(fc+*)/8]
e"
(13.25)
Network
372
H(s)
then given as
is
H($)
=
(S
If
we
i{
e ')(s
13.2 for n
(,
e>
*>*)
_
2
0.76536s
l)(s*
to n
8,
1.84776s
and
(13 26)
"
(s
To
*Xa
H( s\
13.1
.)(,
in factored
(12 27)
1)
given in Tables
is
form as in Eq.
13.27, or
multiplied out as
H(s)
-^
a ns
fln-xs"-
TABLE
a ts
(13.28)
1
13.1
+1
+ V2s + 1
(**+* + 1)(* + 1)
(s2 + 0.76536* + IX** + 1.84776* + 1)
8
2
(* + 1)(* + 0.6180* + IX* + 1.6180* + 1)
2
(** + 0.5176* + IX* + ^2* + IX*2 + 1.9318* + 1)
2
2
2
(* + IX* + 0.4450* + IX* + 1.2456* + IX* + 1.8022* + 1)
(s2 + 0.3986s + IX*2 + 1.1110* + IX*2 + 1.6630* + IX*2 + 1.9622* +
5s
4
5
6
7
TABLE
1)
13.2
Butterworth Polynomials *
a*
V2
2.613
3.414
2.613
3.236
5.236
5.236
3.236
6
7
3.864
7.464
9.141
7.464
3.864
4.494
10.103
14.606
14.606
10.103
4.494
5.126
13.138
21.848
25.691
21.848
13.138
1
1
1
5.126
Topics
13.4
OTHER LOW-PASS
In Section 13.3,
low-pass
filter
we examined
characteristic.
approximants in
FILTER
design
373
APPROXIMATIONS
the maximally
We
in filter
approximation to a
flat
this section.
equal-ripple approximation
have seen that the maximally flat approximation to the ideal lowpass filter is best at co
0, whereas, as we approach the cutoff frequency
We now
<w = l, the approximation becomes progressively poorer.
consider an approximation which "ripples" about unity in the pass band
and falls off rapidly beyond the cutoff co = 1. The approximation is
The Chebyshev or
We
equally
good
at
co
ripple approximation.
and co = 1, and, as a result is called an equalThe equal-ripple property is brought about by the
|co|
<, 1
|co|
>
(13.29)
(13.30)
(13.31)
^ ^^ ^ ^
Thus
for n
In Table
= 2, we obtain Cg(co) as
Cg(co) = 2<o((o)
= 2eo 8 - 1
13.3,
TABLE
Chebyshev polynomials
(13.33)
10 are given.
13.3
1
1
CO
2co*
4e3
4
5
6
7
8
9
10
-1
3o>
8a>* - 8a>* + 1
16a> 6 - 20a>8 + 5o>
32o>* - 48<o + 18a>* - 1
64o> 7 - 112a> + 56eo8 - lea
128a>8 - 256w + 160w4 - 32o>* + 1
256o> - 576o7 + 432o) - 120a.8 + 9a>
512a 10 - 1280w8 + 1120a>* - 400eo* + 50* -
Network
374
FIG.
The
and synthesis
analysis
13.4. C,(co)
and
Chebyshev polynomials.
C,(u>)
filter
1.
approximation are:
The
C3(o))
Beyond
values of
is,
and
\w\ <, 1,
|C(eo)| <,
the interval
<
\a>\
|eo|
<, 1,
for
1,
\co\
CB (co)
never
<, 1.
\a>\.
filter
in the interval
e2
\u>\
Now we
1.
slightly greater
we
add
<,
|<w|
will associate
IHO"))!*
=
1
<
to this function
varies between
1.
and
making
e2,
it 1
a quantity
we obtain
+ *C n\a>)
(13.34)
\H(ja>)\ 2 oscillates
maximum value is
|eo|
1,
is 1/(1
e 2).
and (1
e 2 )- -*.
Topics
(l
in filter
-^^-
+ e2) *
FIG.
in the pass
<o
because
\H(jl)\
1, |H(;>)| is
Cn\\) =
where
e*
is
given as
1
(13.35)
(!+>"
+ ^a
Cn*{(o)
(13.36)
is,
for
\<o\
1,
as a> increases,
we
reach a point
so that
]ff(Mlas-
>
O)
eC n(a>)
The
filter.
1.
band
=1
Ripple
At
375
design
(13.37)
0) k
Loss
<w,
CB(co)
)\
can be approximated by
its
(13.38)
_1
leading term 2" <w",
so that
Loss
(13.39)
Network
376
We
db/decade after an
band by choosing a
the stop
cosh 0k
sufficiently
large n.
From
we
see
sinh
0*
band fixes n.
The derivation of the system function
H(s) from a Chebyshev amplitude approximation \H(jco)\ is somewhat involved and
not be given here. 2 Instead, we will
simply give the results of such a derivation.
will
First
we
- sinh
where n
is
(13.40)
sinh
cok
a
/S fc
cosh*
(13.41)
/?*
The major semiaxis of the ellipse is on the jco axis and has a value co =
cosh pk The minor semiaxis has a value a = sirih(ik and the foci
are at co = 1 (Fig. 13.6). The half-power point of the equal-ripple
.
amplitude response occurs at the point where the ellipse intersects the^eo
cosh fik . Recall that for the Butterworth response, the
axis, i.e., at co
half-power point occurs at co = 1. Let us normalize the Chebyshev poles
1 instead of at co
sk such that the half-power point also falls at co
cosh fi k ; i.e., let us choose a normalizing factor, cosh P k , such that the
Network
Synthesis,
M.
E.
Van Valkenburg,
New
Introduction to
13.
Modern
Topics
s' t
in filter
377
design
are given by
cosh
(th
cosh
/?*
cosh
= <**+./>*
locations can be derived as
= tanhu &
sin
<r't
/2fc
- lW
^j
(13.43)
/2fc-lW
a>t
= cos ^__j_
*i- -1 +J0
j23
_0.5 /0.866
4 --1(0.444)+ JO
szs
-0.444
-0.5(0.444) ;0.866
-0.222 y0.866
by cosh
/3 t that
,
+;0
Chebyshev
poles,
we simply
multiply
is,
Sk
(a\
(1 3.44)
Network
378
and synthesis
anal/sis
Butterworth
pole locus
Chebyshev
pole locus
FIG.
shown in
13.7.
Chebyshev
poles.
we draw
filter
Finally,
we draw
vertical
dashed lines from the intersections of the smaller circle and the radial lines,
and horizontal dashed lines from the intersections of the large circle and
the radial lines. The Chebyshev poles are located at the intersection of
the vertical and horizontal dashed lines, as shown in Fig. 13.7.
Consider the following example. We would like to obtain a system
function H(s) that exhibits a Chebyshev characteristic with not more than
ripple in the pass
1 -decibel
1-decibel ripple,
\H(jl)\
20 log -t-zu
'(!+)*
We then obtain
(1
and
* )
e
Our next
task
is
to
1,
Solving for ,
1A
= -1
w=
2.
\H(j\)\ is
0.509
134 5)
(13.46)
(13.47)
we obtain
0.891
to find n
20
1)
specification.
20 log 2
(13.48)
2.65. Since
fied.
20 decibels at
that at
decibel so that
20 log
From
at least
we know
First
we must
Topics
find
Pk
From
Eq. 13.40
in filter
379
design
we have
= I sinh"
(13.49)
1.965
0.476
tanh
From Table
13.1, the
*!
pk
tanh 0.476
= 0.443
(13.50)
-1.0,
s 23
-0.5 y0.866
(13.51)
Multiplying the real parts of these poles by 0.443, we obtain the normalized
Chebyshev
poles.
s\
-0.443,
s' 2S
-0.222 /0.866
H(s)
is
H(s)
then
^02
(s
(13.52)'
v
0.502
=
s*
0.992s*
/0.972)
1.255s
0.502
Monotonic
filters
with
optimum
cutoff
with Chebyshev niters, the following
can be said. The Butterworth response is a maximally flat, monotonic
response, whereas the Chebyshev response is equal ripple in the pass band.
In the stop band, the Chebyshev response falls off more rapidly than the
Butterworth (except when e is very, very small). In this respect, the Chebyshev filter is a better filter than the Butterworth. However, as we shall see
in Section 13.5, the transient response of the Chebyshev filter is very poor.
If we require sharp cutoff characteristics for a given degree n, however,
the Butterworth filter is quite unsatisfactory. In 1958, Papoulis 8 proposed
In comparing Butterworth
A. Papoulis, "Optimum
March
filters
Filters with
3,
380
Network
analysis
and synthesis
Radian frequency,
02
0.1
0.6
0.4
0.3
o>
0.8
iI
'-;>< \
-1
2
\
-\\
\\
Amplitude response of n
-/
-8
Chebyshev
1.0-db ripple
in
with
filter
pass band
Buttterworth respons>e (n
\
3)
\\
\\
-10
\\
11
\\
I
12
13
-14
FIG.
13.8.
Amplitude response of n
(n
"L"
filters,
properties:
1.
2.
if monotonicity is
assumed.
3.
The
L filter
M(o>)
filter
=
[1
with
all
zeros at
(13.53)
+JV)l*
L(ft> 8)
L filter by
(13.54)
Topics
in filter
design
381
The polynomial
=
L(l) =
LJO)
(a)
(b)
^^0
(0
dco
dL
>
2
>
=M
(M maximum)
dco
and c are the same as for the Butterworth generating polynomial/^*) = to2 ". Property c insures that the response M(a>) is monotonic and property d requires that the slope of L n(caP) at tu = 1 be the
Properties a, b
Ln (for n
odd) to be
Ln((ot)=
where n = 2k
4
first kind
dx
a<p<(a;)
Llo
(13,55)
=1
P^x) = x
P2(x) = K3* - 1)
P9(x) = i(5x* - 3x)
/>,>(*)
a, are given
by
===
3
(1356)
2k
= 1=
1
V2(fc
(13.57)
1)
Later Papoulis 6 and, independently, Fukada,' showed that the evenordered L n polynomials can be given by
rim*-!
(a>8)
(*
-J
2k
1)[2>
-|
*>(*)]
dx
(13.58)
E. Jahnke and F. Emde, Tables of Functions, Dover Publications, New York, 1945.
A. Papoulis, "On Monotonic Response Filters," Proc. IRE, 47, February 1959,
332-333.
M. Fukada, "Optimum
Trans.
3,
382
Network
a*
(13.59)
V(fc
l)(fc
2)
V(fc
l)(k
2)
polynomials up to n
7 together with
2fc
1
1
M=
Case 2 (k odd):
a*
3
flo
+l
==.*
= 2
=
7
2fc
To
is
LB(w 2)
to
shown
(13.60)
in Table 13.4.
L filter, we must
factor the
3,
M\a>)
is
=
1
L^co*)
(13.62)
1
Substituting
a>*
<w = s we obtain
2
=
) = H(s) H(-s)
2
3>
3eo
fc(s
TABLE
L(a> 2
s*
3s
(13.63)
35*
13.4
Polynomials
dLn(l)
dm
Ln(a>2)
CO*
- 3o>4 + <u2
4
6eo 8 - 6o>* + 3to
10 - 40a>8
20o
+ 28a>8 - 8a>4 + to2
12 - 120a> 10 + 105e8 - 40w* + 6o>4
50to
10 - 355o>8 + 105a>
175*o M - 525a> + 615>
3o>8
12
18
24
-15eo4
a)
32
Topics
in filter
design
383
Frequency, rad/sec
02
0.1
0.3
0.4
0.6
0.8
-2
-4
1.0
Amplitude i esponsecrf
OpB mum vers
Butbsrworth fit ers
-6
\\
\\
-8
\\
\\
tf-10
0.4
|-12
-14
-16
0.5
Frequency
0.6
0.8
1.0
1.2
\ \
\\
y
-1.0
\\
\\
-18
\
-20
ft-
\
A
-22
Explanded
scale
-24
\
\
\\
-26
FIG.
After
we
13.9.
factor A(s*),
we
if(s)
filters
obtain
0.577
=
s*
1.31s
1.359s
+ 0.577
(13.64)
where the numerator factor, 0.577, is chosen to let the d-c gain be unity.
poles of H(s) are sx = -0.62; s2>3 = -0.345 ;0.901. The amplitude response of third-order Optimum (L) and Butterworth niters are
compared in Fig. 13.9 Note that the amplitude response of the Optimum
filter is not maximally flat, although still monotonic. However,
the cutoff
characteristic of the Optimum filter is sharper than the cutoff of the
The
Butterworth
filter.
is
given by
H(s)
where
= Ke~'T
Then
(13.65)
H{jm)
Ke-"" T
(13.66)
Network
384
sponse
#w)
is linear
in
= -mT
{e(t),
re-
(13.67)
co.
to
an excitation denoted by
E(s)} is
R(s)
= K E(s)e-' T
r(t)
(13.68)
can be written as
KO-c-W)]
= Ke(t -
T)u(t
(1369)
T)
We see that the response r(t) is simply the excitation delayed by a time
and multiplied by a constant. Thus no
T,
from
response
<f>((o),
by
<u;
that
is,
Delay
_ - f^2> - r
(13.70)
dco
in Eq. 13.65
a delay
is
line.
= Ke~ ,T by
a rational
(13.71)
o
sinh
sT + cosh sT
=1.
to obtain
Kjsmhs
coth
'
W.
E.
1952, 256-263.
(13?2)
If sinh s
in
power
Topics
in filter
series,
we have
s*
s*
s*
= l++-+-+
385
---
(1373)
,'
.'
6
s
s
sinh5- S + - + - + - +
s
design
'--
sion of coth s as
coth
=-+
3
(13.74)
~*
z+...
5
written as
where
Bt = s+1
From these
formulas,
(13.76)
2?
we
obtain
= s* + 3s + 3
2?, = j + 6s* + 15s +
if,
(13.77)
15
Higher order Bessel polynomials are given in Table 13.5, and the roots of
Bessel polynomials are given in Table 13.6. Note that the roots are all in
the left-half plane. A more extensive table of roots of Bessel polynomi8
als is given by Orchard.
The amplitude and phase response of a system function employing an
unnormalized third-order Bessel polynomial
W=
Ms)
s*
^+
6s*
15s
(13.78)
15
IEEE Trans,
Network
386
analysis
and synthesis
TABLE
13.5
bo
bt
b*
b7
1
1
15
15
105
105
10
945
945
6
45
420
105
15
10,395
10,395
4,725
1,260
210
21
135,135
135,135
62,370
17,325
3,150
378
28
are given by the solid lines in Figs. 13.10 and 13.11. These are compared
with the amplitude and phase of an unnormalized third-order Butterworth
function given by the dotted lines. Note that the phase response of the
constant-delay function is more linear than the phase of the Butterworth
function. Also, the amplitude cutoff of the constant-delay curve is more
TABLE
13.6
2
3
-1.0 +y0
-1.5 y0.866025
[-2.32219 +y0
1.83891 yl.75438
i
/ -2.89621 y0.86723
\ -2.10379 y2.65742
-3.64674 +j0
-3.35196 yl.74266
[-2.32467 /3.57102
C
-4.24836 y0.86751
-3.73571 /2.62627
-2.51593 y4.49267
-4.97179 +j0
-4.75829 yl .73929
-4.07014 y3.51717
1-2.68568 y5.42069
r
Topics
0.1
0.4
0.2
in filter
Frequency, rad/sec
0.6
0.8 1
-2
387
design
6 7
\
>^
-4
-Bessel
-6
-8
,-10
B it* srv fO rth
h
!
\
\
-14
-16
-18
<
-20
-22
-24
\
\
-26
FIG. 13.10. Amplitude response of n
Frequency, rad/sec
0.2
0.4
FIG.
0.6
13.1 1.
0.8
1.0
1.2
1.4
filters.
1.6
1.8
2.0
388
13.5
Network
analysis
and synthesis
compare the
In this section
will
Rise time
The
rise
is
defined here as
from 10%
to
90%
of
its
settling time.
3. Settling time.
The
settling
time
is
from the
that time
final
t,
value by
Delay time,
requires to reach
50%
a percentage of the
FIC.
tB
12
10
Time
t,
sec
delay time.
final value.
ta
= rise
time;
/.
setting time;
Topics
in filter
389
design
as rise time
frequency response.
Rise time and bandwidth have an inverse relationship in a filter. The
wider the bandwidth, the smaller the rise time; the narrower the bandwidth, the longer the rise time. Physically, the inverse relationship could
frequencies slows
TB X
BW = Constant
(13.79)
Bandwidth
(/.
- UTJ
(milliseconds)
0.5
f.
The
table
shows a
2/
0.25
3/c
0.16
4/
0.12
y.
0.10
between
rise
time and
bandwidth.
is
first
moment
or
h{t) dt
(13.80)
TB
(13.81)
IEEE
Network
390
His)
FIG.
13.13.
R-C network.
+ lS + a *s2 +
+ b s + 6js +
x
'
+ a*s n
+ b ns m
(13.82)
and the
rise
time
bl
ax
af
2(a 2
TD =
TD i%
TR =
{2tt[V
H(s)
- b )]}*
(13.84)
is
V(s)
I(s)
(13.85)
sRC
TD = RC_
TB = yJlnRC
so that
It
(13.83)
is
(13.86)
no formal mathematical
Overshoot
excess gain
is
definition.
generally caused
by "excess" gain
at high frequencies.
characteristic with a
By
peak
such as the shunt peaked response shown by the dashed curve in Fig.
13.14. A magnitude characteristic with no overshoot is the magnitude
characteristic of
Log frequency
FIG.
10
V,
13.14.
1,
Series,
Div.
Topics
in filter design
391
1.2
1.1
1.0
0.9
?0.8
N
=5 0.7
5
c
0.6
0.5
In
f 6.4
I
J 3 ~A> = 7/il=lo"
0.3
0.2
0.1
10
Time
FIG.
13.15.
t,
12
16
14
18
20
sec
filters.
characteristics
The
step response of
to the response of
filter
that
The
is characteristic
decision as to which
filter is
best depends
filters.
upon
the particular
situation.
phase
is
time
is
not suitable, because the pulses would "smear" over each other
Network
392
1.1
1.0
/
t
t
0.9
0.8
and synthesis
analysis
'
^ ^ s
!
1
0.7
S 0.6
3 Bessel fitter
-Step response of n
-Step response of n 3 Chebyshev
filter with 1-db ripple
nr.
a 0.5
i
<0.4
1
|
0.3
0.2
~r
0.1
ly
12
10
14
16
18
20
Time, sec
System
with short
rise
and
settling
Ti
times
Input pulse train
ALA
T2
Output pulse
train
(a)
System
with long
rise
and
settling
Ti
times
rise
and
setting times.
as seen in Fig. 13.176. The same can be said for long settling times.
Since a pulse transmission system must have linear phase to insure undistorted harmonic reconstruction at the receiver, the best filter for the
system
13.6
is
a linear phase
filter
with small
rise
and
settling times.
IN FILTERS
We
present here a
Topics
393
in filter design
\2
1.0
0.8
tep re sponsc
0.6
0.4
/
/
02
.'
Second
derivative
"^ ^>
r
/
\
\
-1.2
10
12
Time t, sec
16
14
18
20
FIG. 13.18
first
is
negative
when
when
is
than unity.
If we add the second derivative to the step response, we reduce the overshoot and ringing. 11
Suppose a(0 is the step response and H(s) is the system function of the
greater than unity,
filter.
The
and
positive
is less
a1(0
= (0 +
L>
(13.87)
tW*)]
Xs
L
(13.88)
H(s)
s
11
F. F.
Theory, CI-9,
in Filters," Trans.
IRE on
Circuit
Network
394
1.2
/%
1.0
'
--.
^-.^
k
i.
0.8
lo.6
<E
0.4
= i
1.0
a>= *
1.5
a)
Jtt
0.2
y/f
Jl
12
10
Time t, sec
16
14
18
20
FIG. 13.19
From
s
Eq. 13.88
we
see that
axis at
-o
-2
-4
-6
-8
-10
.o
-o
-12
c -14
<3
_i 6
-18
-20
"
-22
-24
-26
-28
0.1
0.2
0.3
0.4
0.6
0.8 1.0
Radian frequency,
FIG. 13.20
1.5
2.0
3.0
Topics
in filter
395
design
will have negligible effect. Therefore the zeros should be added somewhere near the band edge. Figure 13.19 shows the effects of adding
zeros at <w = 1 (i.e., right at the band edge), and at m = 1.5. We see
that the further away the zeros are placed from the band edge, the less
they will have. The addition of the zeros will decrease the 3-decibel
bandwidth of the filter, however, as seen in Fig. 13.20. Therefore, a
effect
in
bandwidth and
reduction in overshoot.
13.7
we
Budak. 1 * The
will
mation.
*-
= pSr.
<
<. 1
(13.89)
and then approximate independently c-** and e~lk~lU with all-pole Bessel
polynomial approximations. Thus the resulting approximation for e~*
will have Bessel polynomials for both numerator and denominator. The
poles of the e~ * -1) ' approximant will be the zeros in the final approximant,
(
while the poles of the e~** approximant remain as poles in the final
11
A. Budak,
Trans,
of IEEE on
Circuit Theory,
CT-12, No.
2,
Network
396
anal/sis
and synthesis
A0
M(o>)
1.0
1.0
-0.6
T*
0.8
0.8
0.7^
-1.0
ft
0.6
0.6
(Bessel)"^
A=1.0_
0.4
0.4
k-
0,8-
02
km 0.7-j^
*-0.6tC
0.2
(a)
(b)
FIG.
As an example,
13.21
poles.
105
(ks)*
10(fcs)
+ 45(fcs)* +
[(fe
(13.90)
105
l)sf
6[(fc
l)s]
15[(ik
l)s]
(13.91)
15
-._ 7{[(fc -
15
g-tt-i).^
We
105(jfcs)
(ks)*
1)5]'
6[(k
10(fcs)
as a parameter.
A:
as a parameter.
lines,
radial lines.
Topics
in filter
design
397
l.l
1.0
k> 0.6-
0.9
0.8
*>0.7-
-km
).8
0.7
0.6
_*-1.0
(Bessel)
0.5
%
0.4
0.3
02
0.1
0.0
-0.1
-0,2
02
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
2.2
2.4
FIG. 13.22
13.8
or a transfer admittance
= -^a-
Yn(s) =
1
(13.93)
(13.94)
Network
398
analysis
and synthesis
Z21W
1.31s
1.359s
(13.95)
0.577
Z21
is
we
even,
s3
Thus we have
1.359*.
0.577
Za
s
Z S2
The
infinity is given in
Chapter
we
fraction expansion of l/ 22
1.31s 2
(13.96)
0.577
1.359s
filter
12.
1.359s
1.31s
We
must synthesize
to give the
it
+ 0.577 ( 1.415s
For the n
13.23.
= 3 Butterworth filter
Z2l(s)
we have
z n (s)
2s
2s
(13.97)
+ 2s-rl
,.
=
s
We
~ss +
2s
-t 7T7s
(13.98)
filter
shown
in Fig. 13.23.
(a)
Optimum
(b) Butterworth
filter
FIG. 13.23
filter
Topics in
filter
>ia v2
c7 :i
niters described in
respectively. 13
transfer
impedance
desired,
we
vice versa.
399
design
7) for
and Bessel
Z21(s)
shown
simply replace
all
in Fig. 13.24.
If a
shunt capacitors by
all
Y21(s)
realization is
series inductors
and
carry over.
In Chapter 14,
terminated
filters.
voltage-ratio
precisely the n
3 Butterworth function.
(s)
constant-resistance
in-
We
Some common
t (s)
2 (s).
1 3.27a
Za(s)
FIG. 13.25. n
= -1
13,
RjL
3 double-terminated Butterworth
(13.99)
Cs 2 + sR/L+(llLC)
filter.
400
Network
analysis
and synthesis
TABLE
13.7
Cx
Lt
c,
L*
Lt
1.000
0.707
1.414
0.500
1.333
1.500
0.383
1.082
1.577
0.309
0.894
1.382
1.694
1.545
0.259
0.758
1.202
1.553
1.759
1.553
0.222
0.656
1.055
1.397
1.659
1.799
C7
1.531
TABLE
1.558
13.8
Cx
L,
C,
L4
Cs
0.509
0.911
0.996
1.012
1.333
1.509
1.050
1.413
1.909
1.282
1.067
1.444
1.994
1.591
1.665
1.077
1.460
2.027
1.651
2.049
1.346
1.083
1.496
2.044
1.674
2.119
1.649
TABLE
1.712
13.9
Lt
c,
Z*
c.
1.000
0.333
1.000
0.167
0.480
0.833
0.100
0.290
0.463
0.710
0.067
0.195
0.310
0.422
0.623
6
7
0.048
0.140
0.106
0.225
0.301
0.382
0.170
0.229
0.283
0.036
A.
0.560
0.349
0.511
Topics
in filter design
401
Pentode
Interstage
Interstage
Interstage
network
network
network
FIG.
13.26.
We sec that Ztl(s) has a real zero and a pair of poles which may be complex depending
upon
the values of R, L,
transfer
all
the
filter
Cs +
(13.100)
1/RC
up
to this point
are
It
Z(s)
by
+ s + l)(s + 1)
1
s+
1
s* + s+ls + ls +
(s*
(13.101)
We
R-C
ci
ci ^*
.R
(o>
FIG.
13.27. (a)
Shunt-peaked interstage.
(b)
(f>)
R-C interstage.
Network
402
analysis
and synthesis
FIG.
Butterworth amplifier.
13.28.
13.9
In Section 13.8,
cutoff frequency of
we
1 Q. Filters designed
with these restrictions are considered to be normalized in both cutoff
frequency and impedance level. We will now discuss methods whereby
the normalized
cutoff frequency
and impedance
filters
level specifications.
(13.102)
sL
From
this
equation
we then
= sL =
a>o5 n
(13.103)
as
L-^
(13.104)
Similarly,
C,
1
sn
Cn
(13.105)
sC
"
Topics
C=
in filter
design
403
is
(13.106)
>o
Since resistances, ideally, are independent of frequency, they are unaffected by frequency normalization.
Consider,
impedance
impedance
next,
level
Z is
impedance denormalization.
Z
where R
Rn
is
(13.107)
R=
RoR
sL
is
(13.108)
relationship
= R^sL n)
(13.109)
is
L = R Ln
Similarly, for a capacitor
is
(13.110)
we have
_L
= Jk
sC
sC
(13.111)
is
C=
(13.112)
R = RoR n
C=
R^o
(13.113)
L = ^O^n
co a
Network
404
and synthesis
anal/sis
mh
66.7
Q>
^=0.3 nf
=4=0.1 nf
500 a
From the
original
V2
filter.
and Rt
Then
<o
10*
500.
R=
C,=
- 500
500*
i
500004)
0.1 /*f
(13.114)
L = 1(500) = 0.0067 h
10,000
C,
ST
4
500(10
The
final design is
13.10
shown
0.3^f
in Fig. 13.29.
FREQUENCY TRANSFORMATIONS
We
not by
as a
introducing new design procedures but through a technique
low-pass
normalized
frequency transformation, whereby, beginning from a
Using frequency transfilter, we can generate any other form of filter.
and band-elimination
filters.
will
remedy
known
since
Topics
in filter design
405
= a>
is
(13.115)
+ jm =
>o
ff +./'>
(13.116)
Since
axis,
in
how
maps
into theycu
co
(13.117)
a>.
mo ^
^ > as shown in Fig. 13.30.
Now let us see how the frequency transformations change the network
For convenience, let us denote the normalized low-pass network
elements with a subscript n, the high-pass elements with a subscript h, the
band-pass elements with a subscript b, and the band-elimination elements
with a subscript e. For the low-pass to high-pass case, let us first consider
the changes for the capacitor C. The transformation is given by the
elements.
ju
JVn
+j
plane
/<*>
* plane
<r*
-J
FIG.
13.30.
-Jox>
Network
406
equation
= LhS
Cs n
L n we
,
fo
(13.118)
C,
have
(13.119)
Chs
We
normalized low-pass
filter
Lh
(13.120)
<o
Cn
(o
Ln
Ch =
and
(13.121)
From
filter
with
pass
filter
level
of 500 Q.
its
From
the low-pass
High-pass
Low-pass
elements as
cu
let
filter,
Band-pass
Band-elimination
1*1
o-^fiflp-o
1{
Ca
1_
<>
Cbl
npnr- -H(BW
BW
2L
<">0
Cd = LnBW
r. BW
Lh woC
n
o-'lHRP-o
-
La
o-j^nnp-
Cc2
CnBW
Cb2=
FIG.
13.31.
-gyp
Topics
Ch =
1.5
x 10~ 9
in filter
407
design
Hf-
FIG.
high-pass-filter circuit
shown
in Fig. 13.32.
Its
filter,
10 (i)
1
(13.122)
son
4
10 (f)
Next, let us examine the low-pass to band-pass transformation (also
an L-C function):
S
where,
if
+ -)
= BW
S(\a>
s I
<
13123 >
m ct and wp, denote the upper and lower cutoff frequencies of the
filter, BWis the bandwidth
band-pass
BW^cocz-cOd
and
<o
is
the geometric
m =
co cl
Vo)c2>ci
The low-pass
to band-pass transformation
the segments
m^ ^
|a|
^ eu^,
(13.124)
shown
maps
(13.125)
the segment
in Fig. 13.33.
\a> n \
<,
to
The normalized
**.
bs + BWs
BW
(13.126)
408
Network
0
C1
+j
Plane
s plane
<Tn
-J
-<oo
FIG.
13.33.
We
LM
BW
BW
CM =
>0
(13.127)
Ln
LfiS
BW
(13.128)
C/h9
BW =6x10*
rad/sec.
We
(T\ isg
13.34.
in Fig. 13.23
filter
Li
FIG.
filter
into a band-pass
Band-pass
filter
shown
filter
in Fig. 13.34
by the
rules
Ct
nroip
1(-
d=c3
"=r-Ci
yCi
Sjij
i*
filter
v2
in Fig. 13.23.
Topics
given above.
The element
in filter
filter
design
409
following equations:
10
= 0.75
U 6x
x io*m)
*
**
x KT'h
(4
Ci-
X
X
10*
10-*f
12
-^ = ^xl(T
6
10*
9
(13.129)
X 10At)
(4
10*
6X104
=
(4
10-*f
32
= o.25
x l(T*h
X lO^f)
Q = 6x
= 0.25
X KT*f
10*
mation
s.
BW
Is
\O)
BW and
(13.130)
eu
S I
where
a>
are defined in a
je>
Jf>n
wo
-J
-o
-Wei
(a)
(b)
410
Network
analysis
and synthesis
element changes:
Ls
n
"
(s/L n BW)
ML BWs)
n
Ctls +
(1/L a s)
(13.131)
Cn s n
C nBW\a>
= Lsis
-\
C<gS
Observe that the normalized low-pass inductor goes into a paralleltuned circuit and the capacitor C goes into a series-tuned circuit, as
shown in Fig. 13.31. In Table 13.10, we have a composite summary of the
various transformations.
TABLE
13.10
Transformation
Equation
Low-Pass to
High-pass
Band-pass
Band-elimination
sn
BW
fe + ?)
Problems
13.1
figure.
Z =
What should L be
PROB.
13.1
Topics
13.2 Find the poles of system functions with n
Butterworth characteristics. (Do not use the tables.)
Show
13.3
response
is
at
m =
cosh
for e
/?*
in filter
3,
design
= 4,
41
and n
1.
filter specifications:
(a)
13.5
(a)
/(a>*)
(6)
/(a,*)
(c)
/(>)
at
to
= a>*
- iC(2a* = Ln(8).
13.6
1)
+J =
L4(co 2) and
3):
Q(.)
!(>*).
13.7 Expand cosh j and sinh s into power series and find the first four terms
of the continued fraction expansion of cosh */sinh s. Truncate the expansion at
= 4 and show that H(s)
KjB^s).
13.8 Synthesize the n
ated in a l-O load.
13.9
will
=
= 3 linear phase filter as a transfer impedance terminfilter,
which,
when terminated
PROB.
in
1-il resistor,
figure.
13.9.
Determine the asymptotic rate of falloff in the stop band of: (a)
filters; (b) linear phase filters.
13.11 Synthesize the n = 3 and n = 4 Butterworth responses as transfer
impedances terminated in a load of 600 SI with a cutoff frequency of 10* rad/sec.
13.12 Synthesize a Chebyshev low-pass filter to meet the following speci13.10
optimum
fications:
RL =
600
(a)
load
(b)
resistor,
(c) cutoff
frequency
(d) at 1.5
x 105
rad/sec
Network
412
13.13
analysis
Synthesize n
and synthesis
specifications:
(a) load resistor
(b) cutoff
13.14
10*
frequency
Design an n
Q
10* rad/sec.
=4
specifi-
cations:
(a)
impedance
(ft)
cutoff frequency
13.15
level
= 500 ft
= 10* rad/sec.
Synthesize a high-pass
filter
in a 10*-O load,
whose amplitude
frequency of
a>
characteristic is
10* rad/sec.
13.16
2 x
10*.
w-
chapter 14
The
14.1
INCIDENT
AND
scattering matrix
REFLECTED
POWER FLOW
we will devote our attention to certain power relationand two-port networks. The characterization of a network in
terms of power instead of the conventional voltage-current description is a
helpful analytical tool used by transmission engineers. It is especially
important in microwave transmission problems where circuits can no
longer be given in terms of lumped R, L, and C elements. In the powerflow description, we are concerned with the power into the network, which
we call the incident power and the power reflected back from the load,
which is the reflected power. A convenient description of the network in
terms of incident and reflected power is given by the scattering matrix,
which is the main topic of discussion in this chapter.
It is convenient to think of incident and reflected power when dealing
with transmission lines. Therefore, we will briefly review some concepts in
transmission line theory. For a more comprehensive treatment of transmission lines, the reader is referred to any standard text on wave propagation. 1 Consider the transmission line shown in Fig. 14.1. The voltage
at any point down the line is a function of x, the distance from the source.
The parameters which describe the transmission line are given in the
In this chapter,
ships in one-
following:
R=
G=
L=
C=
1
Prentice-Hall,
Jersey, 1950.
413
Network
414
anal/sis
and synthesis
l-
h:
FIG.
14.1.
Transmission
line.
Given these parameters, we can now define the impedance per unit
length as
Z = R+joL
(14.1)
Y=G+jcoC
The
characteristic
impedance
of the line
is
(14.2)
given in terms of
z = Vz/r
and the propagation constant
Z and
Fas
(14.3)
is
sfZY
(14.4)
equations for
these definitions in mind, let us turn to the general
line
the
x
down
point
any
the current and voltage at
With
. Vfi->* + Vsyx
I(x) = !*" - iy
V{x)
(14.5)
the
/ refer to the incident wave at point * and
14.5
Eqs.
Solving
x.
wave
at
reflected
the
to
terms with subscript r refer
and reflected
simultaneously, we obtain explicit expressions for the incident
subscript
waves
^*[K(*) + Z,/(*)]
Vre*x = h[V(x)-Z
V(L)
ZtfQ
(146)
l(x)]
is
terminated in
(14.7)
The
Then we
wave
scattering matrix
415
is zero.
f>L
(14.8)
yL
I4J
For the one-port network shown in Fig. 14.2a, consider the following
The incident parameter a is defined as
definitions.
(14.9)
and the
(14.10)
where R^
is
an
arbitrary, positive,
a
Similarly,
Ve~T"
- ^=-
(14.11)
reflected
wave as
= ^=r
(14.12)
VZ
/
One-port
network
(a)
a *
,
One-port
network
Network
416
Thus we
reflected
To
analysis
and synthesis
wave
give further
meaning to a and
b,
one-port network
P = iRe
VI*
(14.13)
where /* denotes the complex conjugate of /. From Eqs. 14.9 and 14.10
we solve for V and / in terms of the incident and reflected parameters to
give
K=(a + b)V
dissipated
(1414)
P = Haa* -
is
bb*)
(14.15)
-KM* -1*1*)
where, again, the asterisk denotes complex conjugate. The term Jaa*
can be interpreted as the power incident, while $bb* can be regarded as the
power reflected. The difference yields the power dissipated by the one-port
network.
The incident parameter a and the reflected parameter b are related by the
equation
where
S is
coefficient.
= Sa
(14.16)
From
the definitions of a
substitution:
Solving for S,
where
we obtain
= Z ~ **
Z+R
one-port network
Z=
excited
(14.19)
(14.18)
by a voltage source
set
equal to the
V9 with a source
The
we
when we choose the
impedance Rq to be equal to
show
reference
AAA
"VW
that
One-port
network
!>
-&
The proof
By
follows.
incident parameter a
is
(,4 20>
-
definition, the
FIG. 14.3
given as
^)
-K+^)
From
Fig. 14.3,
we have
V,
and
-IRB
we
(14.21)
=V
(14.22)
(14.23)
Ra
417
scattering matrix
+Z
obtain
_
_
VBR,
17.,
2V^
RqV,_~\
\
,
r r,+z
^+ zJ
(14.24)
R' + zJ
2VV
\a\*
2 \
becomes
|a|V
(14.26)
=
When we
ance,
i.e.,
l^- (i-isi)
resist-
n _ nil!
~Z~ ~ on
2
ePA. ~
iR
(14.27)
Network
418
analysis
and synthesis
**
PA
where
~vVA
Av
>
^
*"
^*^
available
network
FIG. 14.4
Z is
source/?,.
power
dissipated in
Pa
with
Z = Rt
network shown in
is
= ^-
The
(14.28)
maximum
available
power
at the
From this discussion, it is apparent that the value of the reference impedance Rq should be chosen equal to the source impedance R. A
standard procedure is to assume a 1-Q source impedance and denormalize
when
necessary, that
is, let
where
Next,
us
let
(14.29)
(14.30)
briefly consider
scattering parameter
1.
- 2^-i
z + 1
S for
axis
is
always
less
or equal to unity
is,
|S(/>)| <. 1
(14.31)
This property follows from the fact that the power dissipated in a passive
network is always greater or equal to zero. Since the power can be
expressed as
P-^fl-ISft^O
we
see that
2.
For a
\S(ja>)\*
(14.33)
reactive
(14.32)
The
3. For an open circuit 5=1, and for a short
shown to be true from the equation
S
For an open
therefore
Before
5=
circuit z
oo,
circuit
S=
1.
= ^|
z + 1
so that
419
scattering matrix
This
is
(14.34)
5=1.
For a short
circuit z
= 0;
1.
we proceed
definition.
DEFINITION.
is
ditions
|^)|
()
(6) F(s)
In
^ AT
is real
Rea^O
for
when
s is real.
(a),
4. If z
Z/Rq
a positive
is
is
a bounded
real
function.
S
From
the equation,
= (z-
l)/(z
Re z(s)
1).
^ 0, when Re s ^ 0,
we
see
that
s
so that
when Re j
|S(s)|
;> 0. (b)
Im z(s)-[l- Re z(s)]
jlmz(s)+ [I + Rez(s)]
j
+
= ( Im'^
llm , z(s) +
[l
Re2(S)]/
'
3 f
U-Re^ *
(14.35)
l)/(2
1)
network
is
420
Network
and synthesis
anal/sis
-o
lo>-
-*o2
Ol
o-
02
v2
Two-port
network
Vi
*i
62
-o
FIG.
network, that
D2'
l'o-
14.5. Scattering
o-
is,
(14.37)
"
=
l(vfe
where
R01
and
i?
+vr"'')
and output
ports respectively.
The
scattering parameters
Stj for
equations
bt
(14.38)
iSjitfi
Si2a2
6i
Si1
^im
(14.39)
(14.40)
is
From
Eqs. 14.38,
Sn
S
Si.
=h
a* 010
SM
=
a* ai0
(14.41)
The
scattering matrix
5M
is
Su
is
and SM
coefficient;
is
421
Slt
is
Observe that
all
to zero in the defining relations in Eqs. 14.41. Let us see what the conimplies in the definition for the forward reflection coefficient
dition a4
tlt^O
Figure 14.6 shows the terminating section of the two-port network of Fig.
14.5 with the parameters a, and b a of the 2-2' port shown. If we treat
the load resistor
where Rgx
is
(14.42)
Ro
and b% are
related
by*
aa
= SA
is set
(14.43)
then
S,=
Rot
Rat
Ri*8 =
+ Rto*
(14.44)
From
reflected
is
b, and the
Network
422
impedance
and synthesis
anal/sis
R01 =
We see
(i.e.,
conditions a x
/?i).
and aa
respectively.
Next,
let
ances at ports
reflection coefficients
Let us
(14.45)
-M
a
the equation
5U and Su
and 2 as
h
From
(14.46)
l lai-0
We can write
=
*K W*oi)
which reduces
Similarly,
easily to
Sn
we have
These expressions
tell
us
if
Rqi
Z + Rqx
z Rq
Z + R oi
(14.47)
Rnli)
2q
(14.48)
Rt-Rot
h,_h
we choose
(14.49)
is
terminated
321
-*l
(14.50)
a
As we have just
JtogaJfa
FIG. 14.7
The
then
we can
express
ax as
< 14- 5I >
""vfe
Since a,
423
scattering matrix
y
^
=
bt
Consequently,
>/*'
(14.53)
= "(-/= - V*z
(14.54)
v*
Finally,
we can
vjj**_
(14.55)
Krl^ ^* Ri-fi.i,Ri-Boi
In similar fashion, we find that when port 1 is terminated in Rqi = R t and
when a voltage source V,t with source impedance Rt is connected to port 2,
then
"-
?MF
(1456)
We see that both Slt and SM have the dimensions of a voltage-ratio transfer
function. Indeed, if
It is
R^ =
R^, then
ratio
'Vi-jtV*
A---/,
n
(14.57)
-^-
(14.58)
Network
424
anal/sis
and synthesis
_Jf_
1:1|
2o
+
V2
Ideal transformer
(a)
FIG.
14.8.
From Eq.
14.48,
we have
we
^ = n*
Z1 =
so that
did for
Su
Sn
(14.59)
= n*-l
n* + l
1-1' port in
a 1-G
(14.60)
We obtain,
SM
= (1/w*)
(!/*)
-1_1+1 1+
n*
(14.61)
The
425
scattering matrix
= Q at the 1-1' port and terminating the 2-2' port with a resistance
- * the equivalent circuit
jRo, = 1 ii, as seen in Fig. 14.8rf. Since V^h
jRo!
of the ideal transformer as seen from the voltage source as a 1-Ii impedance in series with an *-ohm resistance (Fig. 14.8c). Then Vx can be
expressed in terms of Vtl , as
K,=
Since
Va =
(14.62)
n8
Vjn, we have
Kin
Since
R^ = R^ =
Q,
(14.63)
n*
SM is
2F,
2it
n*
r i
(14.64)
l
(14.65)
1
v-1
n*
As a second example,
let
shown in
Fig. 14.9. If we
given as
In
n*
(14.66)
l-n
In
M*
is
fl*
+l
terminated in
assume that
R^ =
!*
coefficients are
= ?oZlZo _ Q
z, + z
(14.67)
x^-
Zo
1'
FIG,
14.9. Lossless
transmission line.
Network
426
This result
is
characteristic
then
From
the equation
Z> x
v,
= *>- ri
a2
=
(14.68)
~ 2 \*J vgl
Sn - 2e-*
(14.69)
Stl
we obtain
we
In similar fashion,
(14.70)
find that
Sit
2e- rL
(14.71)
14.4
is
2-" z, i
(14.72)
VW-
let
From
scattering
in
power delivered to
P=
all
we
Hii*
a*fh*
W)
(14.73)
Equation 14.73 follows from the fact that the power delivered to the 1-1'
port
is
Pi
KW -
The
total
power
i(^*
(14.74)
bib,*)
is
W)
is
(14.75)
then
P = P1 + Pt
which is exactly the expression in Eq.
delivered to the network is
(14.76)
P - Hla'fW ~
lb*flb]}
power
(14.77)
The
where
lb]
[S][a],
then
[b*]
now
IP
=
-
[a*]
427
and
-a
-a
[a]
Since [b]
scattering matrix
(14.78)
T [S*] T
(14.79)
be rewritten as
{[a*]
T [a]
la*] llu]
Dct
[[]
(14.80)
[[]
[5*] T [S]]
[S*] [S]]
(14.81)
[S*] T [S]
[u]
(14.82)
For a
lossless
two-port network
From
this equation,
we have
matrix
Sj*Su
+ Stl *Stl =
+ Sn'Su =
Sn *Si,
+ Sn *S =
Su*Su
Su*S11
14.85
is reciprocal,
and
then
S*S
(14.84)
(14.85)
(14.86)
(14.87)
If the
+
IVHI +
|Sii(/>)|
=
|5(/))| -
|S(/>)|
(14.88)
1
428
Network
Two-port
network
<*2
FIG. 14.10
from which
it
\SJ(ja>)\ <. 1
(i.e.,
=
=
and \Su(ja>)\
when
1.
This
network
from port
1-1'
is
reflected
scattering formalism.
Many
For example, an
VSWR =
iM
(14.89)
is
iso)i*-*
pA
< 14 - 90>
Section 14.5.
The
429
scattering matrix
INSERTION LOSS
14.5
In Section 14.4,
coefficients in
Stl
in the two-port
14.5,
From
we
the equation
obtain
l-S^OOl*
= Sn(/w)Sa *(/a>)
|M-)|-
(14.92)
l^>
I^iO)IV8*x
(14.93)
Pal
where Pt is the power dissipated by the load Rt, and
gain of the generator Vgl Similarly, we have
PAl is
the available
|Si.O)l*
We
relate the
power dissipated
and
- -
ISutyco)!* are
(14.94)
power
transfer ratios
power
which
in the
other port.
Now
let
shown in
IVR
^V
(14.95)
is
to measure the
430
Network
analysis
and synthesis
-J
vW^-
r
(a)
Hi
-i_
r
Inserted
)yi
*2* >V2
two-port
network
r
lb)
FIG. 14.11
If
we
_ Ew
a
where a
is
(14.96)
we
=10 log
obtain
(14.97)
Then
Pm is
|K|
dissipated
(14.98)
2R t
*.
2(R t
The power
R>
by the load
ir.il'
(14.99)
RJ*
network
is
given by
P.= Hi!!
2R t
(14.100)
The
The
insertion
power
ratio
2a
when
431
_ wr
P.-VXOk +
In the special case
scattering matrix
<14101)
is,
Rx = R2 = Rn = R^
(14.102)
When Rt
j&
Rt
|S*iO)|
,
then
PjQ
4R1 R 2
(*i
In any event,
i\Sa (jai)\ t
power
we
(14.103)
1
2
R*) |s*iO)l
(14.104)
insertion
14.6
ratios.
In this section,
by Darlington in a
Fig. 14.12.
FIG. 14.12
S. Darlington, "Synthesis
Loss Characteristics,"
/.
Network
432
is,
20
(14.105)
|S21 (>>)| a
P*
power
ratio is
A =10 log %*
P
-101og|S 21(| a
db
(14.106)
insertion loss
to the specification
of the
14.13ft.
amplitude-
ingenious techniques given in Darlington's synthesis procedure is the reduction of insertion loss synthesis to an equivalent L-C driving-point
synthesis problem. This technique can be developed in terms of scattering
parameters. Our initial specification is in terms of |Su (/<u)|. For an L-C
two-port network
\Sn(jcoW
Next,
Sn(s)
is
the equation
Su
=1-
\S ia (Jo>)\
=
+
^i
we
|S*iO)l
(14.107)
Su (s)Sii(- s)
Then from
=1-
2
\
im ^,
(14.108)
(14.109)
Z^s)
=R
+ Su(s)
- Su (s)
(14.110)
in Fig. 14.12.
will restrict
Chebyshev
|S21
2
,
that
=
0)f _
1 +
is,
(14.111)
,2n
CO
The
10
scattering matrix
433
Li
/-
~*np
T-nnnp
c2 z
T-
C-1-
FIG.
14.14.
where
10.
Zi(s)
niters.
C B8(<u)
v2
(14.112)
Example
14.1.
filter
._.
"*W-TT^
(14.113)
IWI* = !--*
1
Lettingy'eo
= s in
|Su
(/a>)| a
we
(14.114)
+0)6
obtain
<Sii(>S_(-*)
= 1
-s*
(14.115)
8
A-*
)
"
g +2> +
^ + ^ 1^ +2j2 _ ^
_______
(14.116)
a*
sll(f) =
(14 117)
.
+ Su(s)
- SuM
2s*
+2s* +25
2s?
2s
(14.118)
1
Network
434
FIG. 14.15
+ 2* + l)^ + 2s + 2s +
l(s
2s*+2s*+s
s
1)2**
2s*
+2s +
l{2s
Zs
1)5
The low-pass
filter is
An equivalent
we
Su(5)
= YM - G
G = 1 mho
1
1
14.2,
and
14.3
(14.119)
Yt(s) + G
Y1(s) =
The canonical
1(5
Then, assuming
+ Sn (s)
- Su(s)
shown
(14.120)
in Fig. 14.16.
Tables 14.1,
7) for double-terminated
FIG. 14.16.
and
vice versa.
filters
and
14.3.
The
TABLE
scattering matrix
435
14.1
Ci
2.000
Lt
C3
Lf
Lg
Cj
C?
1.414
1.414
1.000
2.000
0.765
1.848
1.848
0.618
1.618
2.000
1.618
0.618
0.S18
1.414
1.932
1.932
1.414
0.518
0.44S
1.248
1.802
2.000
1.802
1.248
1.000
0.765
TABLE
0.445
14.2
71
Cx
L*
C8
2.024
c5
L*
1.018
2.024
0.994
2.135
1.091
3.001
1.091
2.135
2.167
1.112
3.094
1.174
3.094
TABLE
J-6
c7
1.112
2.167
14.3
C3
La
Cj
C3
Lt
Le
Cj
2.000
1.577
0.423
1.255
0.553
0.192
1.060
0.512
0.318
0.110
0.930
0.458
0.331
0.209
6
7
0.838
0.412
0.316
0.236
0.148
0.051
0.768
0.374
0.294
0.238
0.178
0.110
is
0.072
0.038
Chebyshev niters
filters do not
book. 8
New York,
'Ibid.,
1962, p. 589.
Chapter
13.
and
Synthesis,
Network
436
anal/sis
and synthesis
Problems
14.1 Determine the reflection coefficient
in the figure.
S for
shown
?o
-VWrvW
4=tt'
w
For the one-port_network in Fig. 14.3, let iJ, . R
If the incident
B
is a = VJl^R^ find the reflected parameter
A.
14.3 For the network in Prob. 14.1, determine
\S(jio)\.
Show that the
scattering elements S for the networks in Prob. 14.1 are
bounded real functions.
14.4 For each of the networks shown, find the
scattering matrix for ik, =
14.2
parameter
-o2
(a)
(b)
+
Vi
/2-*-
^-J-?-
Gyrator
/i
A
V
Negative
impedance
converter
*!,
^ W*
*1
Vi
w
PROB.
14.4
1.
z
T
w
v2
The
scattering matrix
437
14J5 Find the insertion voltage ratio and insertion power ratios for each of
the networks shown. These networks are to be inserted between a source
impedance R = 2 ft and a load impedance RL = 1 ft. From the insertion
power
2h_
i
2_h
o-'TRRTH-^WnP-02
Jsjf
io
_lh_
.
nptpP 1
=f=if
=r=i'
l'o-
(6)
2h
40
:*f
PROB.
14.6
(a)
|SaQ'>)l*
l-W)!
=
1
+0)*
+afl
14.7
at
14.5
chapter 15
Computer techniques
in
circuit analysis
15.1
IN
CIRCUIT ANALYSIS
The advent of the high-speed computer has made
We
iterative design.
difficult
In succeeding sections,
we
will
discuss
some
specific
circuit-analysis
computer programs.
Circuit analysis
The primary objective of a linear circuit-analysis program is to obtain
responses to prescribed excitation signals. These programs are based on
many different methods: nodal analysis, mesh analysis, topological
formulas, and state variables. Most of them can handle active elements
such as transistors and diodes by means of equivalent circuit models.
The
state-variable
matrix formula-
domain
via numerical
T. R. Bashkow,
3,
Computer techniques
The majority of
circuit-analysis
in circuit anal/sis
439
the poles and zeros of these functions, and can also provide transient
response and steady-state response, if desired. With versatile inputoutput equipment, the output can also provide a schematic of the original
440
FIG.
Network
15.3.
analysis
and synthesis
filter
evaluation by Laplace
transform.
high-order polynomials.
list
of data points at
the input.
Computer techniques
In Section 15.2
we
examine further
will
in circuit analysis
details
441
of a typical steady-state
analysis program.
Circuit
The
(filter)
filter
design
for the use of computers in circuit design. Designing insertion loss filters"
to meet certain amplitude requirements requires considerable numerical
calculation even in the simplest cases.
insertion loss
filter
design
is clearly
The use of
digital
logical alternative.
computers in
The amount of
program
is
bands.
In the specifications, the designer could specify both the zeros of
4
If
transmission (loss peaks) and the network configuration desired.
his specifications include neither, the computer is free to pick both configuration and zeros of transmission. The computer's choice is one in
which the inductance values are kept at a minimum. The program was
written so that the same network could be synthesized from both ends.
Finally, the computer prints out the network configuration, its dual,
the normalized element values, and the denormalized ones. It also
provides information such as amplitude and phase response, as well as
plots of these responses obtained from a microfilm printer.
Figures 15.4, 15.5, 15.6, 15.7, 15.8, and 15.9 show the results of a
band-pass
filter
"On
December
by
Figure 15.4
Synthesis," Trans.
IRE on
1958, 287-327.
cit.
Network
442
analysis
and synthesis
CASE NUMBER
3.1
OEGREE OF FILTER
MULTIPLICITY OF PEAK AT ZERO
MULTIPLICITY 0F PEAK AT INFINITY
NUMBER OF FINITE PEAKS BELBM THE BANO
NUMBER #F FINITE PEAKS ABBVE THE BANO
EQUAL RIPPLE PASS BANO REQUESTED
PASS BANO RIPPLE MAGNITUDE
L0HER PASS BANO E06E FREQUENCY
MID-BAND FREQUENCY
UPPER PASS BANO EDGE FREQUENCY
13
3
2
1
0.09000
DB.
1.0000000E+04 CPS.
1.3416408E*04 CPS.
i.80oooooe*04 cps.
l.OOOOOOOE+00
MA
TERMINATIONS
INPUT TERMINATION
OUTPUT TERMINATION
6.0000000E02 OHMS
0.
OHMS
VALUE OF M
0.6229266
VALUE OF H
2.3788111
1.9296560
1.4968796
is
equal ripple with ripple magnitude of 0.05 db, and the degree of the
is
As we
to be 13.
to be
filter
filters
cycles;
Vl.8 X
10*
1.3416408
zeros of transmission
10* cycles;
at/=
10* cycles.
(three),
/=
oo (two),
and four
finite
zeros
of transmission: one below the pass band and three above the pass band.
The positions of these finite zeros of transmission are chosen by the
designer as indicated by the notation "arbitrary" stop band requested.
Computer techniques
CASE NUMBER
FMUtD
1.1
OtMEE OF FILTER
13
MM
1
2
0.0500
ARBITRARY' STEP
DB
cps
1.3416408E*04 CPS
i.aooooooE04
100
300
NORMAL IZEO
UNNORHALIZEO
1.0000000E*00
6.0000000E*02
3.0T04337E-01
6.0706102E-09
S.9892B16E-03
3.85TA228E-09
2.9723019E00
9. 089744 OE-08
1.0083380E00
4.44298O6E00
..L.e..:
7.1783955E-03
9.1797212E-08
TERMINATION
803
PEAK FREQUENCY
3.5200000E+04
901
PEAK FREQUENCY
4.2000000E03
801
2.6982703E-08
1.364T491E00
L C
9.3749272E-01
7.S1321S3E-01
7.9423894E-01
1.44M444E*00
i""c
2.T112T46E-01
t.2441S6*E00
3.S2A810TE-03
1.9447656E-0B
PEAK FREQUENCY
2.0700000E04
1.4931737E-08
100
2.8S46S14E-08
802
1.429T93TE-03
2.499S449E-08
PEAK FREQUENCY
2. 310000 0E04
1.0127M7E01
2.0022669E-07
400
9.97702ME-02
.0165648E-04
200
100
l.A3787S9EQ0
3.2382 800E-08
4.2943194E-01
4.4B148I6E-03
300
TERMINATION
SHORT
SNORT
(AND)
- i.oooooooso* cps
L*"e
443
MULTIPLICITY OF PEAK AT
MULTIPLICITY OF MAX AT INFINITY -
T.8921337E-01
1.8499769E-0I
in circuit analysis
FIG. 15.5
The terminations
filter
are:
input
60 Q, output
= 0Q
configuration
inductance configuration. 8
W.
Saraga,
"Minimum
Network
444
FREQUENCY
IN CVCt.ES
16000
16250
16300
16730
17000
17250
17500
17750
10000
10250
16500
16750
19000
19250
19500
19750
20000
20250
20500
20750
21000
21250
21500
21750
22000
22250
22500
22750
25000
25250
anal/sis
and synthesis
VOLTAGE RA
TIO IN 08
IN OEGREES
5.716532
5.704379
5.707644
5.729113
5.750516
5.743140
5.708979
5.720983
5.702947
4.131556
-1.119811
-7.753694
-14.148566
-20.210264
-26.118735
-32.100833
-38.473164
-45.657895
-56.325030
-71.701988
-59.444189
-57.483949
-57.567501
-58.703992
-60.546362
-63.101161
-66.602572
-71.867460
-63.403713
-80.597817
145.549717
128.958686
111.534102
92.934806
72.763652
50.659800
26.219204
356.194819
322.320795
272.913465
225.689589
195.922988
177.085862
163.748654
153.507803
145.227029
138.296917
132.354567
127.166241
302.572268
298.458582
294.740796
291.334706
288.250341
285.388013
282.735723
280.267304
277.961066
275.798866
93.765362
PHASE
NAG
REAL I IN
IN 8HNS
2 IN
IN 8HNS
134.691
130.664
125.883
119.931
112.687
104.489
95.717
REAL V IN
IN
56.981
61.193
64.762
67.928
71.154
74.696
77.919
79.138
78.767
87.140
111.450
139.624
164.004
184.583
202.605
218.952
234.147
248.501
262.211
275.407
286.178
300.592
312.696
324.536
336.138
347.527
358.726
369.753
380.621
391.345
65. 165
67.626
40.032
16.386
3.462
1.725
.541
.167
.049
.013
.002
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
NILNMS
6.2973
6.2766
6.2814
6.3129
6.3445
6.3337
6.2835
6.3011
6.2748
4.3533
1.2913
-2797
.0641
.0158
.0040
.0010
.0002
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0000
INA6 V IN
IN NILNHOS
-2.6641
-2.9395
-3.2315
-3.5756
-4.0061
-4.5277
-5.1151
-5.8552
-7.3083
-9.4759
-6.7826
-7.1517
-6.0967
-5.4176
-4.9357
-4.5672
-4.2708
-4.0241
-3.8137
-3.6310
-3.4701
-3.3268
-3.1980
-3.0813
-2.9750
-2.8775
-2.7876
-2.7045
-2.6273
-2.5553
FIG. 15.6
20
(
1
-20
-40
-60
-80
f\
1
/
/
'
-100
-120
0.000
0.400
Add 1.00000
0.800
1.200
1.600
2.000
03 to abscissa
Frequency, cycles
2.400
FIG. 15.7
10*
2.800
3.200
3.600
4.000
Computer techniques
445
in circuit analysis
--
JE
;
J
-1
0.000
0.400
Add 1.00000S
0.800
1.200
2.000
1.600
03 to abscissa
Fl6queflcy
2.800
2.400
3.200
3.600
4.000
x ,
FIG. 15.8
0.800
0.600
0.400
0.200
0.000
(*"
-0.200
-0.400
--
-0.600
'*
\
1,
-0.800
u
-1.000
0.000
0.400
Add 1.00000E
0.800
1.200
+ 03 to abscissa
1.600
2.000
}nmm
FIG. 15.9
2.400
10
2.800
3.200
3.600
4.000
Network
446
The peak
anal/sis
and synthesis
the
circuits are
left.
Figure 15.6
cycles
Frequency
(a)
filters
Computer techniques
447
in circuit anal/sis
and
at
Optimization
Network design cannot always be accomplished by way of
analytical
means.
rapidly, if
+ T-* +
mdf=T-*i
OX
OXi
---
+ ^*n
OX
6xt
(15.D
is
=-C
(15.2)
OXi
where
is
a constant. After
all
An
new
excellent paper
the
optimum
is
obtained.
York, 1956.
7
C. L. Semmelman, "Experience with a Steepest Descent Computer Program for
Designing Delay Networks," IRE International Convention Hec, Part 2, 1962, 206-210.
448
Network
analysis
and synthesis
xi so
<*,)=i[K,-n*<,/,)]
(i5.3)
i-X
Machine-aided design
The concept of real-time
interaction between
in the field of
MAC
MAC
MAC
Circuit
Utility
Approach,"
IEEE
Computer techniques
in circuit analysis
Remote
Data
Memory
449
terminal
bank
link
Memory
Central
bank
processor
Buffer
Data
memory
link
Remote
terminal
Queueing
Data
logic
link
Remote
terminal
Memory
bank
Memory
bank
Data
Remote
terminal
link
FIG.
15.1 1.
program into the central processor again. The queueing time and processing time on the multiple-access computer may amount to only a
minute or two, which, for the user, is probably not significantly long.
Dr. H. C. So has written a paper on a hybrid description of a linear
10
in which he has shown that the hybrid matrix is ideally
ii-port network
suited for such problems such as multielement variation studies and
iterative design. Suppose there are n variable elements in the network.
These can be "extracted" and the rest the bulk of the network can be
'
Network
450
analysis
and synthesis
Command
Initialize
i
1
Frequency
range
Engineer
Control
Computer
*j
Parameter
adjust
'
Display
Output
some
is
repeated a
number of
times until the engineer has obtained the desired results. Such a program
if
the
Purpose
Our purpose
is
to
rational
function
"(s) - ?Hr?
1
(154)
Computer techniques
in circuit anal/sis
451
over a set of frequencies (o^ In Eq. 15.4, C and C are real constants,
d
and E(s) and F(s) are polynomials in s = a +ja>. The program described
here computes the amplitudes M(co,) and phase #a> ) over a set of
t
frequencies cok where
,
JU
W =
\CnE(jlOh)\
(15.5)
\Ca FOcok)\
and
<f>((ok)
= arctan
arctan
ft
Fj(ja> k)
(15.6)
real
respectively.
<Ko>k)
<f>
*s
and
subscripts
and imaginary
Note that in Eq.
limited to the
<,it radians.
range
parts,
15.6,
it
<>
In order to circumvent
-JW
FIG. 15.13. Calculation of magnitude and phase at a> ( due to zero z.
and phase.
Method used
Let H(s) be factored into poles and zeros such that
H(s)
= C n (s - z )(s - zt)
Cd(s ~ Pi)(s - P)
t
(S-PJ
(15.7)
Consider the amplitude and phase due to any pole or zero, for example,
is
{a>,)
4>.{a>d
arctan
(<<
- ft*}*
ft,
'~
(15.8)
"
l
overall function
(15.9)
is
M(wJ =
(15.10)
QIT^M)
and
<K>i)
= 2 tJfOi) - j ,,(>,)
(15.11)
452
Network
zero, respectively.
Input
1. The numerator E(s) and denominator F(s) can be read in either as
polynomials (high degree to low) or in terms of their roots. If numerator
and/or denominator are read in as polynomials, their roots will be printed.
only the
minimum
if equally
spaced,
in.
NN
Read CN,
KK =
Test
Read
L1
CD
KK
KK=1
Read WMIN,
(I)
=
Test LI
Ll
DW
Read E(l)
J
L2
=
TestL2
L2=l
Read
F(l)
*~r*
To main program
FIG.
15.14.
Computer techniques
453
in circuit analysis
Fig. 15.14.
Def-
Nl
We
then read in
all
to,.
LI
=0
Read
read
L2
=
=
Read
Read
read
in zeros (if
in).
in).
Output
The main output of
15.3
sections, beginning
The program
bining R, L,
calculates the
series
Network
454
vm
Zm
v -l
Zn
la"
Vo
Vl
l
Zi
1
1
Zn-l
|
|
1
Ym
Yi
r-i
Y
l
FIG. 15.15
The network
Fig. 15.15.
is
These
section to
and current
computed by the equations
section, the voltage
^n =
network are
(15.12)
^n^n
+ Vn-1
section to
make
its
calculations.
and current
V and
/ allow
branch impedances of the section are required. To simplify the preparation of the input data specifying these impedances, the following procedure
is used. Each branch impedance is formed by the addition in series or
parallel of basic R, L, C impedance arms. A basic R, L, C impedance
arm is a series combination of a resistance, an inductance, and a capacany or all of which may be absent. The impedance arms are speciitor
fied by the element values R, L, C and an instruction indicating how the
arm is to be combined.
The elements of the impedance arms are frequency and impedance
normalized to aid computation, and then the arms are combined as
specified to form the appropriate branch impedance.
The roots of the voltage and .current polynomials for each L section
are
computed by a root-finding
routine.
Computer techniques
455
in circuit analysis
For the normal load and open circuit termination, the frequency
responses of the input impedance, and voltage-transfer ratio are computed
for short-circuit termination, the input impedance and current gain are
computed. Provision is made for either a linear or logarithmic increment
Program operation
The ladder network-analysis program requires two sets of input data.
The first set consists of the control instructions. These tell the computer
which of the various options, such as information concerning polynomial
roots or frequency responses, are to be exercised.
In addition, the control instructions provide the computer with necessary
parameters such as normalizing factors, and the number of L sections in
the ladder. The second set of data determines the branch impedances of
the network by specifying the R, L, C impedance arms and their combining
instructions.
Impedance data
The data specifying
The
first
combining instruction. The branch impedance is set equal to the impedance of this arm. Subsequent impedance arms are added in parallel
or series with the existing branch impedance, according to the combining
instructions of the arms. A blank card tells the computer that a complete
branch impedance has been formed and that the next arm begins a new
branch impedance. If blank cards are not properly inserted, the computer
will calculate
The order of
^a
1/
Jg,
struction of
impedances.
For example, suppose the series combination of two tank circuits (Fig.
15.16) is desired as a branch impedance, this impedance may only be
Network
456
anal/sis
and synthesis
50m*f
lMf
2mh
.05 h
-^OOOy-
Tank
circuit
Tank
circuit
XIN
5.00000
1.0
2.00000E
2.0
3.0
11
- 03
.05
.000001
2.0
4.0
Card
Blank
and an additional
Tank
circuit
more
instructions
(XIN
3.0, 4.0,
and
5.0)
set
A is
XIN
Blank card
Operation
A complete
the next
arm
1.0
2.0
This
3.0
4.0
Add
branch impedance.
5.0
Computer techniques
HWV-i
in circuit analysis
457
r-VW-i
-Wrr-VArn
[-VS/Sr-j
L-vs/Sr-l
FIG.
15.17.
Output
coefficients of the voltage and current polynomials are printed in
frequency normalized form. To calculate the denormalized coefficients,
divide by the normalizing frequency raised to the power of the corre-
The
sponding exponent.
The
poles
15.4
Two double precision Fortran programs have been written that help
in the synthesis of double-terminated filters using the Darlington procedure. The mathematics of the programs is described here.
Given a forward transmission
coefficient
S8i(s)
KN(s)
D(s)
(15.13)
Network
458
we
analysis
and synthesis
Su(s) Sn(-s)
D(s)
Su(s) from
the equation
S(s) S gl (-5)
D(-s)
K* N(s) N(-s)
(15.14)
D(-s)
D(s)
The denominator of Su(j) is D(s) because all the poles of Su(s) must be
in the left-half plane. The zeros of Sn(s) are not so restricted. If Sn(s)
Su(s) has zeros &tajb and a jb, the zeros of Sn(s) can be chosen
as either a jb or a jb. The only difference is that certain choices
lead to niters with unity-coupled coils and others without. 12
Once the zeros of Sn(s) are chosen, the input impedance of the filter
is
then
Zi(s)
1
1
ReadCN
~
+
Sii(*)
(15.15)
Su(s)
Numerator
multiplier
I
Read
CD
Denominator
Number
ReadLN
I
Read
A (I),
B(I),N(I)
Read LD
multiplier
of zeros
Read
real
and imaginary
N (I)
Number of poles
(conjugate poles
count as one pole)
Read
H(I),G(I), M(l)
Read
real
and imaginary
To program
FIG.
11
15.18. Input to
T. Fujisawa, "Realizability
Sn(5)
Theorem
5u (j) program.
Low Pass
on Circuit Theory, CT-2, December
IRE
Computer techniques
459
in circuit anal/sis
Program descriptions
We
K* in Eq. 15.14.
The second program performs
1.0 so
A(I) +y'B(I).
C(I) +y'B(I)
denominator polynomials of
The
Sxl(s) may
be represented as
Su (s)
= E(s)
(15.16)
F(s)
ReadLD
Read
ReadLN
Read
of pole pairs of
Su
of
Sn
and
multiplicity,
M (I)
Number
of right-hand plane
zeros of
Sn (a) Sn(-s)
Number
(real poles
H(I).G(I),M(I)
N 0)
I
Read SO)
Multiplicative constants
(i 1.0) (each
combination of zeros has LN cards)
To program
FIG.
and
Su(s)
15.19.
Input to
Zta(j) program.
460
Network
analysis
and synthesis
ZM= i^ = m^m
The input
instructions are
(15 17)
.
in Fig. 15.19.
Problems
15.1
Organize a flow-chart for computing magnitude and phase given only
the unfactored numerator and denominator polynomials. Do not use a rootfinding subroutine.
15.2
_,.
= 0,
1, 2,
3(5
1)
10.
Repeat Prob. 15.2 for a general transfer function given in terms of its
unfactored numerator and denominator polynomials. The frequency points co t
are to be read in by the computer.
15.4 Suppose you have calculated the phase of a transfer function at points
m = 0, 1, 2, .... 50. Devise an algorithm to test for phase linearity. The
deviation from phase linearity is to be called phase runoff.
15.5 Write a program to analyze a two-mesh network made up of only
15.3
R, L, and
15.6
C elements.
15.7 Write a program to calculate the step and impulse response of a linear
system whose system function contains only simple, real poles and zeros.
Repeat Prob. 15.7 if simple complex poles and zeros are allowed.
15.9 Write a program to calculate the residues of a transfer function with
multiple as well as simple poles. The poles and zeros are to be real.
15.8
appendix
A.I
FUNDAMENTAL OPERATIONS
Matrix notation
is
more
quickly.
The theory of
linear transformations
from one
As an example of (2),
of variables to another.
set
u*i
^l^X
+
+
ami^i
aM";3
+
+
amixa
i**a
+
+
alnxn
"
a* *
= yx
= y*
**
(A.1)
+ amnxn = ym
Definition
coefficients
is
of order
is
of a linear transformation.
mx
n.
(The
Oil
12
fl gl
att
first
ln
number here
461
is
the
462
Network
second
is
the
analysis
and synthesis
matrix
is
a single complete
entity, like
a position in chess.
Two
<hi
olt
Oi
r*r
On.
on
oin
xa
Xn
i
'mi
'ml
"yr
=
.
Vt
(A.2)
Vm .
which
x-
Row
is
discussed later.
letters,
such as
or y. Note
elements of matrices need two.
A.2
ELEMENTARY CONCEPTS
Square matrix
Diagonal matrix
diagonal matrix
matrix
is
is
(i.e.,
diagonal.
'1
01
0-2
(A.4)
3
_0
Unit matrix
A unit matrix
for
=j, and
is
0"
"1
U=
(A.5)
,0
1_
Appendix
463
Equality
2,
in matrix form,
if
yt
3, and ya
6.
of equations
we have
"
2"
~Vi
= -3
y*
(A.6)
-6.
.y*.
Transpose
The transpose of a matrix A denoted as A r is the matrix formed by
interchanging the rows and columns of A. Thus, if we have
,
A=
-6
(A.7)
-CIS
A*
then
(A.8)
Determinant of a matrix
The determinant of a matrix
12
det
l
4j
L-5
14
(A.9)
matrix
Cofactor
The
cofactor
A i}
of a square matrix
is
A tl
-n
An =
IS
For
of the matrix
(-iy*+i
6= -6
(A.10)
(A.11)
Adjoint matrix
The
element of
by
its
cofactor
Network
464
and synthesis
analysis
we have
adjA
h
i
l:
-l;
(A.12)
OPERATIONS
A.3
ON
A=
0.
non-
MATRICES
Addition
may be added
Two
matrices
if
matrix
is
(A.13)
[-in-'-H-<-2
Thus
if
matrices A, B,
A+B+
The
associative
K are all
+K=
+b +
[a u
it
+k
tj ]
(A.14)
Associative:
Commutative:
A + (B + C) = (A +
A+B=B+A
B)
+C
(A.15)
Multiplication by a scalar
A[a w ]
[AaJ
scalar, multiply
(A.16)
each of
its
elements by
the scalar.
Example
Al
"
-1
"
0"
1
-3 2
0"
-3
-9
(A.17)
Appendix
465
oA
where and
/SB
[*ait
+ pb
(A.18)
it ]
are scalars.
/9
Multiplication
In order for matrix multiplication AB to be possible, the number of
columns of the first matrix A must equal the number of rows of the second
matrix B. The product C will have the number of rows of the first and
A has
the number of columns of the second matrix. In other words, if
product
the
then
columns,
rows and n columns, and B has n rows and/>
m
C
will
have
rows and
columns.
The
individual elements of
are
given by
(A.19)
Example A.2
4
(A.20)
2
[-:-:;]-[-,:;]
-3
Example A3.
ZtJi
+ Ws = Vx
+ z/s = Vt
(A.21)
caca-ra
We
matrix notation.
Matrix multiplication
is
AmnBB * B^A,
Observe that the product BA is not defined unless p =
of square matrices
is
is,
(A.23)
m. Even a product
may be
seen in the
following example.
L-i
onr-i
2JL
<n
_ r-i
2j~L
oi
(A.24)
f]
Network
466
If
we
anal/sis
and synthesis
-1
-1
-2
-1
2
we
obtain
(A.25)
= ABC
+ C) = AB + AC
Associative:
A(BC)
Distributive:
A(B
(AB)C
(A.26)
Transpose of a product
The transpose of a product AB is equal to the product, in reverse
of the transpose of the individual matrices A and B, that is,
(AB)
(ABC)r
and
- B rA T
order,
(A.27)
= C T(AB) T = C^B^A 2
"
(A.28)
xT x
The product xx r
Common
(a)
[ Xl
*i
xt
+ x*
T
is a square matrix C such that C = C
+
*,*
It is also
xx r
if
is
expression auxf
where
The
aA + ajb +
(c)
vector).
(A.29)
x]
may be
+ xt* +
+ anb,
written as
x n*
a nnxn *
is
x Tx
(x,
column
x rAx
(x,
column
thus
a row vector.
+ atixta +
is
is
2 Z aux x
i
t>
a,
Appendix
is
467
is
1a x%xx xt
*!*
+ 2a ln x1 x
2a u xt xa H
<J*
h 2a, B a;2 x B
(A.30)
x TAx
Inverse
Division
is
matrix
obtain
= AA- = U
1
the determinant of A.
that
inverse
A-1
is
of a
To
The
The
inverse
A-1
is
(A.31)
is,
A" 1
adj
(A.32)
|A|
Example A.4.
Let
A be given as
(A.33)
Its
determinant
is
|A|
=3
(A.34)
An = 1
Afn = 1
The
adjoint matrix
Au =
An =
(A.35)
is
adJA =
[-l
2j
(A.36)
-C
so that
A-1 is
-C 1
-EH
(A.37)
Network
46.8
As a check we
and synthesis
anal/sis
see that
-r
"
'
-1
-l
(A.38)
_i
If the determinant
In
A.4
Consider a
set
m*i
The h t are
we have
constants.
H
+
+ a^xt + +
r-
<*iss*
+ a n **i +
gives
In matrix notation
(A.40)
=A
(A.41)
Ax
Premultiplying by
(A.39)
+ a nnxn = h n
It is
-1
= hx
aSnx = hs
alnxn
is
x.
^si
J_
ia
|A|
^32
-"m
(A.42)
Thus
|A|
*1
Oia
*is
ht
On
ha
a S2
|A|
and
similarly for
xt and *3
This
is
such equations.
Example A.5.
Solve for x, y,
z.
2x
+2 =2
+y =
3y
+ = -1
-* +
(A.43)
Appendix
In matrix form, these equations are written as
1
-1
-1
Also, |A| = 10. Thus,
-&
Therefore
A.5
The following
is
A_1h
-r
7-2
-i
-0.4, z
(A.44)
-l
= 0.7, y =
REFERENCES
-2
469
Ax = h where
~x~
we have x =
*1
y
r
=
X
io
-4
(A.45)
= 0.9
ON MATRIX ALGEBRA
a short
list
examine.
A. C. Aitken, Determinants and Matrices, 9th Ed., Interscience Publishers,
New
York,
1956.
New
York, 1959.
Hohn, Elementary Matrix Algebra, The Macmillan Company, New York, 1964.
McGraw-Hill Book
L. P. Huelsman, Circuits, Matrices, and Linear Vector Spaces,
Company, New York, 1963.
and Sons, New York,
P. LeCorbeiller, Matrix Analysis of Electric Networks, John Wiley
F. E.
1950.
1964.
1952.
of Matrices, Addison-Westey, Reading, Massachusetts,
Englewood Cliffs, NJ.,
L. A. Pipes, Matrix Methods for Engineering, Prentice-Hall,
S. Perlis, Theory
1963.
London,
1962.
Electrical Engineers,
appendix
GENERALIZED FUNCTIONS
B.I
The
is
a mathematical anomaly. P. A.
He denned
d(x)
Its
by the equations
d(x)
dx
(B.I)
for
jt
is
f(x)6(z)dx=f(0)
(B.2)
is continuous at x
0. Dirac called the delta function an
improper function, because there existed no rigorous mathematical justi-
where f(x)
fication for
entitled
P.
1930.
1
and
470
and
II,
Hermann
Appendix B
physicists.
It
was not
until 1953,
elementary (although no
less rigorous)
new branch of
Our treatment of
deserved.
it
The treatment of
471
and
its
work of
Temple4 and Lighthill. 8
To get an idea of what a generalized function is, it is convenient to use
as an analogy the notion of an irrational number a beng a sequence {oc}
of rational numbers <x n such that
impulse.
= lim a
n-ao
where the
<x
on
on the
on the sequence {a} defining
a.' We can also think of a generalized function as being a sequence of
functions, which when multiplied by a test function and integrated over
number a
irrational
(00,
oo) yields
function,
and
(2)
it is
a regular sequence.
DEFINITION
B.I
everywhere, any
is differentiable
[t
m
<f>
C[#f) G C]
M(t)]-*0
raised to
for all
is
one that
that (2)
when
m&fe^O
it
(1)
or
limit is
(B.3)
*->oo
Any
testing function is
a function of class C.
M.
Press, 1935.
true,
it
Fourier Analysis
could be made."
* This defines an irrational
Press,
any
text
Network
472
and synthesis
anal/sis
DEFINITION
A sequence {/(*)}
B.2
of functions of class
is
said
<f>(t)
lim(/,#=lim
n-*ao
J oo
n-*oo
fn(t)<Kt)dt
(B.4)
exists.
n-oo
DEFINITION
alent if for all
B.3
Two
lim(/B ,#=lim(g B ,#
n-*oo
Example B.2.
The
(B.5)
n-*ao
and
equivalent.
DEFINITION B.4
generalized Junction g is denned as a total, or
complete, class of equivalent regular sequences. The term total implies
here that there exists
no other equivalent
Any member
both g and the
to this class.
represent
denning g.
We denote this
Example B.3.
total class
example, {g n },
is sufficient
to
1 '"4
Or*
},
~ {gn
}.
function
~ {e~**
B.5
<f>(t)
{--/*}]
{e-"'-}
DEFINITION
g and a
class, for
represent the
of the
C is defined
(g,
lnt
}.
<f>)
of a generalized function,
as
(g,#=lim f"gn(t)<Kt)dt
n-*oo
is
(B.6)
J co
(g,
+)
(B.7)
g(t)<Kt)dt
actual integration.
DEFINITION
If
defined by the
Note
and
h; therefore
+ h n}
~ {h
n },
the
sum g
+h
is
A}.
defining
DEFINITION
and a constant a
B.7
is
is
473
generalized function;
~ {#}
DEFINITION
{#}
Appendix B
g~
gx ~ {*-*
1/b *}
the derivative
is
represented
and
(g\, *)
= lim
"
(
- ^ e"**'"1
*(/)
(B.8)
//
by a
scalar
defined in general.
We
Lighthill, 7
which
as a step function
Theorem B.l.
by a generalized function
for
C.
lent in terms
write
dt
<
co
(B.9)
some
for all
1/(01
given in
equivalent.
1'
is
/ =/.
In other words, an ordinary function satisfying Eq. B.9 is equivaof inner products to a generalized function. Symbolically, we
If, in addition, / is continuous in an interval, then lim fn
=/
""*"
Furthermore,
multiplication,
it
and
(a
when
differentiation is permitted
m'
= (/i +
6ti'
on the ordinary
<M
function.
'
italic letter
Network
474
DEFINITION
total class
analysis
B.9
and synthesis
The
(,#
= lim
n-*oo
{(*)}
is
defined as the
such that
fuJit)4(t)dt
J ao
-f"ii(t)#0A
(B.12)
J 00
where
Chapter
That
2.
{(/)} exists is
Example B.5.
The sequence
/>o
t0
which
is
is
class
(B.13)
of equivalent regular
DEFINITION
is
,(/)
Appendix B
It
475
total class
when we
we
actually
mmdt
mean
f " <5(0 <Kt) dt
= (d,
J oo
Example B.6.
<j>)
= lim
n-oo
The sequence
^-(5-2r)p[-i(J
,>o
+ |.)]
(B.15)
=
in Fig. B.2
is
(B.14)
J oo
1.0
1.2
1.4
1.6
_1_
_L_
'(/).
476
B.2
Network
anal/sis
and synthesis
Sifting
unit impulse
is
represented symbolically by
J<o
W(0*-/(0);
is
(B.16)
(M,lj8|<>)
The
/J].
left
hand
side of
defined formally by
f>
\ \t)f(t)dt
Js<0
slim
The proof of
n-ao
is
(B.17)
\"*u'JLt)f{t)dt
Ja<0
parts, as follows.
>O
lim
n<x>
n-oo
f'ii,(0/'(0 dt
J*
-/tf)-f'limu.(0/'(OA
_,
= /C8)-[/(/5)-/(0)]=/(0)
Pictorially
impulse
is
we
represent <$(0
centered at
by a spike as shown in
Fig. B.3.
If the
as
d(.t
fl
)/(f) dt
- /(a)
(|a|,
KO
HI
<
ao)
(B.19)
Appendix B
where/'(0 most exist over
are
infinite,
we
actually
[a,
/?].
mean
477
sifting property, if
both
a,
(I
>
or a,
dt
<K0/(0
/?
(B.20)
< 0, then
(B.21)
/:
the sifting
Integration
The
f>0
d(x)
dx
/.<o
(B.22)
d(x) = 0;
i*
These are actually properties of the delta function as viewed from the
generalized function standpoint. The proof can be obtained directly from
the sifting property. Suppose we have the integral
f>
1
and we
let f{t)
1.
/?
(B.23)
Then we have
Ia<0
If both a,
\t)f(t)dt=f(0)
d(t)dt=f(p)
(B.24)
are greater than zero or both are less than zero, then
1
This property
is
d(t) dt
stated symbolically
(B.25)
by the conditions
d(t)
for
ft 0.
<
At)=MQ + Au(t-T)
(B.26)
Network
478
analysis
and synthesis
*M
h(t)
dt <
&\N
I (1 + O
for
some N; we can
co
functions
(ft)
=Mt) +
n(0
(B.27)
fV)=f\(.t)
(B.28)
Au'(.t)
no =A(o + A m
We
we
(B.29)
differentiate across
a discontinuity, we
Example B.7.
The
step response of
Hf)
shown
in Fig. B.Sa.
an R-C network
is
= Ae-" T u(t)
is
shown
in Fig. B.Sb.
(B.30)
is
h'(t)=Ad(t)-j,e- l T u(t)
and
given as
(B.31)
Appendix B
479
Differentiation
The
derivative of
symbolically as d\t)
f"(i) exists over [a,
{u" n(t)}.
It
call
a doublet,
is
defined
fit].
/V>o
<J'(0/(0 dt
J
J<0 '
-/'(0);
(||,
\fi\
successive integration
<
oo)
by
parts.
(B.32)
>0
lim
T" V(0/(0 dt = n-oo
f "B(0/(0 dt
Ja<0
Ja<0
fi>0
- Km
We see that since
'.(0/(0
lim u'Jfi)
lim
-lim
<0
n-oo
u' n (a)
>0
limH' B (0/0)'
n-*oo
<p
(B.33)
Ct>9
u'n (t)f'(t)dt
J<0
= 0,
=0
(B.34)
480
Network
<$'(')
-lim
u' B
['u n (t)f\t)dt
(B.35)
\yu{t)f{t)dt
-/'(/?)+
-/'(0)
J
J<0
A -(-DT"'()
(B.36)
/3].
d'(t)
is
The
unit impulse
Dirac and others have obtained a host of identities concerning the unit
impulse. We will merely give these here without proof.
- ,5(0
d'(-t) - -<5'(0
t <3(0 = o
td'(t) = -d{t)
<5(-o
(3)
(4)
(5)
a)
d(t
a)}
(6)
(7)
(2)
diat)
The proofs of
(1)
<f>(t)
'
C.
is sufficient,
appendix
AND OPERATIONS
CI ELEMENTARY DEFINITIONS
(CI)
The
Re(z),
= Im(z)
(C.2)
abscissa represents the x or real axis, and the ordinate represents the y or
imaginary axis. The plane upon which x and y are plotted is called the
complex plane. Any point on the complex plane, such as z = 3 +j2, can
be represented in terms of its real and imaginary parts, as shown in Fig. C. 1
From the origin of the complex plane, let us draw a vector to any point z.
The distance from the origin to z is given by
|x|
and
is
known
(x*
y)*
(C.3)
as the modulus of
arg z
tan
-i*
V
x
is
The
z.
known
3 *
(C.4)
\z\,
we can
arg z and r
Letting 6
represent z in polar coordinates as
re
(C.5)
481
FIG.
CI
Network
482
Expanding
anal/sis
this last
and synthesis
obtain
+jr sin 6,
z == r cos 6
(C.6)
= r cos 8
y = r sin
x
so that
The
we
where j* = 1.
given as
(C.8)
(C.9)
If
we
+jd)
(c
is
+ d)
+jb)
(C.7)
(a
are multiplied,
+j(b
c)
we have
we
obtain
m
(c+jd) = r^ m
+ jb) =
(a
and
When we
we
(C.ll)
r^rtf"*
If
(CIO)
ri e
divide these
two numbers
jv^**-
-*'
(C.12)
7^ = 7^'^
(C13)
(C.14)
it is
useful to note
1*1*1*1
z-z*
where z*
is
= Uil
=
= |z|
|il
tal
l*i*l
z*
is
2
l*il
(CIS)
defined as
= x +jy = x jy
(C.16)
Appendix
483
The following rules deal with operations involving the conjugate definition
Zi
+ zt = zx * + z,*
^=
Finally, if z has
*!**.*
(C.17)
-1
(C.18)
z*
and
= jMne sn+tk,)M
k0l
(C
19)
ANALYSIS
C.2
If to each z
is
1/n
a function of z or
w =/(*)
The following are examples of complex
(C.20)
functions,
i.e.,
functions of a
complex variable:
= 2z
w = log, z
w = 1/z
w = za + 4
w=
w
(C.21)
|z|
We
see that
w may
be complex, pure
real,
upon the particular relationship with z. In general, the real and imaginary
parts of w are both functions of x and y. That is, if we let w = j + jv,
then
u-A*,V)
As an example,
let
we
v=f(x,y)
w=
Simplifying,
and
z*
+4=
(x
+jy)*
+4
(C.22)
= z* + 4.
(C.23)
obtain
= (* - y* +/'2ry) + 4
(C^4)
Network
484
and synthesis
analysis
jy
jy
Ax
J&
z
+ Az
Pathl
+ A?
Path 2
-jy
-jy
FIG. C.2
so that
The
derivative of a
FIG. C.3
a:*
yl +
and
= 2y
i>
is
(C.25)
defined as
(C.26)
If
one
we have what
function
is
is
known
Az approaches
zero, then
to possess a derivative at
all,
complex
same
at
= constant
(C.27)
dz
for all directions of approach of Az.
in
=lim
,/(z
lim'
A*->0 Av->0
If
we
+ Az)-/(z)
Figs.
C.2
(C.28)
Az
substitute
Az
into Eq. C.28,
f'(z)
we
= Ax + j Ay
(C.29)
obtain
= lim
lim
f[x
(C.30)
Aac-0 Aff-*0
Since
and
(C.31)
(C.32)
Appendix
we
485
finally arrive at
/'()
Am +jAi>
= lim
lim
+ j'Ay
=+
-*/Ao
= hm Au +
Ax
a*-o
,.
For path
2,
9u
(C.33)
.
dx
dx
we have
Au + Ad
Km " T J A
/
/'(z)=lim
a-oas-oAx
==lim
+ J Ay
AiLM
(C.34)
/'Ay
a-o
dt>
dy
du
dy
_ =
dy
dy
From this
last equation,
we
4-
derivatives
(C 35)
9x
dx
are
dv
du
dx
dy
du__dv
dx
dy
We have just seen that in order for a function to have a derivative, the
Cauchy-Riemann equations must hold. A function which is single valued
and possesses a unique derivative is called an analytic function. A set of
Cauchy-Riemann equations
For example, consider the function
are obeyed.
/(z)
/(z)
is
= z* + 4
(C.37)
analytic because
^ = 2x = ^
dy
9
dx
= 2
dy
(C.38)
dx
Network
486
On the
other hand,/(z)
= +1
du
j.,
and
a
and
*>
SINGULARITIES
AND
(C39)
= -1
ay
ox
C.3
= y
1
RESIDUES
at a point z
- z )B
(z
-z
+
Om(2
- So)"* +
(C.40)
In the expansion,
term a_i
is
if
m is finite,
then z
is
m.
The
Example C.l.
= e*\z
about
*=
I/i
z\
+* +
+ -*+
2!
3!
/
(C.41)
= l+l + Lz+l.z* + -z
According to the
2!
3!
z(z-l)
(z
Note
then *o
is
|z
1|
that if
an
<
,
(z
1.
l/z(z
l)2
equal to
1.
1,
[1
1)
- (z -
1)
+ (z -
1)
- (z -
I)*
1)
is
equal to 1.
(C.42)
1)
1
<
is
- i)M + (z 1
for
of the pole at z
,
1)*
- (z -
1)
(z
we have an
infinite
essential singularity.
negative exponents,
Appendix
Example C3.
and s
C
-1
487
of the
function
s
To find
we simply perform a
the residues,
(C.43)
partial fraction
expansion
(C.44)
+2
s =
is
CONTOUR INTEGRATION
IC
f(z) dz
= lim 2 f(z
n-oo j 1
i)
A*/
(C.45)
Thus the
integral
r f(z)dz
(C.46)
or path C8 as
a closed path, say from a to b and then to a again, we are integrating along
a closed contour. The path shown in Fig. C.5 is an example of a closed
contour. The following theorem, known as Couch/ s residue theorem gives
Jy
jy
Closed
'contour
FIG. C.4
Theorem C.l.
If
FIG. C.5
z^, z,,
...
z,
Network
488
anal/sis
C is
I
where
and synthesis
f(z)dz
(C.47)
Example C.4.
s+2 ds
sH.s + 1?
along the
within the
circle
\s\
(C.48)
circle, at s
s\s
is
+ 2 ds - 2*j(-3 +
+ iy
3)
(C.49)
C=|.|-2
Example C.5.
C.7.
The
Find the
function /()
is
35+5
IX* +
'*"-(* +
in Fig.
(C.50)
2)
Jy
'
'
L
shown
given as
-2 -1
-Jy
FIG .
C7
Appendix
489
m ri
+
+ + Txi
1
singularities lie is
j> /(s) ds
If f(z) is analytic in
iTTJil
a domain with no
is
zero, that
is
known
2)
then
= 6tt/
singularities,
(C.52)
is,
f(z)dz
if
This result
(C51)
(C.53)
D
Proofs of some theorems on
appendix
Theorem D.l.
If Z(s)
is
also
positive real.
Proof.
also.
Z(W(s))
satisfies
Theorem D.2.
W(s)
Proof.
Theorem D.3.
Z(s)
Proof.
it is
positive real.
real.
hence Z(W(s))
by
Theorem D.l.
Theorem D.4. The sum of positive real functions is positive real.
Proof. Suppose Zx(s) and Z2(s) are both positive real. When Re s
ReZx k
then
ReZ2 ^
so that
is
and
^ 0,
0.
Also, when s is real, both Zx and Z2 are real. The sum of two real numbers
a real number. Therefore, Zx + Zt is positive real.
Theorem D.5.
(i.e., lie
poles
Suppose there
Proof
Laurent
The
series
(*
- 5o)n
(S
- * )"
490
Let us
make a
Appendix
491
ReZ<)
FIG. D.I
finite.
Z(s)
,-^V(s - s n
)
We can represent Z(s) in polar form by substituting each term by its polar form;
(j - o)+" = rV and k_n = Kt** so that
i.e., let
z<*) = ^ **~**.
Re Z{s) =
* cos (* -
nfl)
which is represented in Fig. D. 1 When 9 varies from to lit, the sign of Re Z(j)
change 2n times. Since ReZ(s) t.
when Re s 2. 0, it is seen that any
change of sign of Re Z(s) in the right-half plane will show that the function is not
positive real. Therefore, we cannot have a pole in the right-half plane. Since the
.
will
function 1/Z(s)
is
it is
Theorem D.6.
can
exist
on the
ya> axis.
As a consequence of the
may exist on they'd) axis if n =
Proof.
poles
and real.
is
Proof.
real
when
If
1,
and ^
It is readily
Theorem D.7.
derivation of
on
seen that
it is
implies
they' to axis
is real.
<j>
Theorem D.5,
= 0. The condition n
cannot be
it is
seen that
+ fln-iJ"-1 +
m + bn.lS"
*>m*
?"
Z{s)
+ OyS + a.
+
+b
Pis)
Q(s)
Network
492
anal/sis
and synthesis
FIG. D.2
Theorem D.9. The lowest powers of P(s) and Q(s) may differ by at most unity,
Proof. The proof is obtained as in Theorem D.8 by simply substituting 1/s
for s and proceeding as described.
Theorem D.10.
has poles on
they'a> axis,
real, positive
residues.
(c)
Re F(ja>) ^
for all
o>.
We need only show that these three conditions fulfill the same requirements as Re Z(s) ^ for Re s ^ 0. We will make use of the minimum modulus
Proof.
theorem which
appendix
An
aid to
of
E.I
the improvement
filter
approximation
INTRODUCTION
contours with this axis. Corrections which best reduce the errors in the
original approximant are then sought by variation of the overlay position
and orientation. Either phase or amplitude may be corrected. However,
it is not always possible to simultaneously improve both the phase and the
amplitude characteristics of an approximant by a single pair-shaped
correction.
F. F.
Kuo and M.
CT-9, No.
1
4,
December
Network
494
E.2
analysis
and synthesis
in
GAs)-GM<K)
will
CE.1)
this
paper,
O^-C <'-)<'->
p)(s
(s
(E.2)
p)
GMcsZtzA
(E.3)
s-p
Let us consider the effect when the pole-zero pair in Eq. E.3 is used to
augment any given rational function in the complex frequency plane.
The added
gain, in decibels,
D=
where we have neglected the
then
For
due to
20 log10
effect
fc
-q
s- p
s
of the constant
(lO)*'
20
E.3. If
we let
(E.5)
circle
is
kjp^q\
II
its
C in Eq.
(E.6)
P
and
(E.4)
s-p
p. Its radius
center
is
- *"l
(E7)
at
Press, Oxford,
*
E. C. Titchmarsh, The Theory of Functions, Oxford University
England, second edition 1939, pp. 191-192.
Appendix E
495
Let
(E.9)
Then
5
II
(E.10)
k*\
(E.ll)
Here, s
is e. It is
is the midpoint between the pole and zero, and their separation
easy to see from Eq. E.l 1 that the center of each circle of constant
gain
E.3
4>
9,
(E.12)
Network
496
and synthesis
analysis
equal to one half the subtended arc ps'q. Therefore, the arc qsp is a
constant phase contour. The angle at the center of the circle between cp
and the perpendicular bisector of chord pq is also equal to
All of the
<f>.
circular contours of constant phase have their centers on this perpendicular
bisector.
is
phase angle
is
negative, as indicated
<f>
(E.13)
E.4
CONTOUR DRAWINGS
drawn
in Figs. E.2
and
E.3.
decibel line, except that the gain curves are of opposite sign.
Therefore,
7.5*
=^5*
-K s
~~yf
/i?\
9'
v>
I
10*
^.s\
lb"
30
i
Ills
2\
1.71 1
5|
]l
|] .2
\
1
0.7
0.6
'
TO*
20
'
lb"
10"
9*
7.5"
^v
*Sf
v^jy
-J'l
rV.
db
0.5
Appendix E
497
The
zero-decibel line
is
The
line
Fig. E.4.
The corrections in
Figs. E.2
Network
498
analysis
and synthesis
Zero gain
.1
G-
G+
<t>-
-x^-*-Zero phase
G-
G+
<t>+
G and phase
<f>
corrections.
only half of the symmetrical pattern has been drawn. The signs may be
obtained from Fig. E.4, and are important in selecting the orientation of
the pole-zero pair.
E.5
CORRECTION PROCEDURE
There are several variables at the designer's disposal. The first is the
distance between the correcting pole and zero. Since the scale of the
contours in Figs. E.2 and E.3 is not specified, the frequency scale of
the underlying complex plane determines the distance between the pole
and zero. The second design variable is the location of the center of the
pole-zero pair. The distance of the center from a given band on theyeo
axis determines the
The
third variable
Appendix E
499
zero phase axis should be rotated clockwise with respect to the a (real)
axis.
We thus see that by varying the frequency scale of the complex frequency
plane, the position of the center of the pole-zero pair,
may be helpful.
Unless the pole and zero both lie on the real axis, one must remember
that another pole and zero are located at conjugate positions. The contributions from both pole-zero pairs may be added algebraically. In most
practical cases, the desired correction will have a band-pass character.
statements
any frequency.
way
The broadness
in
the best
fit.
different pole-zero pair, also shown in Fig. E.7, has been chosen to minimize
the deviation of the slope of the phase response from its slope at to = 0. This
pair-shaped correction decreases the phase for m < 0.7 and increases it for
co
> 0.7.
Figure E.8 shows the deviation of the phase responses from linear phase.
It is clear from Figs. E.5 and E.8 that the gain corrected approximant has a
poorer phase response than the original, while the phase corrected approximant
500
Network
anal/sis
and synthesis
0.2
0.5
0.3
Frequency,
0.7
1.0
1.5
2.0
filter
designer.
It is possible,
filters.
however,
both gain and phase by using two pairshaped corrections. A useful approach to this objective lies in localizing the
gain correction further out of band, and the phase correction further in-band
than in the separate corrections just discussed. This can be done by shifting
the pole-zero centers to higher or lower frequencies and also by experimenting
with nonsymmetrical corrections.
to achieve moderate corrections in
fall-off
of third-degree Butterworth
filter.
Appendix E
Loss
1.2
1
correction
o
1.0
Butterworth
poles
0.8
X o
Phase
0.6
correction
0.4
0.2
_L
J_
-0.2
\
\
-0.4
-0.6
x o
-0.8
-1.0
X
-1.2
filter
20
>
10
With >hase :orrect on
s -10
^'
Tl ird-orc er
^^v
/
//
//
/
/
/
zL
Biitterwo rvn'/
\\ 1
\
-20
>
/w th amp litude
correct on
-30
-40
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
Frequency, rad/sec
501
Network
502
E.6
anal/sis
and synthesis
few words are in order concerning the synthesis of the equalizer from
the pole-zero pair obtained by the method described.
In order to provide optimum power transfer and to facilitate cascading
the correction network with the original
should be of the constant-resistance type.
filter,
We will restrict
whose voltage
our discussion
transfer function
G(s)
is
is
(E.14)
Z (j)Z2(i) =
1
will let
(E.15)
R =
Q.
+as+a
fc(s* + b lS + b
s
minimum
phase,
we know
We
(E.16)
)
kb
that {a ( , b t
^a
^ 0}.
In
_L_ 1= (
l)s*
G(s)
so that
all
a)
(Ai')
be nonnegative
k- ^0
1
kbi
kb
<*i
^
^
(E.18)
Zi
So
Vi
So
So
Z2
So
v2
Appendix E
503
-ju
FIG. E.I0. Poles and zeros of Zt(s).
must apply. 8
Z^j) are
the locations of the poles and zeros in terms of the polar angles
and
their distances
from the
origin,
n p and n t
finite
^<2
(E.20)
If
A.
J.
Minimum
Function,
IRE
Network
504
E.7
CONCLUSION
We
L. Weinberg,
"RLC lattice
Bibliography
SIGNAL ANALYSIS
Goldman,
S.,
McGraw-Hill,
Frequency Analysis, Modulation and Noise,
'
New
Javid,
York, 1948.
M. and
E. Brenner,
McGraw-Hill,
New York,
1963.
New
5>ons,
York, 1965.
Lighthill,
Functions,
York, 1958.
Cambridge University
Circuits, Signals and
Mason, S. J. and H. J. Zimmerman, Electronic
1960.
Systems, John Wiley and Sons, New York,
McGraw-Hill, New
Applications,
Its
and
Integral
Papoulis, A., The Fourier
York, 1962.
Rowe, H.
E., Signals
and Noise
in
McGraw-
NETWORK ANALYSIS
Allyn and Bacon, Boston,
Balabanian, N., Fundamentals of Circuit Theory,
Systems, Addison-Wesley,
Bohn, E. V., The Transform Analysis of Linear
York, 1959.
505
New
Network
506
W.
Cassell,
L.,
New York,
1964.
Chen,
W.
New
York,
1963.
New
York, 1953.
Guillemin, E. A., Theory ofLinear Physical Systems, John Wiley and Sons,
New York,
1963.
Harman, W. W. and D. W.
McGraw-Hill,
New York,
Lytle, Electrical
1962.
1961.
Legros, R. and A. V.
New
W.
R. and
York, 1952.
LePage,
J.
Prentice-Hall, Inc.,
Englewood
S. Seely,
New York,
Lynch,
W. A. and
Hill,
New York,
McGraw-
1959.
J.
1961.
Englewood
Pearson, S.
I.
and Sons,
and G.
J.
New York,
Bibliography
Pfeiffer, P. E.,
Reza, F.
507
M. and
S. Seely,
York, 1959.
Sanford, R.
S.,
Cliffs, N.J.,
1965.
R.
E., Elements
of Linear
Circuits,
Addison-Wesley, Reading,
Massachusetts, 1965.
Seely, S.,
Seshu, S.
Sons,
New York,
1959.
Skilling,
1954.
NETWORK SYNTHESIS
Balabanian, N., Network Synthesis, Prentice-Hall, Englewood
Cliffs, N.J.,
1958.
York, 1964.
Geffe, P. R., Simplified
Modern
Filter Design,
John F. Rider,
New York,
1963.
New York,
1957.
Network
508
Saal, R.,
analysis
and synthesis
Tiefpdsse, Telefunken
Skwirzynski,
J.
GMBH,
1961.
Electrical Filters,
D. Van
Truxal,
Tuttle,
J.
D.
1958.
Van Valkenburg, M.
1962.
Yengst,
York, 1964.
Name
Eisenman, R. L., 469
Elmore, W. C, 390
Ende, F., 381
Barnes,
428
Cassell, W. L., 506
Cauer, W., 324
Chen, W. H., 506
Chien, R. T., 506
Carlin,
H.
I.,
Darlington,
S.,
431, 457
Davis, H. F., 51
de Pian,
L.,
506
Jahnke, E., 381
James, R. T., 389
Javid, M., 505
Dutta Roy,
S.
Index
C, 283
509
Name
510
Jones, E.
Jordan, E.
Index
M. T., 507
C, 413
Rowe, H.
E.,
505
27
Justice, G.,
W.
Kautz,
Sanford, R.
Semmelman, C.
Le
Seshu,
Corbeiller, P.,
Lighthill,
R.,
M.
.,
506
Lynch,
Lytle,
W.
447
Skwirzynski,
506
J.,
L.,
503, 507
Skilling,
Ley, B.
S.,
469
Le Page, W.
507
262
I.,
S.,
A., 506
D. W., 506
505
Tellegen, B. D. H., 272
Valley, G. E.,
Van
A. C, 291
507, 508
Von
Weber,
Paley, R. E.
506
Pederson, D. O., 507
Perlis, S., 469
Pfeiffer, P. E., 507
Pipes, L. A., 469
Pearson, S.
Rack, A.
J.,
I.
E., 507
Weinberg, L., 400, 435, 504
Widder, D. V., 136
Wiener, N., 291
Yengst,
W. C, 508
Young,
L.,
507
503
Subject Index
ABCD
parameters, 262
Admittance, driving-point,
All-pass network, 221, 357
15,
187
Amplifier, 11
Characteristic impedance,
Cbebyshev
414
Characteristic value, 78
filter,
373
approximation, 366
locus of poles, 376, 378
tables of, 373, 400, 435
transient response, 392
Chebyshev polynomial, 373
Circuit, bridged-T 258, 275,
502
double-tuned, 238
Band-elimination
filter
255
229
symmetrical, 259
reciprocal, 8,
transformation,
single-tuned,
409
Band-pass filter transformation, 407
Bandwidth, half-power, 235
spectral, 67, 389
Bessel
filter,
383, 435
Complex
analysis,
400, 435
483
484
487
differentiation,
392
Bessel polynomial, 386, 395
transient response of,
integration,
programs,
438,
439,
Computer
Black box, 15
filter,
500
pole locus of, 371
step response of, 391, 394
tables of, 372, 400,
Contour
18,
487
integration,
435
Damping
factor,
225
d-c value, 25
time, 388
511
245
Subject Index
512
65
properties of, 67
145
76
nonhomogeneous, 75
ordinary, 76
linear,
193
18,
11,
438
Dot reference
470
for transformer,
123
Duty
cycle,
symmetry conditions, 68
Free response, 106
Frequency, angular, 2
complex, 4
domain, 14, 134, 367
forced, 192
natural, 192
Frequency normalization, 18, 402
Frequency transformation, 18, 404
26
Energy density, 71
Energy spectrum, 71
Equal ripple approximation, see Chebyshev approximation
Essential singularity, 486
Excitation, 1
379,
Ideal low-pass
filter,
457
441, 457
computer programs
for,
driving-point,
transfer,
1,
46,
15, 253,
50
amplitude spectrum, 57
complex form, 55
cosine series, 53
evaluation of coefficients, 52, 58
188
13, 77,
Initial conditions,
phase spectrum, 57
Fourier transform, 1, 63
amplitude spectrum, 64, 65
inverse,
63
64
of unit impulse, 65
of unity, 68
86 107, 176
Initial
conditions, 53
276
orthogonality conditions, 49
discrete, 56,
315
matrix, 254
transformer, 263
Fourier series,
424
symmetry
292, 367
383, 493
filter synthesis,
Insertion
power
431
ratio,
430
513
Subject Index
136
Node equations,
Norm, 47
255
106,
Optimum (L)
486
L-C immittance, 315
properties of, 315
synthesis of, 319
series,
379
filter,
parameters,
254, 343
Laurent
impedance
Open-circuit
filter
Orthogonal set, 47
Orthonormal set, 47
Overcoupling, 242
Overshoot, 388, 392
Parseval's equality,
ideal elements, 12
50
theorem, 71
ideal models, 10
MAC,
Project,
448
Minimum
inductance
transformation,
443
filter,
see
Optimum
filter
Passive network, 8
Peaking
233, 241
circle,
shift, 1
distortion,
213
3,
57
Phasor, 5
Plancheral's theorem, 71
n-port,
449
100
reciprocal, 8, 100,
255
symmetrical, 259
synthesis, 290, 315, 341, 397, 431
time-invariant, 9,
100
passive, 8,
100
(p.r.)
function,
16, 299,
490
Power, average, 301, 315
available, 418
Propagation constant, 414
Proportionality, principle of, 8
514
Q,
Subject Index
Spectra, continuous, 3, 63
236
circuit, 233,
57
line (discrete), 3,
Ramp
Stability,
function, 31
R-C admittance,
R-C impedance,
290
marginal, 293
Steady-state solution, 106, 109
342,
116,
16, 290,
301,
490
filter
Reciprocal network,
8,
255
Rect function, 65
Reference impedance factor, 415
Reflected parameter, 415
Reflected power, 416
Reflection coefficient, 416, 421, 458
Remainder function, 308
Residue, definition of, 162
evaluation by vector method, 162
Residue condition for two-ports, 344
Resistor, 13, 103, 175
Response, critically damped, 118
overdamped, 118
underdamped, 119
Ringing, 388
Rise time, 388
R-L admittance, properties of, 325
synthesis of, 329
R-L impedance, properties of, 331
synthesis of, 331
Sampled-data system, 142
Sampler, 18, 142
element, 416
R-L-C
functions, 333
System function,
14, 187,
admittance
circuit,
Signal, continuous,
decomposition
parts, 21
parameters,
248, 400
23
into
even
20
46
symmetrical, 21, 54
Time constant, 24
Time delay, 32, 212, 245, 384, 388
Time delayer, 11
Time domain, 14, 134, 366
Time-invariant system, 8
Transfer function,
14,
486
16,
188,
266,
341, 352
impedance,
Transformed
16,
188, 347
175
Transformer. 122. 177
ideal, 263
circuit,
filters,
388
and odd
common
emitter,
Transmission
coefficient, 421,
Transmission
line,
286
457
413, 425
273
Sine function, 65
Singularity,
194
Terminals, 12
Transistor,
257, 344
periodic, 20,
deterministic,
397
synthesis,
matrix, 420
Shunt peaked
L-C
R-C
R-L
Short-circuit
Undercoupling, 241
515
Subject Index
of,
ratio,
428
40
Unit ramp
Unit step function, 28, 474
derivative of, 34, 37, 475
480
z-transform, 142
o
This Wiley international Edition
gram of paperbound
text
is
hardbound
It
is
m
o
O
z
D
m
g
o
for stu-
an unabridged
edition,
which
is
also
titles in
Chemical Engineering
Civil
Engineering
Electrical Engineering
Psychology
Agricultural Engineering
Biological Science
Metallurgy
Chemistry
Mechanical Engineering
Earth Sciences
Business
Management
(A
and Agriculture
Mechanics
Statistics
3
o
in
>
2
O
Mathematics
Physics
c/>
C/>
WILEY
TOPPAN
[P 10)