Transfer Function
Transfer Function
Spring 2015
Lecture 2: Linear Circuit Analysis Review
Sam Palermo
Analog & Mixed-Signal Center
Texas A&M University
Announcements
Reading
Fundamentals of Circuit Analysis (Dr. Silva)
1.1, 1.2, App. D, E, F (Sedra/Smith)
Agenda
Laplace Transform
Passive Circuit s-Domain Models
Transfer Functions
Sinusoidal Steady-State Response
Poles & Zeros
Bode Plots
Second-Order Systems
References
Continuous & Discrete Signal & System Analysis,
3rd Ed., C. McGillem and G. Cooper, Saunders
College Publishing, 1991.
Feedback Control of Dynamic Systems, 3rd Ed., G.
Franklin, J. Powell, and A. Emami-Naeini,
Addison-Wesley, 1994.
Design of Analog Filters, R. Schaumann and M.
Van Valkenburg, Oxford University Press, 2001.
4
Motivation Example
Given vo 0 0
Write a KCL at vo
vo t sin 105 t
dv t
10nF o 0
1k
dt
dvo t
vo t
sin 105 t
1k 10nF 1k 10nF
dt
Now, if we remember anything from out Diff. Eq. class, we can solve this equation.
vo t
1 105 t
e
2
transient response
(can go to zero quickly)
sinusoidal steady-state
response
Laplace Transform
Laplace transforms are useful for solving
differential equations
One-Sided Laplace Transform
Lxt X s xt e st dt
[McGillem]
7
[McGillem]
8
Complex Frequency
Representation:
1
i t vt
R
V s RI s
I s
1
V s
R
1t
vt i d v0
C0
i t C
Complex Frequency
Representation:
dvt
dt
1
1
V s
I s v0
sC
s
I s CsV s Cv0
10
Time-domain Representation:
Complex Frequency
Representation:
di t
dt
1t
i t v d i 0
L0
V s LsI s Li 0
1
1
I s V s i0
sL
s
11
V s I s R
V s I s
Z s R
Z s
1
sC
1
sC
V s I s sL
Z s sL
12
Transfer Function
H s
L vo t Vo s
L vi t Vi s
RC Transfer Function
Vo s
1
sC
ZC
1
Vin s
Vin s
Vin s
1
1 sRC
Z R ZC
R
sC
Vin s 1 sRC
14
Convert to
Laplace Domain
Vo s
10 5
1
1
H s
Vin s 1 sRC 1 s
s 10 5
10 5
5
10 5
10
Vo s H s Vi s
2
s 10 5 2
s 10 5
with partial fraction expansion
1
1
1 5
s
10
2
2
2
Vo s
s 10 5 s 2 10 5 2 s 2 10 5 2
v o t
1 105 t
e
2
15
If input vi t is sinusoidal
vi t A cost
The steady - state output will be
v ss t H j A cost H j
17
H j
H s
Vin s 1 sRC
1 jRC
Output Magnitude
1
1
H j H j H j
1 jRC 1 jRC
*
H j
1 RC 2
Output Phase
ImH j
ImNum
ImDen
tan 1
tan 1
H j tan 1
Re
H
j
Re
Num
Re
Den
H j tan 1 RC
0
RC
1
H j tan 1 tan 1
tan RC
1
1
18
1
1
s
10 5
with s j j10 5
H j10 5
H j10
1
1 j
1
1
2
2
H j10 5 tan 1 1 45
v ss t
1
2
sin 10 5 t 45
19
Numerical Example
1 j1010 j10
100 j101000 j10
1 j1010 j10
100 j101000 j10
12 10 2 10 2 10 2
100 2 10 2 1000 2 10 2
1.41 10 3
20
21
s z1 s z2 ...s zm
s p1 s p2 ...s pn
Poles are the roots of the denominator (p1, p2, pn) where H(s)
Zeros are the roots of the numerator (z1, z2, zm) where H(s)0
105
Example 1 : H s
s 105
Example 3 : H s
s 105 0
s 15 0
p1 s 105 rad / s
s
Example 2 : H s
s 105
z1 s 0rad / s
s 105 0
100s 15
s 2 50 s 1500
z1 s 15rad / s
s 2 50 s 1500 0
p1, 2 s1, 2
50 2500 6000
25 j 29.6rad / s
2
p1 s 105 rad / s
22
Bode Plots
Technique to plot the Magnitude (squared) and
Phase response of a transfer function
Magnitude is plotted in Decibels (dB), which is a power
ratio unit
dB
H j 10 log10 H j
2
dB 20 log
10
H j dB
H j tan
ReH j
1
23
H s
H s
Vo s
1
1
1
1 j10 5
1
, where p1 10 5 rad/s
j
1
p1
Phase:
1 10 5
20 log10 1 20 log10 1 10 5
PhaseH j tan 1 10 5
24
Phase:
~20log10 (1)
= 0dB
~-20log10 (10-5)
= -20dB/dec
1 10 5
20 log10 1 20 log10 1 10 5
PhaseH j tan 1 10 5
(rad/s)
|H(j)|
|H(j)|2
20log10|H(j)| (dB)
Phase (H(j)) ()
103
0.9999
0.9999
~0
~0
104
0.995
0.990
-0.043
-5.71
5x104
0.894
0.800
-0.969
-26.6
105
0.707
0.500
-3.01
-45.0
5x105
0.196
0.039
-14.2
-78.7
106
0.100
0.010
-20.0
-84.3
107
10-2
10-4
-40.0
-89.4
108
10-3
10-6
-60.0
-89.9
-45/dec
25
-20dB/dec
-45/dec
26
Transient Response
= 103 rad/s = -p1/100
vo t 1
Phase Shift 0
vo t
1
2
vo t 0.1
Phase Shift 84.3
27
2. Where to end?
a.
Calculate |H(j)| as
H s
s 10s 100 1 s 1 s
10 100
DC Magnitude 10 20dB
HF Magnitude 0 dB
20 log10 H j 20 log10
10 1 2
1 10
1 2
1 10
2 2
2
2
20 log10 10 20 log10 1 2 20 log10 1 101 20 log10 1 10 2
z1 1, p1 10, p2 100
-20dB/dec.
+20dB/dec.
-20dB/dec.
+20dB/dec.
-20dB/dec.
29
b.
c.
For negative poles: 1 dec. before the pole freq., the phase will
decrease with a slope of -45/dec. until 1 dec. after the pole
freq., for a total phase shift of -90
For negative zeros: 1 dec. before the zero freq., the phase will
increase with a slope of +45/dec. until 1 dec. after the zero
freq., for a total phase shift of +90
Note, if you have positive poles or zeros, the phase change
polarity is inverted
30
H s
s 10s 100 1 s 1 s
10 100
H j 180 tan 1 tan 1 tan 1
1
10
100
LF Phase 180
z1 1, p1 10, p2 100
+45/dec.
-90/dec.
+45/dec.
-45/dec.
-45/dec.
-45/dec.
31
Second-Order Systems:
Real or Complex Poles?
H s
k102
s2 s
0
Q
02
0
02
2 poles p1 , p 2
2Q
2Q
2 poles : p1 1, p2 1000
Note, Q 0.032
-20dB/dec.
-45/dec.
-40dB/dec.
-45/dec.
s 11s 10 s 1s 10
2 poles : p1 1, p2 10
Note, Q 0.287
-45/dec.
-20dB/dec.
-40dB/dec.
-90/dec.
-45/dec.
34
k102
s2 s
02
Q
What is the low frequency magnitude?
H j 0 k1
What is the high frequency magnitude?
H j
k102
k102
02 j
02
Q
02
k1Q
Note, if Q 1 then the magnitude exceeds the low frequency value, i.e. frequency peaking occurs!
35
2
4
k1 0
d
0
2
d
2
2 2 0
0
Q
pk 0 1
1
2Q 2
0 for large Q
k1Q
1
1
4Q 2
[Schaumann]
36
k102
02
02
j
Q
02
k1Q
37
Next Time
OpAmp Circuits
38