ECE514F 2018

Transfer Functions
 Zero-State Response
• Linear constant coefficient differential equation
Input x(t) and output y (t )  y zeroinput (t )  y zero state(t )
Zero-state response: all initial conditions are zero
yt   Y s  xt   X s 
dr dk
r
y t   s r
Y s  k
x t   s k
X s 
dt dt
Laplace transform both sides of differential equation with
all initial conditions being zero and solve for Y(s)/X(s)
y' (t )  y(t )  x(t ) sY ( s)  Y ( s)  X ( s)
y (0  )  0 H ( s) 
Y ( s)

1
X (s) s  1 19 - 2
 Transfer Function
• H(s) is called the transfer function because it
describes how input is transferred to the output
in a transform domain (s-domain in this case)
Y(s) = H(s) X(s)
y(t) = L-1{H(s) X(s)} = h(t) * x(t)  H(s) = L{h(t)}
• Transfer function is Laplace transform of
impulse response

19 - 3
 Transfer Function Examples
• Laplace transform

X s     xt  es t dt
0

• Assume input x(t) and output y(t) are causal

yt   xt  T  x(t)

Y s   X s  e  s T
Y s 
 
H s   es T
X s 
19 - 4
 Transfer Function Examples
• Ideal integrator with • Ideal differentiator
y(0-) = 0 with x(0-) = 0
y t     x d
t
y t   x(t )
d
0
dt
Y s   X s   y 0 
1
s
1
s
   
Y s   s X s   x 0   s X ( s)
Y s 
H s   s
H s  
Y ( s) 1
 X s 
X (s) s

x(t)
  dt
y(t) x(t) d
  y(t)
t

0 dt 19 - 5
 Cascaded Systems
• Assume input x(t) and output y(t) are causal
 x d
t
• Integrator first, x(t) 0
x(t)
then differentiator X(s) 1/s s
X(s)/s X(s)
• Differentiator first, x(t) d
x(t ) x(t)
then integrator dt
s 1/s
X(s) s X(s) X(s)
• Common transfer functions
A constant (finite impulse response)
A polynomial (finite impulse response)
Ratio of two polynomials (infinite impulse response)
19 - 6
 Block Diagrams
X(s) H(s) Y(s)

W(s)
X(s) H1(s) H2(s) Y(s) = X(s) H1(s)H2(s) Y(s)

H1(s)
X(s)  Y(s) = X(s) H1(s) + H2(s) Y(s)
H2(s)

E(s)
X(s)  G(s) Y(s) = X(s)
G(s)
1 + G(s)H(s)
Y(s)
-
H(s)
19 - 7
 Cascade and Parallel Connections
• Cascade
W(s) = H1(s) X(s) Y(s) = H2(s)W(s)
Y(s) = H1(s) H2(s) X(s)  Y(s)/X(s) = H1(s)H2(s)
X(s) H1(s) H2(s) Y(s)  X(s) H2(s) H1(s) Y(s)

One can switch the order of the cascade of two LTI systems
if both LTI systems compute to exact precision
• Parallel Combination
Y(s) = H1(s)X(s) + H2(s)X(s) Y(s)/X(s) = H1(s) + H2(s)
H1(s)
X(s)  Y(s) = X(s) H1(s) + H2(s) Y(s)
H2(s) 19 - 8
 Feedback Connection
• Governing equations • What happens if H(s) is
E s   F s   H s  Y s  a constant K?
Y s   Gs  E s  Choice of K controls all
poles in transfer function
• Combining equations
Common LTI system in
Y s   Gs   F s   H s  Y s  
EE362K Introduction to
Y s   G s  H s  Y s   G s  F s  Automatic Control and
Gs  EE445L Microcontroller
Y s   F s 
1  Gs  H s  Applications/Organization
E(s)
F(s)  G(s) Y(s) = F(s)
G(s)
1 + G(s)H(s)
Y(s)
-
19 - 9
H(s)
 External Stability Conditions
• Bounded-input bounded-output stability
Zero-state response given by h(t) * x(t)
Two choices: BIBO stable or BIBO unstable
• Remove common factors in transfer function H(s)
• If all poles of H(s) in left-hand plane,
All terms in h(t) are decaying exponentials
h(t) is absolutely integrable and system is BIBO stable
• Example: BIBO stable but asymptotically
unstable H ( s) 
 s  1   1  s  1   1 
 
      
 s 1 
2 s  1  s  1   s  1 
19 - 10
Based on slide by Prof. Adnan Kavak
 Internal Stability Conditions
• Stability based on zero-input solution
• Asymptotically stable if and only if
Characteristic roots are in left-hand plane (LHP)
Roots may be repeated or non-repeated
• Unstable if and only if
(i) at least characteristic root in right-hand plane and/or
(ii) repeated characteristic roots are on imaginary axis
• Marginally stable if and only if
There are no characteristic roots in right-hand plane and
Some non-repeated roots are on imaginary axis
19 - 11
Based on slide by Prof. Adnan Kavak
Frequency-Domain Interpretation
y t   ht   e s t
est h(t) y(t)

  h e s t  d
• y(t) = H(s) e s t 


for a particular value of s e st

H s 

y t   ht   e j 2p ft

• Recall definition of 
frequency response:   h e j 2p f t  
d


ej 2 p f t h(t) y(t) e j 2p f t
h  
d
e j 2p f 

Hf 

19 - 12
Frequency-Domain Interpretation
• Generalized frequency: s = s + j 2 p f
• We may convert transfer function into
frequency response by if and only if region of
convergence of H(s) includes the imaginary axis
H freq  f   H s  s  j 2pf

H s   for Res  0
1
• What about h(t) = u(t)?
s
We cannot convert H(s) to a frequency response
However, this system has a frequency response
• What about h(t) = d(t)? H s   1 for all s  H freq  f   1
19 - 13
 Frequency Selectivity in Filters
• Lowpass filter • Bandpass filter
|Hfreq(f)| |Hfreq(f)|
1 1

f f
• Highpass filter • Bandstop filter
|Hfreq(f)| |Hfreq(f)|

f f
19 - 14
Linear time-invariant filters are BIBO stable