13.

1 The Substitution theorem ◀ 1

Chapter 13
13 Network Theorems
December 2020
By Mohammad Mazi
Dedicated to the Department of Mechatronic Engineering

Introduction
In this chapter we shall study four very general and useful network theorems: the
substitution theorem, the superposition theorem, the Thévenin-Norton equivalent
network theorem, and the reciprocity theorem. The usefulness of these theorems lies
in that (1) they are applicable to very large classes of networks encountered in practice
and (2) their conclusions are very simple. This generality and simplicity may be
deceiving; often people do not perceive the breadth of application of these theorems,
or know precisely what they mean.
The principal assumption underlying all these network theorems is the uniqueness
of the solution for the network under consideration. We discussed, in Secs. 7 and 8 of
Chap. 10, the fact that for linear networks (with the exception of degenerate cases)
the solution is always unique for any given set of inputs and for any given initial state.
The substitution theorem is a very simple but general theorem which holds for all
networks with a unique solution. It can be applied to linear and nonlinear networks,
time-invariant and time-varying networks. The other three theorems apply only to
linear networks. Recall that a linear network, by definition, consists of elements which
are either linear or independent sources. These independent sources are the inputs to
the network. The superposition theorem and the Thévenin-Norton equivalent network
theorem apply to all linear networks (time-invariant or time-varying). Thus, these
networks may include linear resistors, linear inductors, linear coupled inductors, linear
capacitors, linear transformers, linear dependent sources and independent sources. The
reciprocity theorem applies to a more restricted class of linear networks; the elements
must be time-invariant, and they can not include dependent sources, independent
sources.

13.1 The Substitution theorem

13.1.1 Theorem, Examples and Application
The substitution theorem allows us to replace any particular branch of a network
by a suitably chosen independent source without changing any branch current or any
branch voltage. The primary reason for the substitution is that the substitute network
is easier to solve than the original one. With this idea in mind we are ready to state
the theorem.

Substitution Theorem
Consider an arbitrary network which contains a number of independent sources.
Suppose that for these sources and for the given initial conditions the network has a
unique solution for all its branch voltages and branch currents. Consider a particular
2 ▶Network Theorems

branch, say branch k, which is not coupled to other branches of the network. Let jk  · 
and vk  ·  be the current and voltage waveforms of branch k . Suppose that branch k
is replaced by either an independent current source with waveform jk  ·  or an
independent voltage source with waveform vk  ·  . If the modified network has a unique
solution for all its branch currents and branch voltages, then these branch currents
and branch voltages are identical with those of the original network.

Remark on uniqueness
Applying this theorem requires in principle that we check that the modifled network
has a unique solution, i.e., that the specified initial conditions and the inputs determine
uniquely all branch voltages and all branch currents. For linear networks (time-
invariant or time-varying) this is always the case except for some degenerate cases (see
Secs. 7 and 8 of Chap. 10). In fact, we have shown that if a network consists of resistors,
inductors, and capacitors that are linear, time-invariant, and passive, then the initial
currents in the inductors, the initial voltages on the capacitors, and the inputs
determine uniquely all branch voltages and all branch currents. The statement is still
true when coupled inductors are present, provided every set of mutually coupled
inductors has an inductance matrix which is positive definite.

Example 13.1 Consider the circuit shown in Figure 13.1a. The problem is to find the
voltage V and current I of the tunnel diode given E , R , and the tunnel-diode
characteristic (shown in Figure 13.1b). The solution is obtained graphically in Figure
13.1b. According to the substitution theorem we may replace the tunnel diode either
by a current source I or by a voltage source V as shown in Figure 13.3. It is easy to
see that in both cases the solutions are the same as that obtained originally.

𝐸𝐸 − 𝑣𝑣
𝑖𝑖 =
𝑅𝑅

Figure 13.1 A tunnel-diode circuit and its characteristic; the point is the solution of the circuit.
 13.1 The Substitution theorem ◀ 3

Figure 13.3 The modified networks of Figure 13.1a obtained by the substitution theorem.(a) With a
current source ; (b)with a voltage source .

To illustrate the importance of the condition that the modified network must have
a unique solution, let us consider the voltage source E and resistor R of Figure 13.1
as forming the branch k of the theorem. Let us replace it by a current source I as
shown in Figure 13.2a. The possible solutions of the modified network are located at
the intersection of the diode characteristic and the current source characteristic. Figure
13.2b shows that there are three possible solutions, namely V1, I  , V2, I  , and V3, I 
. Clearly, only one of these solutions can coincide with that of the given network.
Hence, as a tool for solving networks, the substitution theorem is effective only when
the modified network has a unique solution.

Figure 13.2 Example showing the importance of requiring that the modified network have a unique
solution; in the circuit of (a), has three solutions , , and .

The substitution theorem has many applications. The substitution theorem is

particularly useful in the analysis of networks which contain a single nonlinear or time-
varying element. We shall now give a simple application to illustrate the power of the
theorem.

Example 13.2 Consider any linear time-invariant network which is in the zero state
at time zero and has no independent sources (see Figure 13.4). We shall use the
frequency-domain characterizations of the network, i.e., network functions in terms of
4 ▶Network Theorems

the complex frequency variable s . Assume that there are two accessible terminals to
the network which form a port. Let us apply at time zero a voltage source e0 to the
port and let the response be the current i entering the port. Call E 0  s  and I  s  the
Laplace transforms of e0  t  and i  t  , respectively, as shown in Figure 13.4a. The
network function, by definition the ratio of the Laplace transform of the response to
that of the input, is in this case the driving-point admittance
I s 
Y s   (13.1)
E0 s 

Figure 13.4 An application of the substitution theorem showing that the driving-point impedance of a
linear time-invariant network is the reciprocal of the driving­point admittance of the same network;
is in the zero state at and contains no independent sources.

Next let us apply a current source i0 to the port and let the response be the voltage
v across the port. Call I 0  s  and V  s  the Laplace transforms of i0  t  and v  t  ,
respectively, as shown in Figure 13.4b. The network function is then the driving-point
impedance
V s 
Z s   (13.2)
I 0 s 
The driving-point impedance is equal to the reciprocal of the driving-point admittance;
i.e.,
1
Z s   (13.3)
Y s 
The proof of this relation can be made by standard loop and node analyses; however,
it is lengthy. On the other hand, if we use the substitution theorem, the conclusion is
obvious. Let us consider the voltage source E 0 in Figure 13.4a as the specific branch
k of the theorem. Applying the theorem, we replace the voltage source by a current
 I  s  , where I  s  is the current entering the port in Figure 13.4a. The
source I 0  s   
network after substitution is shown in Figure 13.4b. The substitution theorem means
that the voltage across the port in Figure 13.4b must be the same as the voltage across
the port in Figure 13.4a; that is, V  s     E 0  s  . Hence we conclude from Eqs.(13.1)
and (13.2) that Z  s   1
  / Y s  .
 13.2 The Superposition Theorem ◀ 5

13.2 The Superposition Theorem

The importance of the superposition theorem is hard to exaggerate, for it is the
foundation of many engineering systems in daily use, such as high­fidelity audio
systems, telephone systems, broadcasting systems, analog computer components, and
numerous measuring instruments and tech­mques.
Roughly speaking, the superposition theorem means that, for a linear network, the
zero-state response caused by several independent sources is the sum of the zero-state
responses due to each independent source acting alone. Let us illustrate this concept
by a high-fidelity microphone­amplifier-loudspeaker system. If we concern ourselves
only with the electric circuit aspects of it, we may think of the microphone as a voltage
source in series with an impedance; the output of the circuit is the current in the
driving coil of the loudspeaker. Suppose we want to amplify the music produced by a
violin and a piano. If superposition applies, the response when both the violin and the
piano play simultaneously is the sum of the responses due to each one of them acting
alone. If superposition did not apply, we would hear the sum of their respective
responses plus some “interaction”. Imagine the effects that would then appear in the
case of a 140-piece orchestra! Because high-fidelity enthusiasts demand that a violin
sound like a violin whether or not the piano is played simultaneously, the designer of
high-fidelity systems must make sure he ends up with a linear system, for then he is
assured that superposition applies.

13.2.1 Theorem,Remarks,Examples and Corollaries

With this general picture in mind, we give the theorem’s precise formulation as follows.

Superposition Theorem
Let N be a linear network; i.e., let each of its elements be either an independent
source or a linear element (linear resistor, linear inductor, linear capacitor, linear
transformer, or linear dependent source). The elements may be time-varying. We
further assume that N has a unique zero-state response to the independent source
waveforms, whatever they may be. Let the response of N be either the current in a
specific branch of N , or the voltage across any specific node pair of N , or more
generally any linear combination of currents and voltages. Under these conditions, the
zero­state response of N due to all the independent sources acting simultaneously is
equal to the sum of the zero-state responses due to each independent source acting one
at a time.

Example 13.3 Consider the linear network shown in Figure 13.5. The network is in
the zero state; that is, 𝑣𝑣(0− ) = 0. Let the voltage across the capacitor be the response.
the independent sources are is  t   Iu  t  and es  t   E   t  , where 𝐼𝐼 and 𝐸𝐸 are
constants. Let vi   Be the zero-state response due to is acting alone on the network
(i.e., with es  0 , hence a short circuit); vi is given by
 t 
vi (t )  IR  1  e RC

  
for t  0
 
6 ▶Network Theorems

Let ve  ·  be the zero-state response due to es acting alone (i.e., with is  0 hence an
open circuit); it is given by
E  t
ve  t   for t  0
e RC         
RC
Consider the differential equation of the network with both sources present as follows:
v  es dv
is  C
R dt
Or
dv v e E
C   is  s  Iu  t      t 
dt R R R
It is easy to verify that the zero-state response due to is and es acting simultaneously
is
 E   t
v  t   IR    IR e RC        
for t  0
 RC 
Clearly, we have, as the superposition theorem requires
v  t   vi  t   ve  t           
for t  0

Figure 13.5 Example 1: A linear network with two independent sources.

Example 13.4 Consider the nonlinear circuit shown in Figure 13.6. Let is be a dc
current source of 4 amp and es be a dc voltage source of 10 volts. Let the response be
the voltage v . For is acting alone, the response is vi   4
 volts. For es acting alone, the
response is ve   0
 (the ideal diode is reverse-biased). For is and es acting simultaneously,
 (it is easy to check that with both is and es
the response due to both sources is v   0
turned on the diode is still reverse-biased). Hence, for these element values, source
location, source waveform, etc., the superposition theorem does not apply.
Consider next the same circuit, but with is   10
 amp and es   10 volts. Then it
is easy to show that vi   10
 volts, ve   5
 volts, and v   15
 volts. For these element
values, source location, source waveform, and choice of response, the superposition
theorem applies. In the present situation, the source waveforms are such that the diode
is, in all three cases, forward-biased; thus, the nonlinearity of the diode does not enter
the picture.
 13.2 The Superposition Theorem ◀ 7

Figure 13.6 Example 2: A nonlinear circuit with an ideal diode.

Example 13.5 Balanced networks often furnish examples of nonlinear networks in

which a nonlinear element does not affect the application of the superposition theorem.
Consider the balanced bridge shown in Figure 13.7. Let v be the response. Since the
bridge is balanced, it is clear that neither is nor es can cause current to flow through
the diode. Therefore, if v is the response, the superposition theorem applies (as far as
the sources es and is are concerned). It should be stressed, however, that if the voltage
source were placed in series with the diode and the current source in parallel with it,
then the superposition theorem could not possibly hold for all source waveforms. For
one waveform the diode might be conducting; for the other waveform and for their
sum it might be reverse-biased.

Figure 13.7 Example 3: A

Up to now the superposition theorem has been stated exclusively in terms of the
zero-state response of a linear network. Since the sinusoidal steady state is the limiting
condition (as t   ) of the zero-state response to a sinusoidal input, it follows that
the superposition theorem applies in particular to the sinusoidal steady state. More
formally, we state the following corollary.

Corollary 1
Let N be a linear time-invariant network; i.e., let each of its elements be either an
independent source or a linear element. Suppose that all the independent sources are
sinusoidal (not necessarily of the same frequency). Then the steady-state response of
8 ▶Network Theorems

N due to all the independent sources acting simultaneously is equal to the sum of
the sinusoidal steady-state responses due to each independent source acting one at a
time.
For linear time-invariant networks it is usually more convenient to use the
frequency-domain characterization. The superposition theorem can then be stated in
terms of network functions.

Example 13.6 Consider the linear time-invariant network shown in Figure 13.8. The
inputs are the two independent sources e1 and i2 , and the output is taken as the
voltage v across the resistor with resistance R2 . More precisely, v is the zero-state
response to e1  ·  and i2  ·  acting simultaneously on the network. Let V  s  , E1  s  ,
and I 2  s  be the Laplace transforms of the waveforms v , e1 , and i2 , respectively.
Then, the superposition theorem gives
V  s       
 H 1  s  E1  s   H 2  s  I 2  s 

Figure 13.8 Example 4: Illustration of the superposition theorem in terms of transfer functions.

The network function 𝐻𝐻1 (𝑠𝑠) is given by

V s  1
H1 s   
E1  s  I 2 0 1   R1  Ls G2  Cs 

Note that in calculating H 1  s  ,the current source i2 is set to zero. Similarly

V s  R1  Ls
H 2 s   
I 2 s  E1  0 1   R1  Ls G2  Cs 

Again in calculating H 2  s  , the voltage source e1 is set to zero

This idea can be generalized to any number of inputs. Since it is important in practice
we state it formally as a corollary.

Corollary 2
Let N be a linear time-invariant network. Let the response be the voltage across
any node pair or the current through any branch of N . More specifically, call X  s 
 13.3 Thévenin-Norton Equivalent Network Theorem ◀ 9

the Laplace transform of the zero-state response due to all the independent sources
acting simultaneously. Then
m
X s   H k s  I k s   
k 1

where h  s ,   k 1, 2, ... , m, are the Laplace transforms of the m inputs and H k  s  , k
 1, 2, ... , m , are the respective network functions from the m inputs to the specified
output.

13.3 Thévenin-Norton Equivalent Network Theorem

The Thévenin-Norton theorem is a powerful tool in calculating the response of
complicated networks. It has added importance because it gives a mental picture of
any linear network as seen from any two of its terminals. It is a very general theorem;
it applies to an extremely broad class of networks, and whatever the network, it gives
an equivalent network of the same form.

Figure 13.9 Circuit illustrating the conditions for application of the Thévenin-Norton equivalent network
theorem.

13.3.1 Theorem, Examples, Remarks, and Corollary

The Thévenin-Norton network theorem considers the following situation: a linear
network N is connected to an arbitrary load by two of its terminals,1 and 1' , as shown
in Figure 13.9. We assume that the only interaction between N and the load comes
from the current flowing through terminals 1 and 1' · In particular no other coupling
(e.g., magnetic or through dependent sources) is allowed between N and the load. It
is important to stress the fact that we make no assumptions concerning the load; it
may be nonlinear and/or time-varying. The network N is only required to be linear;
it may include both dependent and independent sources. In broad terms, the Thévenin-
Norton theorem asserts that the terminal current waveform i  ·  and the terminal
voltage waveform v  ·  will not be affected if N is replaced by either a “Thévenin
equivalent network” or by a “Norton equivalent network.”
The Thévenin equivalent network is shown in Figure 13.10a. It consists of a two-
terminal network N 0 in series with a voltage source eoc The waveform eoc  ·  of the
voltage source is the open-circuit voltage of N , that is, the voltage across the terminals
10 ▶Network Theorems

1 and 1' when the load is disconnected, as shown in Figure 13.10b. The voltage eoc is
caused by all the independent sources of N and the initial state; it is measured with
the polarity indicated on the figure. The network N 0 is obtained from the network N
by setting all independent sources to zero (i.e., by replacing every independent voltage
source by a short circuit and every independent current source by an open circuit) and
by setting all the initial conditions to zero. Note that the dependent sources reman
unchanged.

Figure 13.10 Thévenin equivalent network; (b) circuit defining the voltage source of the Thévenin
equivalent network; the subscript “oc” emphasizes that is an open-circuit voltage.

The Norton equivalent network is shown in Figure 13.11a. It consists of the same
two-terminal network N 0 placed in parallel with a current source isc · The waveform
isc  ·  of the current source is the short-circuit current of N , that is, the current
flowing in the short circuit which connects terminals 1 and 1' , as shown in Figure
13.11b. The current is caused by all the independent sources of N and the initial state;
it is measured with the polarity indicated in the figure.

Figure 13.11 Norton equivalent network; (b) the circuit defining current source of the Norton
equivalent network; the subscript “sc” emphasizes that is a short-circuit current
 13.3 Thévenin-Norton Equivalent Network Theorem ◀ 11

Thévenin-Norton Theorem
Let the linear network N be connected by two of its terminals 1 and 1' to an
arbitrary load. Let N consist of independent sources and linear resistors, linear
capacitors, linear inductors, linear transformers, and linear dependent sources. The
elements may be time-varying. We further assume that N has a unique solution when
it is terminated by the load, and when the load is replaced by an independent source.
Let N 0 be the network obtained from N by setting all independent sources to zero
and all initial conditions to zero. Let eoc be the open-circuit voltage of N observed at
terminals 1 and 1' , as shown in Figure 13.10b. Let isc be the short­circuit current of
N flowing out of 1 into 1' as shown in Figure 13.11b. Under these conditions, whatever
the load may be, the voltage waveform v  ·  across 1 and 1' and the current waveform
i  ·  through 1 and 1' remain unchanged when the network N is replaced by either
its Thévenin equivalent or by its Norton equivalent network.

Example 13.7 Consider the resistive circuit shown in Figure 13.12a. We want to
determine the voltage across the tunnel diode. We can use the Thévenin theorem and
consider the tunnel diode as the load. First we determine the Thévenin equivalent
network of the one-port faced by the tunnel diode. By inspection, the open-circuit
voltage is given by
R2E
eoc 
R1  R2

The network� N 0 is obtained by shorting out the battery; it is thus the series
combination of R3 with the parallel combination of R1 and R2 . The equivalent
resistance is therefore
R1R2
Req  R3 
R1  R2

Given eoc and Req , the Thévenin theorem asserts that v and i will remain
unchanged if we consider the circuit shown in Figure 13.12b. With the notations
defined on the figure, the terminal characteristic of the Thévenin equivalent network
is
v  eoc  Req i (13.4)

The circuit of Figure 13.12b can be solved by plotting the characteristic (13.4) on
the same graph as the tunnel-diode characteristic. Any intersection of the
characteristics will give one solution of the problem. The plot is shown in Figure 13.12c,
and the solution is obvious.

13.3.2 Spetial.Case
If the given linear network N is time-invariant it is more convenient to use network
function concepts. The Thévenin-Norton equivalent network theorem can be stated in
terms of the driving-point impedance or admittance of the “relaxed” network N 0 .
12 ▶Network Theorems

Figure 13.12 Example 1: a tunnel-diode circuit

Corollary
Let the linear time-invariant network N be connected by two of its terminals, 1
and 1' , to an arbitrary load. Let Eoc  s  be the Laplace transform of the open-circuit
voltage eoc  t  observed at terminals 1 and 1' , that is, the voltage when no current
flows into N through 1 and 1' . Let I sc  s  be the Laplace transform of the current
isc  t  flowing out of 1 and into 1' when the load is shorted. Let Zeq  
1  / Yeq be the
impedance (seen between terminals 1 and 1' ) of the network obtained from N by
setting all independent sources to zero and all initial conditions to zero. Under these
 13.3 Thévenin-Norton Equivalent Network Theorem ◀ 13

conditions, whatever the load may be, the voltage V  s  across 1 and 1' and the
current I  s  through 1 and 1' remain unchanged when the network N is replaced by
either its Thévenin equivalent network or its Norton equivalent network, as shown in
Figure 13.13. Furthermore,

Eoc  Zeq I sc (13.5)

Formula (13.5) is easily verified by referring to the definitions of Eoc and I sc ·
Indeed, if we replace the load in Figure 13.13a by a short circuit, the current I becomes
the short-circuit current I sc and (13.5) follows by KVL.

Figure 13.13 The Thevenin and Norton equivalent networks for linear time-invariant networks.

Dually, if we replace the load in Figure 13.13b by an open circuit, the voltage V
becomes the open-circuit voltage Eoc and KCL gives
I sc  Yeq Eoc (13.6)

Multiplying this equation by Zeq and noting that Zeq  s Yeq  s   1
 , we obtain (13.5)
again. The formula (13.5) is especially useful in computing the Thévenin equivalent
network in circuits with dependent sources.
14 ▶Network Theorems

Example 13.8 Consider the simple transistor amplifier shown in Figure 13.14a. Its
small­signal equivalent circuit is shown in Figure 13.14b. We want to determine the
amplifier voltage ratio VL / V0 and the output impedance (i.e., the impedance faced
by the load resistor RL at terminals 1 and 1' . Let us obtain the Thévenin equivalent
network faced by RL . We shall calculate the open-circuit voltage Eoc and the short-
circuit current I sc at terminals 1 and 1' . First, we convert the series connection of the
voltage source V0 and resistor rx into the parallel connection of a current source V0 / rx
and resistor rx .

Figure 13.14 A simple transistor amplifier and its small-signal equivalent circuit; the bias elements
, , , and (being large by design) are neglected in the small-signal equivalent circuit.

The open-circuit voltage Eoc can be found from the circuit, as shown in Figure 13.15a.
Using node analysis, we obtain
YV  I
 V0 
 gt  s C   C   sC   V1   
     r 
 sC  sC   V2   x 
    gmV1 
 13.3 Thévenin-Norton Equivalent Network Theorem ◀ 15

Vo 
 gt  s C   C   sC   V1   
    r 
 sC   gm  
sC   V2   x 
    0 

Figure 13.15 Example 2: derivation of the Thévenin and Norton equivalent networks

Thus
 Vo 
 gt  s C   C   sC   V1   
     r 
 sC   gm sC   V2   x 
    I sc 

gt  s C   C   V0 / rx
sC   gm 0
Eoc  V2 
gt  s C   C   sC 
sC   gm sC 

 sC   gm V0 / rx
 (13.7)
s 2C C   sC   gt  gm 
Next we compute I sc from the circuit in Figure 13.15b .
gt  sC  V0
IC  
gt  s C   C   rx
16 ▶Network Theorems

IC  gt  sC  V0 1
V1  
YC  gt  s C   C   rx sC   gt

1 V0

gt  s C   C   rx

V1sC   gmV1  I sc

 sC   gm  V0
I sc   sC   gm V1  (13.8)
s C   C    gt rx

Thus the equivalent impedance in the Thévenin equivalent network is

Eoc s C   C    gt
Zeq   (13.9)
I sc s 2C C   sC   gt  gm 

The voltage ratio of the amplifier is obtained next as follows

Eoc
VL  RL (13.10)
Zeq  RL

Combining (13.7) ,(13.9), and (13.10) we have

  sC   gm  
 
 r 
 x 
VL  s 2C C   sC   gt  gm  
   RL
 s C   C    gt 
V0  
 RL
 s 2C C   sC   gt  gm  
 
 sC   gm  / rx
 RL  
s 2C C RL  s C   C   C gt RL  C gm RL   gt

It is interesting to note that the network function relating VL to V0 has a zero in

the right-half plane; the network function is zero when s   
 gm / C  .

Alternate Method
We obtained the impedance 𝑍𝑍𝑒𝑒𝑒𝑒 in Eq.(13.9) by observing that it was the ratio of
𝐸𝐸𝑜𝑜𝑜𝑜 to 𝐼𝐼𝑠𝑠𝑠𝑠 , and we had already easily computed them. In general, the Thévenin
equivalent impedance can always be obtained by (1) setting to zero all the independent
sources of 𝔑𝔑, (2) connecting a “test” current source 𝐼𝐼𝑡𝑡 (𝑠𝑠) to terminals 1 and 1′, and
(3) using node analysis to calculate 𝑉𝑉𝑡𝑡 (𝑠𝑠), the Laplace transform of the zero-state
response to 𝐼𝐼𝑡𝑡 (𝑠𝑠). Then
Vt  s   Zeq  s  I t  s 

For this case the node equations are

 13.3 Thévenin-Norton Equivalent Network Theorem ◀ 17

 gt  s C   C   sC    E1  0
     
 sC   gm sC    Vt   It 
    
s C   C    gt
Vt  It s 
s 2C C   sC   gt  gm 

The ratio Vt  s  / I t  s  is Zeq  s  . It gives the same result for Zeq as before.

13.3.3 An Application of the Thévenin Equivalent Network Theorem

One important practical problem in system or circuit design is the study of
sensitivity of the system or circuit with respect to changes in its components. Such
changes may be due to manufacturing deviation, temperature effects, or aging. The
Thévenin theorem can be used conveniently for such a study. For simplicity, let us
consider a linear time-invariant network which is in the sinusoidal steady state.
Suppose that we know the solution, and we wish to determine the change of the current
in the kth branch due to a small change in the impedance of the kth branch, as shown
in Figure 13.16a. Let us denote the impedance of the kth branch by Z .The Thévenin
equivalent network is shown in Figure 13.16b, with the kth branch as the load. E 0
and Zeq are, respectively, the open-circuit voltage phasor and the equivalent
impedance. The sinusoidal current of branch k is represented by the phasor I ; this
phasor is given by
I  Zeq  Z   E 0 (13.11)

Let the kth branch be changed slightly so that the new impedance is Z  Z , and
let I '  I  I be the new current, as shown in Figure 13.16c. Since the remainder of
the network remains unchanged, we have the same Thévenin equivalent network as
before. Thus, the new current is given by
I   Zeq  Z  Z    I  I  Zeq  Z  Z   E 0 (13.12)

Substituting (13.11) in (13.12) , we obatin

I Z  I  Zeq  Z   I Z  0 (13.13)

Since Z is a small change, I Z represents a second-order term in (13.13) and can

be neglected. We therefore obtain
I Z
I   (13.14)
Zeq  Z
Or
I Z

  (13.15)
I Zeq  Z
Thus, we have obtained a simple relation between Z , the change in the impedance
Z, and I , the change in the branch current I . More significantly, this relation leads
to a simple interpretation of I ; the change in current, I , is equal to the current that
would flow if the only independent source in the circuit were a voltage source I Z
18 ▶Network Theorems

with the polarity indicated in Figure 13.17 ( Z is the change in the impedance and I
is the current that used to flow through Z). A moment of thought will indicate that
under this situation the measured voltages and currents in other branches of the
network are precisely the changes of voltages and currents caused in the original
network by the change Z in the kth branch.

Figure 13.16 Sensitivity study of a network using the Thevenin theorem; the impedance of branch k
changes from Z to .

Figure 13.17 A proposed measurement procedure to study the effect of a change in impedance
of branch k.
 13.4 The Reciprocity Theorem ◀ 19

13.4 The Reciprocity Theorem

Reciprocity is a property that is encountered very often in physics; it occurs in
electrostatics, in mechanics, in acoustics, etc. Therefore, we should not be astonished
that it comes up in circuits. Roughly speaking, when reciprocity applies to a physical
system, the input and the output can be interchanged without altering the response of
the system to a given input waveform. It is extremely important not only in the
analysis and design of systems but also in measurement techniques.
In electric circuits reciprocity applies to a subset of all linear time­invariant
networks. Applicable networks may have resistors, inductors, coupled inductors,
capacitors, and transformers; however, gyrators, dependent sources, and independent
sources are ruled out. We shall use the symbol NR to designate networks satisfying
these conditions (the sub­script R stands for reciprocity).
As an example, consider a telephone link between two points A and B. Suppose that
the circuit includes only elements from the allowed list. Note that since resistors are
included in the allowed list, the circuit may include negative-resistance amplifiers. On
the basis of only this information, the reciprocity theorem allows us to conclude that the
transmission from A to B is identical with the transmission from B to A. It is obvious
that this fact greatly simplifies the design and the testing of the telephone link.

Figure 13.18 Illustration of a soldering-iron entry-the wires are connected to existing nodes of the
network: (b) illustration of a pliers entry-the lead is cut, and the wires are soldered to the open
leads thus created.

Let NR be any network made of elements from the allowed list. The reciprocity
theorem concerns the zero-state response of NR to either an independent current source
or an independent voltage source. This theorem allows considerable freedom in the way
the independent source is applied and in the way the response is measured. A simple
way to visualize this freedom is the following: let us connect to NR two pairs of wires.
The first pair of wires will give us terminals 1 and 1' , and the second pair will give us
terminals 2 and 2 ' . We may choose to connect either or both pairs of wires to existing
nodes of NR · This connection is called, for obvious reasons, a soldering-iron entry. It
20 ▶Network Theorems

is illustrated in Figure 13.18a. We may also choose to connect either or both pairs of
wires as follows: we cut the lead of a branch and solder the wires to the terminals created
by the cut. This connection is called, for obvious reasons, a pliers entry. It is illustrated
in Figure 13.18b. In stating the theorem, we shall connect a source to 1 ,1' and measure
either the open-circuit voltage across 2, 2 ' or the current through a short circuit
connected to 2 , 2 ' .

13.4.1 Theorem,Examples and Remarks

Since in the following we shall have to consider the current waveform in a wire
under two different sets of inputs and since the same reference directions are used in
both cases, we shall use, say j2 for the current response to the first set of inputs and
jˆ2 for the current response to the second set of inputs. For voltages we use v1 and v̂1
, etc. In short, the network variables associated with the second set of inputs are
distinguished by the “hat” symbol �. We state the reciprocity theorem as follows.

Reciprocity Theorem
Consider a linear time-invariant network NR which consists of resistors, inductors,
coupled inductors, capacitors, and transformers only. NR is in the zero state and is
not degenerate. Connect four wires to NR thus obtaining two pairs of terminals 1 ,1'
and 2 , 2 ' .

Statement 1
Connect a voltage source e0  ·  to terminals 1,1' and observe the zero­state current
response j2  ·  in a short circuit connected to 2 , 2 ' (see Figure 13.19a). Next, connect
the same voltage source e0  ·  to terminals 2 ,2 and observe the zero-state current
response jˆ1  ·  in a short circuit connected to 1 ,1' . The reciprocity theorem asserts
that whatever the topology and the element values of the network NR and whatever
the wave­form e0  ·  of the source,
j2  t   jˆ1  t          
for all t
    

Figure 13.19 Statement 1 of the reciprocity theorem asserts that the wave forms and
are equal; observe that the currents and are short-circuit currents and note the reference
directions.
 13.4 The Reciprocity Theorem ◀ 21

Statement 2
Connect a current source i0 to terminals 1,1 and observe the zero-state voltage
response v2  ·  across the open-circuited terminals 2 , 2 (see Figure 13.20a). Next,
connect the same current source i0 to terminals 2 , 2 and observe the zero-state volt-
age response vˆ1  ·  across the open-circuited terminals 1 ,1 (see Figure 13.20b). The
reciprocity theorem asserts that whatever the topology and the element values of the
network NR and whatever the waveform i0  ·  of the source,
v2  t   vˆ1  t       
for all t
   

Figure 13.20 Statement 2 of the reciprocity theorem asserts that the waveforms and are
equal; note that and are open-circuit voltages and note also the reference directions.

Statement 3
Connect a current source i0 to terminals 1,1' and observe the zero-state current re-
sponse j2  ·  in a short circuit connected to 1, 1' (see Figure 13.21a). Next, connect a
voltage source e0 to terminals 2, 2 ' and observe the zero-state response vˆ1  ·  across
the open-circuited terminals 1,1' (see Figure 13.21b). The reciprocity theorem asserts
that whatever the topology and the element values of the network NR , and whatever
the waveform of the source, if i0  t  and e0  t  are equal for all t, then
vˆ1  t   j2  t      
  for all t

Figure 13.21 Statement 3 of the reciprocity theorem asserts that if the source waveforms and
are equal, then the zero-state responses and are also equal; note the reference
directions.
22 ▶Network Theorems

13.4.2 Reciprocity in terms of network functions

Since the reciprocity theorem deals exclusively with the zero-state response (includ-
ing the steady-state response as 𝑡𝑡 → ∞) of a linear time­invariant network, it is con-
venient to describe it in terms of network functions. The equivalent statements which
correspond to those stated in the theorem in terms of network functions are given
below.

Statement 1
Consider the two networks shown in Figure 13.19. In Figure 13.19a the input is a
voltage source 𝑒𝑒0 connected to terminal pair 1,1′, and the response is the short-circuit
current 𝑗𝑗2 . Call 𝐸𝐸0 (𝑠𝑠) and 𝐽𝐽2 (𝑠𝑠) the Laplace transforms of 𝑒𝑒0 and 𝑗𝑗2 , respectively. We
define the transfer admittance from 1 ,1′ to 2 , 2′ as
J 2 s 
y21  s  
E0 s 

In Figure 13.19b, the input is the same voltage source 𝑒𝑒0 connected to terminal pair
2 ,2′, and the response is the short-circuit current 𝚥𝚥1̂ . Using obvious notations, we
define the transfer admittance from 2, 2′ to 1, 1′ as
Jˆ1  s 
y12  s  
E0 s 

The reciprocity theorem asserts that

y21  s   y12  s          
for all s
Statement 2
Consider the two networks shown in Figure 13.20. In Figure 13.20a the input is a
current source 𝑖𝑖0 applied to terminal pair 1, 1′, and the response is the open-circuit
voltage 𝑣𝑣2 across terminal 2, 2′. Let 𝐼𝐼0 (𝑠𝑠) and 𝑉𝑉2 (𝑠𝑠) be the Laplace transforms of 𝑖𝑖0
and 𝑣𝑣2 , respectively. We define the transfer impedance from 1, 1′ to 2, 2′ as
V2  s 
z 21  s  
I 0 s 

In Figure 13.20b, the input is the same current source 𝑖𝑖0 applied to terminal pair 2,
2′, and the response is the open-circuit voltage 𝑣𝑣̂1 across terminal pair 1 ,1′. We define
the transfer impedance from 2,2′ to 1,1′ as
Vˆ1  s 
z12  s  
I 0 s 
The reciprocity theorem asserts that
z12  s   z 21  s        
for all s
Statement 3
Consider the two networks in Figure 13.21. In Figure 13.21a the input is a current
source i0 applied to the terminal pair 1, 1' , and the response is the short­circuit current
 13.4 The Reciprocity Theorem ◀ 23

j2 . We define the transfer current ratio

J 2 s 
H I s  
I 0 s 

In Figure 13.21b the input is a voltage source e0 applied to branch 2, and the
response is the voltage v̂1 across the node pair 1,1' . We define the transfer voltage
ratio as
Vˆ1  s 
HV  s  
E0 s 
The reciprocity theorem asserts that
H I  s   HV  s           
for all s
Example 13.9 The purpose of this example is to illustrate the full meaning of the
reciprocity theorem and, in particular, that it applies to dc conditions, sinusoidal
steady state, and transients. We consider the network NR shown in Figure 13.22.

Figure 13.22 Network used in Example 13.9.

Statement 1
Terminal pairs 1, 1' and 2, 2 ' are obtained by performing pliers entries in the resistive
branches with resistances of 5 and 1 ohm, respectively. Let us apply a constant voltage
source of 1 volt at 1, 1' and measure the current j2 in the short circuit joining 2, 2 '
(see Figure 13.23).
Suppose that we wish only to investigate dc conditions. Since, at direct current, the
capacitor is an open circuit and the inductor is a short circuit, we obtain, by inspection,
that j2 is a constant current of 1 amp. Next, as shown in Figure 13.23b, we apply
6
the same voltage source across terminals 2, 2 ' and calculate the current jˆ1 in the short
circuit joining 1 , 1' . We find jˆ1 = 1/6 amp, as required by Statement 1 of the theorem.

Statement 2
For the sake of variety, let us now pick soldering-iron entries. The terminals 1, 1 ,2
and 2 are shown in Figure 13.24. In terms of network functions, Statement 2 states
that the transfer impedance from node pair 1, 1' to 2, 2 ' is equal to the transfer
impedance from node pair 2 , 2 ' to 1, 1' · This means in particular that the
corresponding sinusoidal steady-state responses to the same sinusoidal current source
will be equal. Let us verify this fact for the network shown in Figure 13.22.
24 ▶Network Theorems

Figure 13.23 Network is used to illustrate Statement 1 of the reciprocity theorem under dc
conditions.

First, as shown in Figure 13.24a, we apply a sinusoidal current source

i0  t   2cos  2t   / 6  at node pair 1, 1' and calculate the open-circuit voltage at node
pair 2, 2 ' .Using phasor analysis, we write the node equations as follows:
 1 1 
 0.2  j 2     j   /6  
 j2 2 j  V1    2e  
    
 1 1 V
  2   0 
   1  
 2j 2j 
We obtain successively
j 0.5  2e j   /6   e  j   /3 
V2    0.542e  j 109.4
2
 0.2  j1.5  1  j 0.5    0.5 j  1.20  j 1.40

Thus, the sinusoidal steady-state response is

v2  t   0.542 cos  2t  109.4 

For the second experiment we apply the same current source to node pair 2,2 ' and
calculate the open-circuit voltage at node pair 1, 1' (see Fig. Figure 13.24b). The node
equations are
 13.4 The Reciprocity Theorem ◀ 25

Figure 13.24 Network is used to illustrate Statement 2 of the reciprocity theorem under
sinusoidal steady-state conditions.

 1 1 
 0.2  j 2     
 j2 2 j  V1    0


 1 1
 
 V2   2e  j   /6    
   1  
 2j 2j 
Then We obtain
j 0.5  2e j   /6  
Vˆ2   0.542e  j 109.4
2
 0.2  j1.5  1  j 0.5    0.5 j 
As before .The sinusoidal steady-state response is
vˆ1  t   0.542 cos  2t  109.4 

Statement 3
For variety, let us pick 1, 1' to be defined by a soldering-iron entry at the terminals
of the 5-ohm resistor, and pick 2, 2 ' to be defined as a pliers entry in the 1-ohm branch
(see ). Let the source waveform be a unit impulse applied at t = 0. To calculate the
required zero-state responses, we use Laplace transform.
First, as shown in Figure 13.25a, we apply a current source at node pair 1, 1   and
calculate the current j2 in the short circuit joining 2 , 2 ' . Since branch 2 is a 1-ohm
resistor, j2  t   
 v2  t  , so we need only calculate v2 By node analysis and Laplace
transform, we obtain
26 ▶Network Theorems

 1 1 
 0.2  s    V1  s   1
 s s      
  V s  0
 
1 1
 1   2     
 s s 
Hence
1/s 1
V2  s   
2
 0.2  s  1 / s  1  1 / s    1 / s  s2  1.2s  1.2

Taking the inverse Laplace transform and noting that j2  t   v2  t  , we obtain

 
 1   1  1 0.6t
j2  t   1    1   e sin 0.916 
t
 s 2  1.2s  1.2    s  0.6    0.916  
2 2 0.916
 
Where we used table 13.1 .Our conclusion is that
j2  t   1.09 e 0.6t sin 0.916           
t for t  0

Figure 13.25 Network is used to illustrate Statement 3 of the reciprocity theorem under
transient conditions.

For the second experiment, as shown in Figure 13.25b, we apply a voltage source
e0 with the same waveform (hence e0    ) across 2 , 2 ' , and we calculate the open -
circuit voltage across node pair 1, 1' · Using mesh analysis, we obtain
 13.4 The Reciprocity Theorem ◀ 27

 1 1 
5     Iˆ  s   0
 s s  1  
 1 1   Iˆ2  s   1
  1  s     
 s s 
And
1/s
Iˆ1  s  
2
 5  1 / s  s  1  1 / s    1 / s 
Since
5 5 1
Vˆ1  s    
 5s  1  s  1  1 / s   1 / s 5s 2  6s  6 s 2  1.2s  1.2
Recognizing this function of s to be the transform of 𝑗𝑗2 (𝑡𝑡), we use previous
calculations and conclude that
vˆ1  t   1.09 e 0.6t sin 0.916t        
for t  0

Thus, the two responses are equal, as required by the theorem.

Example 13.10 Consider the circuit shown in Figure 13.26 with the notations defined
on the figure. The current source i0 is applied across node pair 1,1′, and the open-
circuit voltage v2 is observed across node pair 2,2′. For i0  t   Iu  t  , the zero-state
response is
 t 




v2  t   R1gm R2I  1  e C 2         
R2 for t  0
 


Figure 13.26 Example 13.10: circuit containing a dependent source

28 ▶Network Theorems

If we interchange the source and the response, i.e., if we apply i0 across node pair 2,2′,
and observe the open-circuit voltage across node pair 1,1′, we see that
v̂1  t   0   
for t  0

This example shows that when dependent sources are present, reciprocity does not
hold, in general.

Example 13.11 Consider the circuit shown in Figure 13.27, which includes a
nonlinear element. Clearly, in Figure 13.27a we have v2  t   0.5u  t  . If we
interchange the input and the response as shown in Figure 13.27b, we get vˆ1  t   0 .
Thus, when nonlinear elements are present, reciprocity does not hold, in general.

Figure 13.27 :Example 13.11 circuit with a nonlinear element.

Example 13.12 We shall discuss heuristically the case of a linear circuit which
includes a time-varying resistor, as shown in Figure 13.28a. We shall show that
reciprocity does not hold. Consider only the steady-state response of the circuit to a
sinusoidal current input i0  t   cos10t . Note that the time­varying resistor is in
parallel with the current source and its resistance varies sinusoidally; that is,
R  t   1  0.1  
  cos t . The angular frequency of the source is 10 rad/sec, and that of the
resistor is 1 rad/sec. The effect of this time variation is to create a steady-state voltage
response v1 which contains both the sum and the difference of the two frequencies;
that is, 1  10  1  11rad / sec , and 2  10  1  9rad / sec . In the circuit, we see
that there are two lossless tuned circuits tuned at 1  11rad / sec And 2  9rad / s
, respectively . thus , at these two frequencies the impedance faced by the parallel
combination of the source and the time­varying resistor is infinite. Clearly, the voltage
response v2 across the output 1-ohm resistor does not contain sinusoids at these two
frequencies.
 13.4 The Reciprocity Theorem ◀ 29

Figure 13.28 Example 13.12 : circuit with a linear time-varying resistor.

Next let us interchange the input and the response, as shown in Figure 13.28b. The
input is now applied across the 1-ohm resistor and the response is taken as the voltage
v1 across the time-varying resistor. Let us take advantage of the Thevenin equivalent
network theorem and consider the time-varying resistor as the load, as shown in Figure
13.28c. Obviousely,the equivalent source Eoc has a steady-state component at
s  10 /rad s and Zeq is finite at s  10rad / sec . This equivalentsource combined
with the effect of the rime varying resistor will produce a volateg v̂1 which contains
sinusoids at 1  11 /
rad s and 2  9 / rad s . Thus v̂1 cannot be identical to v2 of
the previous case.this example shows that when time-varying elements are present,
reciprocity does not hold, in general. (See also problem 20.)
30 ▶Network Theorems

Remarks
Any network which satisfies the reciprocity theorem is called a reciprocal network.
Using this concept, we see that the reciprocity theorem guaran­tees that any network
made of linear time-invariant resistors, capacitors, inductors, coupled inductors, and
transformers is a reciprocal network. It is also a fact that some linear time-invariant
networks that contain de­pendent sources are reciprocal, whereas others are not.

Example 13.13 Some texts suggest that all networks made of passive linear time-
invariant elements are reciprocal. This is false. To show this, we introduce a new two-
port element called a gyrator. By definition, a gyrator is a two-port element (shown in
Figure 13.29) described by the equations
v1  t   i2  t 

v2  t   i1  t 
Or, in matrix form,
 v1  t    0    i1  t  
   
 v2  t     0   i2  t  
     
In both equations, the number  is a constant and is called the gyration ratio.
From these equations it follows that the gyrator is a linear time­invariant element. It
is linear because its branch voltages v1 and v2 are given by linear functions of its
branch currents. Furthermore, from the defining equations, it follows that the power
delivered by the external world to the gyrator is, for all t,
v1  t  i1  t   v2  t  i2  t   0

Thus, the gyrator neither absorbs nor delivers energy to the outside world. It is
worth noting that the ideal transformer also has these three properties. However, in
contrast to the ideal transformer, the gyrator as a network element does not obey the
reciprocity theorem. Hence, we say that the gyrator is not reciprocal. We check this
fact as follows: we apply Statement 2 of the reciprocity theorem to the terminals 1,1′ , 2
and 2′ shown in Figure 13.29. With i1  1 amp and i2  0 (a 1-amp current source
at 1,1′ and an open circuit at 2,2′ we have v2   volts; with i2  1 amp and i1  0
, we have v1   volts. If the gyrator were reciprocal, we would have v1  v2 . In
conclusion, as far as linear time­invariant networks are concerned, passivity of every
network element is not sufficient to guarantee that the network is reciprocal.

Figure 13.29 Symbolic representation of

the gyrator; the constant a is called the
gyration ratio.