Transfer Functions
Transfer functions dened
Examples
System stability
Pole-Zero Plots
Sinusoidal steady-state analysis
Linearity and time invariance dened
Transfer function synthesis
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
Transfer Functions
Assume zero initial conditions.
N
k=0
N
dk y(t)
ak
dtk
ak sk Y (s)
k=0
k=0
Y (s)
M
dk x(t)
bk
dtk
bk sk X(s)
k=0
N
ak sk
= X(s)
k=0
Y (s)
J. McNames
M
M
bk sk
k=0
M
k
k=0 bk s
N
k
k=0 ak s
Portland State University
ECE 222
X(s) = H(s)X(s)
Transfer Functions
Ver. 1.67
Initial Conditions
Assume zero initial conditions.
N
k=0
N
dk y(t)
ak
dtk
ak sk Y (s)
k=0
M
k=0
M
dk x(t)
bk
dtk
bk sk X(s)
k=0
All voltages and currents are due to independent sources
(superposition)
Energy stored in capacitors and inductors also act like independent
sources
We will now focus a specic class of circuits
Only one independent source (input)
No energy stored in capacitors or inductors
Greatly simplies analysis
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
Transfer Functions Continued
M
k
k=0 bk s
X(s) = H(s)X(s)
Y (s) = N
k
k=0 ak s
In the time domain, the relationship can be complicated
In the s domain, the relationship of Y (s) to X(s) of LTI systems
simplies to a rational function of s
H(s) is usually a rational ratio of two polynomials
H(s) is called the transfer function
Specically, the transfer function of an LTI system can be dened
as the ratio of Y (s) to X(s)
Usually denoted by H(s), sometimes G(s)
Without loss of generality, usually aN 1
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
Example 1: Transfer Function vs. Impulse Response
Fill in the missing parts to determine how the transfer function of an
LTI system G(s) is related to the impulse response h(t)
x(t) = (t)
X(s) =
y(t) =
Y (s) =
L {h(t)} =
L1 {G(s)} =
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
Transfer Functions and the Impulse Response
x(t)
h(t)
y(t)
x(t)
H(s)
y(t)
Because of their relationship, both H(s) and h(t) completely
characterize the LTI system
If the LTI system is a circuit, once you know either H(s) or h(t),
you have sucient information to calculate the output
You now have three dierent approaches to solve for the output of
an LTI circuit
y(t) = x(t) h(t)
Solve for H(s), X(s), and then y(t) = L1 {H(s) X(s)}
Use Laplace transform circuit analysis to solve for the outputs
of interest
All three have limitations, advantages, and disadvantages
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
Continous-Time Convolution Tradeos
x(t)
h(t)
y(t)
x(t)
H(s)
y(t)
Continuous-time Convolution: y(t) = x(t) h(t)
Advantages
Can nd solution for all t, not just t > 0
Can be approximated using discrete-time convolution
Disadvantages
Cannot account for non-zero initial conditions, requires
complete x(t) and y(t)
Can be dicult to write and solve integrals
Can only be used for single-input single-output (SISO) systems
that have one independent source
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
Transfer Function Analysis Tradeos
x(t)
h(t)
y(t)
H(s)
x(t)
y(t)
Transfer Function: y(t) = L1 {H(s) X(s)}
Advantages
Reduces dierential equation to an algebra problem
Usually the easiest approach
Easy to nd the output for dierent input signals
Disadvantages
Can only solve for y(t) for t > 0
Requires zero initial conditions
Can only be used for SISO systems
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
Laplace Transform Circuit Analysis Tradeos
x(t)
h(t)
y(t)
x(t)
H(s)
y(t)
Laplace Transform Circuit Analysis
Advantages
Elegant method of handling non-zero initial conditions
Can handle multiple sources (multiple inputs) & can solve for
multiple outputs (any voltage or current) MIMO systems
Disadvantages
Can only solve for y(t) for t > 0
Cannot account for full history, x(t) for t < 0. Requires this
eect to be captured in the initial conditions
Can be tedious
Specic to application (circuits), we did not discuss
generalization to other types of systems
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
Example 6: Transfer Functions
R
+
vs(t)
vo(t)
-
Find the transfer function for the circuit above. The input is the
voltage source vs (t) and the output is labeled vo (t).
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
10
Example 7: Transfer Functions
R
C
+
vs(t)
vo(t)
-
Find the transfer function for the circuit above. Do you recognize this
function?
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
11
Example 8: Transfer Functions
C
R
+
vs(t)
vo(t)
-
Find the transfer function for the circuit above. Do you recognize this
function?
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
12
Example 9: Transfer Functions
CB
CA
RB
RA
+
vs(t)
vo(t)
RL
Find the transfer function for the circuit above.
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
13
Example 9: Workspace
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
14
Example 10: Transfer Functions
+ vR(t)
R
vs(t)
+ vL(t)
vC(t)
-
Find the transfer function from the input voltage to an output voltage
across each element of the three passive elements in a series RLC
circuit.
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
15
Example 10: Workspace
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
16
Poles and Stability
Assume all of the poles in a transfer function H(s) are unique. Then
H(s) can be written as follows using partial fraction expansion:
N
N (s) k
H(s) =
=
D(s)
s p
=1
L1 {H(s)}
= h(t) =
N
k e+p t u(t)
=1
Note the expansion is in terms of the poles, rather than p
If |h(t)| dt < , the LTI system is bounded-input
bounded-output (BIBO) stable
That is |h(t)| < < for all t
h(t) is bounded if Re{p } < 0 for all
The system is BIBO stable if and only if all the poles are in the
left half of the complex plane
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
17
Pole-Zero Plots
H(s) =
N (s)
D(s)
Zeros: roots of N (s)
Poles: roots of D(s)
Poles must be in the left half plane for the system to be stable
As the poles get closer to the boundary, the system becomes less
stable
Pole-Zero Plot: plot of the zeros and poles on the complex s
plane
You will use these throughout the junior sequence (ECE 32x)
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
18
Example 11: Pole-Zero Plots
Use MATLAB to generate a Pole-Zero plot for a system with the
following transfer function
s2 1
H(s) = 3
s + 4s2 + 6s + 4
Using the MATLAB, we can quickly nd the roots
H(s) =
(s + 1)(s 1)
(s + 2)(s + 1 j)(s + 1 + j)
Is the system stable? The pole-zero plot, impulse response, and step
response are shown on the following slides.
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
19
Example 11: Pole-Zero Plot
PoleZero Plot
1.5
Imaginary Axis
0.5
0.5
1.5
3
J. McNames
2.5
1.5
Portland State University
0.5
0
Real Axis
ECE 222
0.5
Transfer Functions
1.5
Ver. 1.67
20
Example 11: MATLAB Code for Pole-Zero Plot
sys = tf([1 0 -1],[1 4 6 4]);
figure;
[p,z] = pzmap(sys);
h = plot(real(p),imag(p),bx,real(z),imag(z),ro);
set(h,LineWidth,1.2);
set(h,MarkerSize,5);
hold on;
plot([0 0],[-2 2],k:,[-3 2],[0 0],k:);
hold off;
xlabel(Real Axis);
ylabel(Imaginary Axis);
title(Pole-Zero Plot);
axis([-3 2 -1.5 1.5]);
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
21
Example 11: Impulse Response
Impulse Response
1
0.8
h(t)
0.6
0.4
0.2
0
0.2
0.4
0
J. McNames
Portland State University
3
4
Time (seconds)
ECE 222
Transfer Functions
Ver. 1.67
22
Example 11: MATLAB Code for Impulse Response
sys = tf([1 0 -1],[1 4 6 4]);
figure;
t = 0:0.01:7;
[h,t] = impulse(sys,t);
h = plot(t,h);
set(h,LineWidth,1.5);
hold on;
plot([0 0],[-2 2],k:,[0 max(t)],[0 0],k:);
hold off;
axis([0 max(t) -0.5 1.2]);
xlabel(Time (seconds));
ylabel(h(t));
title(Impulse Response);
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
23
Example 11: Step Response
Step Response
0.1
0
y(t)
0.1
0.2
0.3
0.4
0.5
J. McNames
Portland State University
3
4
Time (seconds)
ECE 222
Transfer Functions
Ver. 1.67
24
Example 11: MATLAB Code for Step Response
sys = tf([1 0 -1],[1 4 6 4]);
figure;
t = 0:0.01:7;
[h,t] = step(sys,t);
h = plot(t,h);
set(h,LineWidth,1.5);
hold on;
plot([0 0],[-2 2],k:,[0 max(t)],[0 0],k:);
hold off;
axis([0 max(t) -0.5 0.2]);
xlabel(Time (seconds));
ylabel(y(t));
title(Step Response);
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
25
Steady-State Sinusoidal Analysis
Assume a system H(s) is BIBO stable. Consider a sinusoidal input
x(t)
=
=
L
cos(t)
sin(t)
X(s)
=
=
J. McNames
Y (s)
Y (s)
A cos(t + )
A cos() cos(t) A sin() sin(t)
s
s2 + 2
s2 + 2
s
A sin() 2
A cos() 2
s + 2
s + 2
A [s cos() sin()]
s2 + 2
H(s)X(s)
A [s cos() sin()]
H(s)
s2 + 2
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
26
Steady-State Sinusoidal Analysis Continued
A [s cos() sin()]
Y (s) = H(s)
s2 + 2
N
k
k
k
=
+
+
s j s + j
s + p
=1
y(t)
2|k| cos(t + k)u(t) +
N
k ept u(t)
=1
= yss (t) + ytr (t)
yss (t)
lim y(t)
= 2|k| cos(t + k)
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
27
Steady-State Sinusoidal Analysis Comments
If x(t) = A cos(t + ),
yss (t) = lim y(t) = 2|k| cos(t + k)
t
If the input to an LTI system is sinusoidal,
The steady-state output is sinusoidal at the same frequency
The amplitude and phase of y(t) dier from that of x(t)
We applied this idea when we did phasor analysis
But how is k related to H(s), A, and ?
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
28
Solving for the Complex Residue
A (s cos sin )
Y (s) = H(s)
s2 + 2
N
k
k
k
=
+
+
s j s + j
s + p
=1
A [s cos() sin()]
k = H(s)
s + j
s=+j
A [j cos() sin()]
= H(j)
2j
A [cos() + j sin()]
= H(j)
2
= 12 H(j)A ej
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
29
Sinusoidal Steady-State Output
Since H(j) is complex, we can write it in polar form as
H(j) = |H(j)| ejH(j)
Then using the results of the previous slide, we have
k = 12 H(j)A ej
|k| = 12 |H(j)|A
yss (t)
= 12 |H(j)|A ej(+H(j))
k = + H(j)
2|k| cos(t + k)
= |H(j)|A cos (t + + H(j))
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
30
Sinusoidal Steady-State Output
x(t)
x(t)
yss (t)
H(s)
y(t)
= A cos(t + )
= |H(j)|A cos (t + + H(j))
The input is sinusoidal
The steady-state signal yss (t) is also a sinusoid
Same frequency as x(t):
Amplitude is scaled by |H(j)|
The phase is shifted by H(j)
If we know H(s), we can easily nd the steady-state solution for
any sinusoidal input signal
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
31
Example 12: Steady-State Sinusoidal Analysis
R
+
vs(t)
vo(t)
-
Find the steady-state sinusoidal response to an input voltage of
vs (t) = cos(t).
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
32
Example 12: Workspace
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
33
Steady-State Sinusoidal Analysis Comments
We will study this in depth shortly
There is analytical signicance to how the magnitude and phase of
H(s) vary with s = j
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
34
LTI Systems
x(t)
h(t)
y(t)
H(s)
x(t)
y(t)
If we know the transfer function, we have sucient information to
calculate the output for any input
This enables us to treat the circuit more abstractly as H(s)
The transfer function may be for another type of system:
mechanical, chemical, hydraulic, etc.
Mathematically they are treated the same
Field-specic analysis is used only to nd H(s)
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
35
; ;
Example 13: Transfer Function Analysis
b
x(t)
y(t)
Find the transfer function for the linear system shown above. The
external force x(t) is the input to the system and the displacement
y(t) is the output. Find the transfer function.
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
36
Transfer Function Synthesis
x(t)
h(t)
y(t)
H(s)
x(t)
y(t)
Thus far we have talked only about circuit analysis
We now know several ways to solve for the output of a given
system
If there are zero initial conditions, then we can nd the transfer
function H(s) of a given circuit
Now we will discuss how to design a circuit that implements a
given H(s)
This is called transfer function synthesis
There are many circuits that have the same transfer function
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
37
Cascade Transfer Function Synthesis
x(t)
H1(s)
H2(s)
HP(s)
y(t)
N (s)
= H1 (s) H2 (s) HP (s)
H(s) =
D(s)
There are many approaches to transfer function synthesis
Will discuss how to specify H(s) to meet the requirements for a
given application later this term
The most common (and perhaps easiest) approach to synthesis is
to break H(s) up into 1st (real poles) or 2nd (complex poles)
order components
Thus each component, Hi (s) has either a 1st or 2nd order
polynomial in the numerator and denominator
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
38
Cascade Transfer Function Synthesis Continued
x(t)
H1(s)
H2(s)
HP(s)
y(t)
There are robust, standard circuits for implementing these
low-order components
The output of each transfer function is generated by an
operational amplier
This is essential for the cascade synthesis to work (will explain
later)
Some of these 1st and 2nd order components are discussed in the
text (Chapter 15)
Others can be found in more advanced analog circuits texts
You will use cascade synthesis in the rst lab for ECE 203
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
39
Summary
Circuits with a single input (independent source) and zero initial
conditions can be represented generically by their transfer
functions
H(s) is the Laplace transform of the system impulse response
The output of the system is y(t) = L1 {H(s)X(s)} for any
causal input signal (x(t) = 0 for t < 0)
For sinusoidal inputs, the output is also sinusoidal at the same
frequency but amplied by |H(j)| and shifted in phase by
H(j)
Thus, transfer functions make sinusoidal steady-state analysis easy
Generalization of phasors
Transfer function analysis used for all types of LTI systems, not
just circuits
Can synthesize a transfer function using a cascade of 1st and 2nd
order active circuits
J. McNames
Portland State University
ECE 222
Transfer Functions
Ver. 1.67
40
