The Inverse Z-Transform: 1 Dr. B. Khan CECOS University

You might also like

Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 48

The inverse z-transform

• Formal inverse z-transform is based on Cauchy


integral.
• Less formal ways, sufficient most of the time are:
• Inspection method which makes use of the z-
transform table.
• Power series method
• Partial fraction method where a rational function is
expanded into simple functions the inverse z-
transforms of which can be obtained from the
table.
1 Dr. B. Khan CECOS University
The inverse z-transform
• A general expression for the partial fraction
expansion of a rational function has already
been mentioned.
• Now we are ready to take the final step in the
inversion of X(z).
• The process of inversion and the necessary
formulae can easily be understood if we start
from the simplest case- when all poles are
distinct and real.

2 Dr. B. Khan CECOS University


The inverse z-transform
• When all the poles are distinct and real, then ni
=1, and r = N (total number of poles), cik = ci1 =
Ai.
X (z) N
Ai
z
  i 1 (z  pi )

 X ( z) 
Ai  ( z  pi ) 
 z  z  pi

3 Dr. B. Khan CECOS University


The inverse z-transform
• Or
N
Ai
X (z)  
i 1
1
(1  p i z )

N
 Ai 
x(n)  z 1
 1 
i 1  (1  p i z ) 

4 Dr. B. Khan CECOS University


The inverse z-transform
• Or

1  A  
 pi  n
u (n), if ROC: z  pi
z  i
1  

(1  p z )    pi  u(n 1) if ROC: z  pi
n
 i

x ( n )  ( A1 p  A2 p  .....  AN p )u ( n )
n
1
n
2
n
N

5 Dr. B. Khan CECOS University


The inverse z-transform
• When some of the poles are complex, then if
x(n) is real, the polynomial appearing in X(z) will
have real coefficients. This means that if pj is a
pole, then its complex conjugate pj* will also be a
pole.
• The corresponding coefficients (residues) will
also be complex.
1  A A *

x k (n )  z  k
1
 k
1 
1  p k z 1  pk z 
*

6 Dr. B. Khan CECOS University


The inverse z-transform
• Or

xk (n)  [ Ak  pk   A p
n *
k   * n
k ]u(n)
j k
Ak  Aand
k e
j k
p k  rk e

x k ( n )  Ak rk e j (  k n  k )
e  j (  k n  k )
u ( n )
x k ( n )  2 Ak r k
n
cos(  k n   k ) u ( n )
7 Dr. B. Khan CECOS University
The inverse z-transform
• When some of the poles are repeated:
ni
X (z) N
Ak Cm
z
 
k 1 , k  i

z  p k m 1  z  p i  m

• The first term is for all first order poles. The second term
is for a pole of order ni . Such terms will be there for
every higher order pole.
• The residues Cm’s will be evaluated through
differentiation as mentioned before.
• Each term can be inversely transformed by inspection
method.
8 Dr. B. Khan CECOS University
Analysis and characterization
of LTI systems using Z-Transform
• As we mentioned, that a relaxed LTI system
can be described by a linear difference
equation with constant coefficients of the
form:

N M
y(n)  ak y(n  k)  bk x(n  k)
k 1 k 0

9 Dr. B. Khan CECOS University


Analysis and characterization
of LTI systems using z-transform
• Take the z-transform of the difference
equation M

k  0
b k z  k

H ( z )  N
1  k  1
a k z  k

which indicates that H(z) is rational that is


A( z )
H ( z) 
B( z )

10 Dr. B. Khan CECOS University


Analysis and characterization
of LTI systems using z-transform
N (z)
• If X(z) is also rational, that is X (z) 
Q (z)
then,

N ( z ) A( z )
Y (z)  H (z) X (z) 
B ( z )Q ( z )
• Assume that the system and the input have
simple poles
11 Dr. B. Khan CECOS University
Analysis and characterization
of LTI systems using z-transform
• If pk ‘s and qk ‘s are the poles of the system
function and the input signal, then the partial
expansion is:

N
Ak L
Qk
Y (z)   1
  1
k 1 1  p k z k 1 1  q k z

12 Dr. B. Khan CECOS University


Analysis and characterization
of LTI systems using z-transform
• The inverse z-transform of Y(z) is:
N L
y(n)   Ak ( pk ) u(n)  Qk (qk ) u(n)
n n

k 1 k 1

• The first part is the natural response, while the


second part is called the forced response

Dr. B. Khan CECOS University


13
Transient and steady state responses
• The output of a relaxed system with real distinct poles
is N L
y(n)   Ak ( pk ) u(n)  Qk (qk ) u(n)
n n

k 1 k 1
• The system is initially relaxed at time n =0, means that
its memory (i.e., the outputs of the delays) should be
zero. Since the memory of the system is described in
“state”, the system is zero-state and the corresponding
output is called the zero-state response.

Dr. B. Khan CECOS University


14
Transient and steady state responses
• The zero-state response of a relaxed system can
therefore be separated into two parts. The
natural response is N
ynr (n)   Ak ( pk ) u(n)n

k 1

• The forced response is


L
y fr (n)  Qk (qk ) u(n)n

k 1

Dr. B. Khan CECOS University


15
Transient and steady state responses
• The zero-state response of a relaxed system can
therefore be separated into two parts. The
natural response is N
ynr (n)   Ak ( pk ) u(n)n

k 1

• The forced response is


L
y fr (n)  Qk (qk ) u(n)n

k 1

Dr. B. Khan CECOS University


16
Transient and steady state responses
• If |pk| <1 for all k, then ynr(n) decays to zero as n
approaches to infinity. The response is called the
transient response. The transient may rapid or
slow.
• The decay is rapid if the magnitude of the pole
position is small (near the origin) and is slow if
the magnitude of the pole is large (near the unit
circle).

Dr. B. Khan CECOS University


17
Transient and steady state responses
• If all the poles of X(z) are inside the unit circle
then yfr(n) decays to zero as n approaches to
infinity indication that the input is transient.
• However, if all the poles lie on the unit circle, the
response is sinusoidal indicating that the input
signal persists for all n ≥ 0.
• The response is called steady-state response.

Dr. B. Khan CECOS University


18
Stability, causality and ROC
• An LTI system is causal if h(n) = 0 for n <0.
• In terms of ROC a discrete-time LTI system with
rational system function is causal if and only if:

(a) The ROC is the exterior of a circle outside the


outermost pole, including z= ∞.
(b) with H(z) expressed as a ratio of polynomial
in z, the order of the numerator cannot be
greater than the order of the denominator.
19 Dr. B. Khan CECOS University
Stability, causality and ROC continued
• An LTI system is BIBO stable implies that:


n  
h(n)  

• In terms of the system function H(z): A


necessary and sufficient condition for a system
to BIBO stable is:

H (z)  
n  
h(n)
20 Dr. B. Khan CECOS University
Stability, causality and ROC continued

• We have:
H (z)   h(n) z
n  
n


H (z)  
n  
h(n) z n

• At |z|=1 
H (z)  
n  
h (n )  

21 Dr. B. Khan CECOS University


Stability, causality and ROC continued
• Hence if the system is BIBO stable, the unit circle is
contained in the ROC. The converse is also true.
• A causal LTI system with rational function is stable
if and only if all the poles are inside the unit circle.
• The conditions on stability and causality are
different and that one does not imply the other. A
causal system may be stable or unstable. The same
is for anti-causal and non-causal.

22 Dr. B. Khan CECOS University


Pole-zero cancellation
• When a pole of rational z-transform is at the same location
as a zero, the pole is cancelled by the zero and,
consequently, the term containing that pole in the inverse
z-transform vanishes.
• No exact pole-zero cancellation can occur in practice as a
result of numerical calculations, and as a result the term in
the response has a very small amplitude.
• A system pole can suppressed by the zero of the input
signal.
• By proper selection of the zeros of the input, it is possible
to suppress one or more system modes (pole factors).
23 Dr. B. Khan CECOS University
Multiple-order poles and stability
• A multiple-ordered poles results in an output
sequence that terms of the form:
b n
Ak n ( pk ) u(n)
where 0 ≤b≤ m-1 and m is the order of the pole. If |pk|
<1, these terms decay to zero as n approaches infinity
because the factor (pk)n dominates the term nb.
Consequent no bounded input can produce unbounded
output if the system poles are inside the unit circle.
24 Dr. B. Khan CECOS University
Stability of second-ordered system
• The second-ordered system forms the basic
building block for the realization of higher
ordered system. Therefore it is better to study
such system in detail. The system equation is

y(n)  a1 y(n  1)  a2 y(n  2)  b0 x(n)


1 2
Y ( z)  a1Y ( z) z  a2Y ( z) z  b0 X ( z)

25 Dr. B. Khan CECOS University


Stability of second-ordered system
• The system function is therefore
Y ( z) b0
H ( z)   1 2
X ( z ) 1  a1 z  a 2 z

2
b0 z
H (z)  2
z  a1 z  a 2

26 Dr. B. Khan CECOS University


Stability of second-ordered system
• The system poles are:

a1 a  4a 2
2
p1 , p 2    1

2 4
• The characteristic of the tow-pole system depend
on the location of the poles

27 Dr. B. Khan CECOS University


Stability of second-ordered system
• When the poles are real and distinct implies:
a12 > 4a2, then
A1 A2
H (z)  1
 1
1  p1 z 1  p2 z
b 0 p1  b0 p 2
A1  , A2 
p1  p 2 p1  p 2

28 Dr. B. Khan CECOS University


Stability of second-ordered system
• The unit sample response of the system is:

h(n) 
b0
p1  p 2
n 1

n 1
p1  p 2 u ( n ) 
• If the poles are equal :

a  4 a 2  p1  p 2   a1 / 2
2
1

29 Dr. B. Khan CECOS University


Stability of second-ordered system
• The system function is:
b0
H (z) 
1  pz  1 2

• The unit sample response h(n):

h ( n )  b 0 ( n  1) p u ( n ) n

30 Dr. B. Khan CECOS University


Stability of second-ordered system
• If the poles are complex conjugate: a12 < 4a2
*
A A
H (z)  

1  pz 1 * 1
1 p z   
b0
H (z) 

1  pz 1
1 p z
*
 1

31 Dr. B. Khan CECOS University
Stability of second-ordered system
• Or *
A A
H (z)  

1  pz 1 * 1
1 p z   

b0  A 1  p z * 1
 A 1  pz
*
 1

*
b0 p b0 p
A  and A *

p  p* p*  p
32 Dr. B. Khan CECOS University
Stability of second-ordered system
• The unit sample response of the system is:
h(n)  A p  u (n)  A
n *
p 
* n
u (n)

• Let p =rejω where 0 < ω <π

j  j
b0 e b0 e
A  and A *
 
j 2 sin  j 2 sin 
33 Dr. B. Khan CECOS University
Stability of second-ordered system
• The unit sample response of the system is:
j ( n 1)  j ( n 1)
b0 r e n
e
h(n) 
sin  2j
n
b0 r
h(n)  sin  ( n  1) u ( n )
sin 
34 Dr. B. Khan CECOS University
Response of Pole-zero system
with non-zero initial conditions
• The two-sided Z-transform requires that the
corresponding signals be specified for the entire
time range (-∞ <n <∞).
• This requirement prevents its use for a very
useful family of practical problems, namely the
evaluation of the output of non-relaxed systems.
• The reason is that these systems are described
by difference equations with non-zero initial
conditions.

35 Dr. B. Khan CECOS University


The one-sided z-transform
• The one-sided z-transform is a power tool for
the solutions of constant coefficients
difference equations with non-zero initial
conditions
• The one-sided or the unilateral z-transform is

 x(n) z
defined as:  n
X (z) 
n0

x(n)  X (z),

36 Dr. B. Khan CECOS University


One-sided z-transform
• It has the same properties as X(z) except the
time shifting which is different.
• Time delay property:

If x(n)  X ( z), then
  n 
k
x(n  k )  z  X ( z)   x(n) z 
k

 n1 

37 Dr. B. Khan CECOS University


One-sided z-transform
• Time advance property:

If x(n)  X ( z), then
  n 
k 1
x(n  k )  z  X ( z)   x(n) z 
k

 n 0 
• The one-sided z-transform does not contain any
information about the negative values of x(n),
whether or not the signal x(n) is zero for n<0.
38 Dr. B. Khan CECOS University
Response of Pole-zero system
with non-zero initial conditions
• Let a causal signal x(n is applied to a non-
relaxed LTI system. The modeling equation of
an LTI (relaxed or not relaxed) system is
N M
y (n )    a k y (n  k )   bk x (n  k )
k 1 k 0

• For non-relaxed system we have initial


conditions which cannot be incorporated in
two-sided z-transform.
39 Dr. B. Khan CECOS University
Response of Pole-zero system
with non-zero initial conditions
• The expression for the one-sided z-transform
shows that the initial condition can be
incorporated as previous inputs and outputs.
N
 k

Y ( z )    a k z Y ( z )   y (  n ) 
 k 

k 1  n 1 
M
 k
n 
  bk z  X ( z )   x ( n ) z 
k 

k 0  n 1 

40 Dr. B. Khan CECOS University


Response of Pole-zero system
with non-zero initial conditions
• Since x(n) = 0, for n <0, therefore , X+(z) =X(z)
N N k
Y 
(z)  Y 
( z ) a k z k
  ak z k
 y ( n)
k 1 k 1 n 1
M
 X ( z ) bk z k

k 0

N N k
Y 
( z )( 1  
k 1
ak z k
)   ak z
k 1
k
 n 1
y ( n )
M
 X ( z ) bk z k

k 0

41 Dr. B. Khan CECOS University


Response of Pole-zero system
with non-zero initial conditions
• Since x(n) = 0, for n <0, therefore , X+(z) =X(z)
M N k


b z k
k
 a z  y(n) z
k
k n

Y ( z)  k 0
N
X ( z)  k 1
N
n 1

1   ak z k
1   ak z k

k 1 k 1

 N 0 ( z)
Y ( z)  H ( z) X ( z) 
A( z )
42 Dr. B. Khan CECOS University
Response of Pole-zero system
with non-zero initial conditions
where N k
N 0 ( z )   a k z k
 y (  n) z n

k 1 n 1

Y ( z)  H ( z) X ( z)
zs

 N 0 (z)
Y (z) 
zi
A( z )

y (n))  y zs (n)  y zi (n)


43 Dr. B. Khan CECOS University
Response of pole-zero system
with non-zero initial conditions
where N k
N 0 ( z )   a k z k
 y (  n) z n

k 1 n 1

Y ( z)  H ( z) X ( z)
zs

 N 0 (z)
Y (z) 
zi
A( z )

y (n))  y zs (n)  y zi (n)


44 Dr. B. Khan CECOS University
Response of pole-zero system
with non-zero initial conditions
• If p1, p2, p3,…,pN are the roots of A(z), then
N
y zi (n)   Dk  p k  u (n)
n

k 1
N L
y (n)  A
k 1
k ( p k ) u (n)   Qk (q k ) u (n)
n

k 1
n

N
   Dk ( p k ) n u (n)
k 1

45 Dr. B. Khan CECOS University


Response of pole-zero system
with non-zero initial conditions
Or
N L
y (n)  
k 1
A k/ ( p k ) n u ( n )  k 1
Q k (q k ) n u (n)

where A k
/
 Ak  D k

• The initial conditions change the natural response


ynr(n) of the system through scale factors {Ak} while the
force response yfr(n) remains the same.
• No new poles are introduced by the initial conditions .

46 Dr. B. Khan CECOS University


Practice problems
• The students are advised to use the tools and
formulae presented in the slides related to the z-
transform in order to develop their problem
solving skill only in the analysis of LTI systems in
the z-domain for the subjective section.

• For the objective section they are advised to


read the textbook in the context of the slides
presented on z-transform.

47 Dr. B. Khan CECOS University


Practice problems
• Exercise for the preparation of the subjective
section:
• Examples: 3.5.1, 3.5.2, 3.5.3, 3.5.4, 3.5.5, 3.6.7
• Related problems from the exercise: 3.35, 3.37,,
3.38, 3.42, 3.49 and Home Assignment problesm

48 Dr. B. Khan CECOS University

You might also like