Introduction

Frequency Domain
In principle, the voltages and currents in analog circuits are arbitrary functions of time (we
call them signals or waveforms). Analytical analysis of the circuit response to an arbitrary
input waveform is difficult and requires solution to a set of differential equations. Even
numerical analysis becomes difficult when there are a lot of circuit elements. Fortunately,
there are ways to find the response of a linear circuit to time-dependent signal. These
approaches are based on the following observations:
1. For circuits driven by sinusoidal sources, the forced response of the state vari-
ables (currents and voltages) are all sinusoidal functions with the same frequency
as the source. We can solve these circuits easily using Phasors, for example
This is derived from the mathematical properties of sinusoidal functions. Forced response of
a set of linear differential equation (circuit equations) to a sinusoidal function is a sinusoidal
function. This property leads to special analysis tools for AC circuits using “phasors,” or
using Fourier transform. AC steady-state analysis of linear circuits are covered in 60A/B.
When we use phasors, the circuit equation do not contain time anymore, but they include
frequency ω. As such, this is usually called analysis in “frequency-domain” to differentiate
that from “time-domain” analysis where we solve the differential equation to find the circuit
response.
2. Any arbitrary but periodic signal can be written as a sum of sinusoidal
functions using Fourier series expansion.
For example, a square wave with period T or frequency ω0 = (2π)/T and amplitude Vm can
be written as:

4Vm 1 1
 
v(t) = sin(ω0 t) + sin(3ω0 t) + sin(5ω0 t) + ...
π 3 5

Signals with frequencies nω0 (n integer) are called harmonics of the fundamental frequency,
ω0 . In general the amplitude of higher harmonics become smaller as n become larger. The
idea of decomposition of a periodic function to a sum of sinusoidal functions can be extended
to an arbitrary temporal function by using Fourier integrals. As such, in principle, any
function of time can be written as a sum of (or an integral of) sinusoidal functions.
3. Proportionality and superposition principles state that response of a linear
circuit to a linear combination of sources is equivalent to the linear combination
of circuit response to each individual source.

ECE60L Lecture Notes, Spring 2002 1

Basically, in a circuit with several independent sources, the value of any state variable equals
to the algebraic sum of the individual contributions from each independent source. So, in a
circuit with a time-dependent source, we can use Fourier series decomposition and replace the
source with a linear combination of several sinusoidal sources. We can then find the response
of the circuit to each sinusoidal source and then use proportionality and superposition to
find the response to the time-dependent source.
For example, suppose we want have a circuit driven by a source that can be decomposed into
v(t) = A cos(100t) + B cos(300t). We want to know the voltage across an element, v̄(t). We
solve the circuit with the source cos(100t) and find the voltage across the element interest,
suppose α cos(100t + φα ). We then repeat the analysis with a source cos(300t) and find the
voltage across the element interest, suppose β cos(300t + φβ ). The response of the circuit to
v(t) = A cos(100t) + B cos(300t), then is v̄(t) = Aα cos(100t + φα ) + Bβ cos(300t + φβ ).
The problem is actually much simpler than the example above. In principle, solution of AC
steady-state circuit is simple and we typically find the response the circuit with frequency,
ω as a parameter. We can then construct the response by replacing ω with frequencies
of interest in the response equation (e.g., set ω = 100 and 300 in the above example).
Another major simplification arises when the circuit response is frequency independent. In
that case, the circuit response can be directly applied to any time-dependent function. For
example, in the above example, if the circuit response to cos(100t) and cos(300t) sources were,
respectively, α cos(100t) and α cos(300t) (frequency independent), then the circuit response
is simply: v̄(t) = αv(t).
Therefore, we focus on circuits driven by sinusoidal sources. We solve these
circuits in frequency domain. We try to find circuit parameters with frequency
ω as a parameter to facilitate construction of response to an arbitrary function
of time. In particular, we are very interested in regimes in which the circuit re-
sponse is independent of the frequency as the output wave-forms will be identical
to input signals.
There are several ways to solve the circuit in frequency domain, all having same mathematical
foundation. We can use phasors (which are really Fourier Transforms). Or, we can use
complex frequency domain which is sometimes called “s-domain” (s = σ + jω). In junior
level courses and beyond, you will probably use complex frequency domain mainly. Circuit
analysis with phasors is sufficient for the work we do in this class.
Analysis in frequency domain is straight-forward. Resistors, capacitors, and inductors are
replaced by impedances, Z: Z = R for a resistor, Z = 1/(jωC) for a capacitor and Z = jωL
for an inductor. Impedances obey Ohm’s Law: V = ZI. Thus, with impedances, the circuit
reduces to a “resistive” circuit and all analysis techniques of resistive circuits (node-voltage

ECE60L Lecture Notes, Spring 2002 2

method, mesh-current method, Thevenin Theorem, etc.) apply. The only difference is that
analysis is performed using complex variables.
Circuit Components

It is not practical to design a complete circuit as a whole from scratch. It is usually much
easier to break the circuit into components and design and analyze each component sepa-
rately. In this manner we can design “building blocks” (such as amplifiers, filters, etc.) that
can be used in a variety of devices. A typical analog circuit is composed of a “source,” one
or several “two-port networks,” and a “load.”

Source 2-port 2-port Load

Network Network

Thevenin Theorem provides the corner stone of the strategy to divide the circuit into compo-
nents and analyze each independently. We can see this by examining the Thevenin Theorem.
Thevenin Theorem and Thevenin or Norton Equivalents
I I
I VT + IN +
+ ZT
+ V ZN V
V
-
-
- -

We know from linear circuit theory (See Textbook for 60A) that the IV characteristics of a
two-terminal network is in the form of (using active sign convention):

V = Vt − Zt I; Zt = Zn ; In Zn = V t

which is similar to the IV characteristics of the Thevenin or Norton forms shown above.
Therefore, any two-terminal network can be replaced by its Thevenin or Norton equivalent.
An important corollary to the Thevenin Theorem is that if a two-terminal network does
not include an “independent source” it will be reduced to a single “impedance” (even if it
includes dependent sources).
How to calculate the Thevenin equivalent: You have seen a detailed discussion of
Thevenin/Norton forms in ECE60A. In summary, the best method is to calculate two of the
the following three parameters: (1) Open-circuit voltage, Voc (found by setting I = 0) , (2)

ECE60L Lecture Notes, Spring 2002 3

Short-circuit current, Isc (found by shorting the terminals of the two-terminal network, i.e.,
setting V = 0), and (3) Direct calculation of Zt which is the resistance seen at the terminal
with the independent sources “killed” (i.e., their strength set equal to zero). Remember, you
should NOT kill dependent sources. The usual “rule of thumb” is to find Voc and Isc if there
is a dependent source in the problem, and to find Voc and Zt if there is no dependent source
in the problem. Then, one can find the Thevenin and Norton parameters from:

Vt = Voc ; In = Isc ; In Zt = V t

Example: Find the Thevenin and Norton Equivalent of the circuit below:

We need to find two of the three parame- 5Ω 4Ω I

ters Voc , Isc , and Zt (Rt here). It is best to 25 V +

find Voc and Zt for this problem (no depen- + 20 Ω 3A V
dent source) but all three are calculated for -
demonstration of the solution technique. -

1. Voc : Using node-voltage method and 25 5Ω V1 4Ω I=0

noting that since I = 0, by KVL, V1 = Voc . +

25 V
+ 20 Ω V
V1 − 25 V1 -
3A oc
−3+ =0
5 20 -
4V1 − 100 − 60 + V1 = 0
V1 = 32V → Voc = Vt = V1 = 32V

5Ω 4Ω
2. Rt (killing the independent sources)
From the circuit, we have
20 Ω
Rt
Rt = 4 + (5 k 20) = 4 + 4 = 8 Ω

3. Isc To calculate Isc by nodal analysis,

note that Isc = V1 /4. Then, 25 5Ω V1 4Ω

25 V
V1 − 25 V1 V1 +
+ −3+ =0 20 Ω 3A I sc
5 4 20 -
4V1 − 100 + 5V1 − 60 + V1 = 0
V1
V1 = 16V → In = Isc = = 4A
4

So, the Thevenin/Norton parameters are: Vt = 32 V, In = 4 A, and Rt = 8 Ω. (note,

Vt = In Rt .)

ECE60L Lecture Notes, Spring 2002 4

How to measure the Thevenin equivalent: Suppose we have given a box with two
terminals and want to find the Thevenin equivalent of the circuit inside the box. In principle,
we cannot use the above technique and try to measure Voc , Isc , and Zt . We cannot turn off
the input signal and use a ohm-meter to measure Rt . Nor can we short the terminals and
measure Isc (there is a good chance that we are going to ruin the circuit if we do that). In
principle, we can use a volt-meter (or scope) to measure Voc but care should be taken as it is
not known a priori if the internal resistance of the volt-meter (or scope) is large enough to
act as an open circuit (there are other complications). There is also the issue of measurement
error that one should consider
I
Instead of measuring Voc , Isc , and Zt directly, it is best +
to measure the IV characteristics of the two-terminal V R
network. We can do this by attaching a variable load -
(a resistance) to the box, vary the load which changes
the output voltage and currents, and measure several I

pair of I and V . These data point should lie on the

IV line of the two-terminal network. Values of Vt ,
Measurements
In , and Rt can be read directly from the graph as In
shown. This method is much better as by using a
“best-fit” line to our data, we can minimize random
measurement errors. slope of -1/R t

V
Vt

How to find the Thevenin equivalent using PSpice: You can use the same technique
described above for measuring the Thevenin parameters with PSpice. Attach a “variable”
load to the circuit. Ask PSpice to compute output voltage V as a function of load resistance
RL . Use PROBE to plot the output current I versus the output voltage V and you will have
the IV characteristics of the circuit.

ECE60L Lecture Notes, Spring 2002 5

How each sub-circuit sees other elements

The strategy of dividing a circuit into individual components works because of the Thevenin
Theorem. Recall that any two-terminal network can be replaced by its Thevenin equivalent.
In addition, if a two-terminal network does not include an “independent source” it will be
reduced to a single “impedance” (even if it includes dependent sources).

Source 2-port 2-port Load

Network Network

Load sees this

Source see this two-terminal network two-terminal network

What Source sees: The source sees a two- +

terminal network. This two-terminal network Source V ZL

does not contain an independent source. So it -

can be reduced to a single impedance.
Zt Ii
What Load sees: The load sees a two- Vt
+
terminal network. This two-terminal network +
Vi Load
contains an independent source. So it can be -
-
reduced to its Thevenin equivalent.

What each two-port network sees: Fol- Zs Ii Io

lowing the logic above, its obvious that each Vs
+ +
two-port network sees a two-terminal network + 2-port ZL
Vi Network Vo
containing an independent source in the input -
- -
side (can be reduced to a Thevenin form) and
a two-terminal network that does not contain
an independent source on the output side (so
it can be reduced to a single impedance).
The above observations indicate that we do not need to solve a complete circuit. For example,
for a particular two-port network, we only need to solve the circuit above (with V s , Zs , and
ZL as parameters). Then, wherever this two-port network appears in a circuit, we can use
these results.

ECE60L Lecture Notes, Spring 2002 6

Two-port networks
As you can imagine, majority of components in electronic circuits are two-port networks.
For example, in a tape recorder, a large number of two-port networks exists between the
source (tape head) and the load (speakers). They amplify the signal, filter out the unwanted
noise, and process the signal. We study several two-port networks in 60L. As we noted in the
previous page, we can design and analyze these two-port networks using a simple model for
the previous stages and a load impedance for later stages of the system as is shown below.
Consider the two-port network below. It “communicates” with the outside world (rest of the
circuit) through 4 parameters: Vi , Ii , Vo , and Io . If we solve the two-port network circuit once
and find the relationship between these four parameters, we do not need to do that again.
While any linear two-port network can be reduced to a combination of four elements (see
your circuit theory textbook), it is customary to use the following parameters to describe
the behavior of a two-port network.
Vo
Voltage transfer function, Hv (jω) =
Vi Zs Ii Io
Io Vs
+ +
Current transfer function, Hi (jω) =
Ii + 2-port ZL
Vi Network Vo
Vi -
Equivalent input impedance, Zi (jω) = - -
Ii
Vo 

Equivalent output impedance, Zo (jω) =
Io Vs =0

The equivalent output impedance as defined above is the equivalent Thevenin impedance of
a two-terminal network consisting of our 2-port network, Zs , and Vs .
What are ZL and Zs : Consider a circuit in which our two-port network above is the “nth”
two-port network (see figure in the previous page). In this case, the output voltage of
“n-1” two-port network is the same as the input voltage of our “nth” two port network:
Vo |n−1 = Vi |n and the output voltage of our “nth” two-port network is the input voltage
to the “(n+1)th” two-port network: Vo |n = Vi |n+1 (with the similar relationship between
the currents). The transfer function definitions above indicate that ZL is actually the input
impedance of “(n+1)th” two-port network (next stage) and Zs is the output impedance of
“(n-1)th” two-port network (previous stage).
Interaction between Components & Voltage and Power Transfer
It is obvious from the definitions of transfer functions above that they depend on the values
of ZL and Zs . This means that when a two-port network is placed in a circuit, the output
impedance of the previous stage (Zs = Zi |n−1 ) and the input impedance of the next stage
(ZL = Zi |n+1 ) affect the two-port network transfer functions.

ECE60L Lecture Notes, Spring 2002 7

One is interested to find if there is a way to ensure good coupling and maximum signal
transfer between connecting two-port networks. Consider the connection between two two-
port network as is shown below.
As we are interested in the interaction between the two-
port network, we replace the “nth” two-port network and
nth (n+1)th
all of the circuit to its left with its Thevenin equivalent. Stage Stage
From definitions of transfer functions above, we know
that Zs is actually the output impedance of the “nth”
two-port network Zs = Zo |n . Similarly the “(n+1)th” Zs IL

two-port network and all of the circuitry to its right can Vs

+
be replaced by its Thevenin equivalent, ZL (as there is +
VL ZL
−
no independent source in there). Again, we note that −
ZL = Zi |n+1 nth (n+1)th
Stage Stage

Good coupling between components typically means largest IL , VL or power, PL = VL IL .

Unfortunately, these three parameters do not maximize simultaneously.

Vs ZL ZL
IL = VL = Vs PL = V L I L = V2
Zs + Z L Zs + Z L (Zs + ZL )2 s

Values of IL , VL , and PL are plotted in the fig-

ure assuming Vs and Zs are fixed. We can see
that best current coupling (maximum IL ) when
ZL = 0 (or effectively, ZL /ZS  1) and the best I Load
V Load
voltage coupling (maximum VL ) when ZL → ∞ P Load

(or effectively, ZL /ZS  1). The best power

coupling (maximum PL ) is somewhere in between
when ZL = Zs∗ (RL = Rs and XL = Xs ).

0 1 2 3 4 5 6
RL/Rs
ZL
Maximum voltage transfer: 1 → VL |max = Vs
ZS
ZL Vs
Maximum current transfer: 1 → IL |max =
ZS Zs
Vs2
Maximum power transfer: ZL = Zs∗ → PL |max =
4Zs

Maximum power transfer is not usually a criteria for coupling components (except the last
stage of coupling to the load). In most cases, we are interested in good voltage coupling to

ECE60L Lecture Notes, Spring 2002 8

keep power dissipation in the circuit small. For example, consider a CD player. The source
produces a low-voltage signal proportional to the information on the CD which needs to be
amplified, translated into sound frequencies, amplified further, filtered, etc. and then fed to
a speaker. In order to keep the circuit small and cheap, we amplify the signal and do the
signal processing with signals of substantial voltage but low current. This keep the power
dissipation in each stage small. Only in the last stage (power amplifier) the signal current
is increased to drive the load (speakers).
The criteria for best voltage coupling is ZL  Zs (VL ≈ Vs ). If we are modeling the inter-
action between two two-port networks, Zs represent the output impedance of the previous
stage, ZL represents the input resistance of the next stage. Therefore, best voltage coupling
condition translates into ensuring that output impedance of previous stage is much
smaller than input impedance of the next stage.:

Zo |n  Zi |n+1

And, a useful goal for designing two-port networks is to ensure that input impedance is
large and output impedance is small.

Zs Ii Io
In this course, we examine many two- Vs
+ +
port networks, calculate their parameters
+ 2-port ZL
(transfer functions and input and output -
Vi Network Vo
impedances), and experiment with them in - -
the Lab.

We showed that in general we need to solve a circuit as is shown. Examination of the defini-
tions of parameters of the two-port network shows that only Zo depends on Zs . Furthermore,
if we follow the good practice of designing circuits with low output impedance and high input
impedance, one can easily show that Zo will become independent of Zs . So, for the rest of
this course, Zs is ignored (then, Vs = Vi ). Furthermore, to facilitate understanding of circuit
behavior, we will first solve all of the circuits assuming ZL → ∞ (this is the same as setting
Zo |n  Zi |n+1 . We will then investigate the impact of adding a load to the circuit (called
terminated networks).

ECE60L Lecture Notes, Spring 2002 9

Voltage divider as an example of a two-port network
Vs Ii
Ii =
Rs + R 1 + R 2 k R L +
Rs
Vo Vs
R
1
Io = −
RL +
Io

R2 k R L Vi
Vo = (R2 k RL )Ii = Vs - +
Rs + R 1 + R 2 k R L
R Vo R
2 L
R1 + R 2 k R L
Vi = (R1 + R2 k RL )Ii = Vs - -
Rs + R 1 + R 2 k R L

From the above we can now calculate the parameters of our two-port network:
Vo R2 k R L Io R2 k R L
Hv = = Hi = =
Vi R1 + R 2 k R L Ii RL
Vi Vo 

Zi = = R1 + R2 k RL Zo = = (Rs + R1 ) k R2
Ii Io Vs =0

Note that the transfer functions are independent of the value of input signal strength, Vi
(because the circuit is linear) and the parameters of the two-port network “depend” on values
of Rs and RL . For this case, the transfer functions are frequency independent (no capacitor
or inductor in the circuit).
Condition of best voltage coupling is RL  Zo (this means that Hv will be independent of
RL and Hv is at its maximum value). For simplicity, let’s consider a case with Rs = 0. In
this case,
R1 R2
RL  Z o = → R L R1 + R L R2  R 1 R2
R1 + R 2
Replacing for R2 k RL = R2 RL /(R2 + RL ) in Hv , we get
R2 RL
Hv (jω) =
R1 R2 + R 1 RL + R 2 RL
Using RL R1 + RL R2  R1 R2 , we can drop R1 R2 term in the denominator of Hv :
R2 RL R2
Hv (jω) ≈ =
R1 RL + R 2 RL R1 + R 2
which is independent of RL . Note that the value of Hv is exactly the value of Hv in absence
of any load (i.e., RL → ∞). It is also straight forward to show that this value represent
the maximum value of Hv in the presence of any load and, thus, represents the best voltage
coupling.

ECE60L Lecture Notes, Spring 2002 10