Professional Documents
Culture Documents
1
Chapter Outline
2
Why do we need another transform?
Think about all the transforms you have seen so far.
Why need yet another?
-Convergence issues with the FT
-The DTFT of a sequence exits if and only if the sequence x[n] is absolutely
summable, that is, if
x[n]
n
-DTFT may not exits for many signals of practical or analytical signals, whose
frequency analysis can therefore not be obtained through DTFT
A generalization of the DTFT leads to the z-transform that may
exits for many signals for which DTFT does not.
Furthermore, the use of the z-transform allows simple algebraic
expressions to be used which greatly simplifies frequency domain
analysis.
Digital filters are designed, expressed, applied and represented in
terms of the z-transform.
3
4.1 Definition of Z-Transform
For a given sequence x[n], its z-transform X(z) is defined as:
X ( z ) zx[n] x[ n ] z n
4
Thus by restricting z to have unity magnitude; i.e., for |z| = 1
(r=1), the z-transform corresponds to the DTFT (finding
frequency response).
Since the z-transform is a function of a complex variable, it is
convenient to describe and interpret it using the complex z-plane.
5
The unit circle in the
complex z-plane.
n
x[n]r n
n
x[n] z
n
8
The ROC as a ring in the z-
plane. For specific cases, the
inner boundary can extend
inward to the origin, and the
ROC becomes a disc. For other
cases, the outer boundary can
extend outward to infinity.
9
4.3 Rational Z-Transforms
The Z-transforms of LTI systems can be expressed as a ratio
of two polynomials in z-1, hence they are rational transforms.
Starting with the constant coefficient linear difference
equation representation of an LTI system:
10
A rational z-transform can be alternately written in factored
form as
M M
b0 (1 k z )
1
p0 ( z k )
H ( z) k 1
N
z ( N M ) k 1
N
k = zeros of H(z)
a0 (1 k z 1 ) d 0 ( z k ) k = poles of H(z)
k 1 k 1
12
Demo
What does this look like?
13
Poles & Zeros In Matlab
14
15
16
4.4 Properties of the ROC
The properties of the ROC depend on the nature of the signal.
Assume that the algebraic expression for the z-transform is a
rational function and that x[n] has finite amplitude, except possibly
at n = and n = -.
The properties are summarized below:
17
(cont)
Property 5: If x[n] is a finite-duration sequence, then the
ROC is the entire z-plane, except possibly z = 0 or z = . The
point z = will be included if x[n] = 0 for n < 0, and the point
z = 0 will be included if x[n] = 0 for n > 0.
Property 6: If x[n] is a right-sided sequence, the ROC
extends outward from the outermost (largest magnitude) finite
pole in X(z) to (and possibly including) z = .
Property 7: If x[n] is a left-sided sequence, the ROC extends
inward from the innermost (smallest magnitude) nonzero pole
in X(z) to (and possibly including) z = 0.
Property 8: If x[n] is a two-sided sequence, the ROC will be
the intersection of the two ROC areas corresponding to the left
and right sides of the sequence.
18
19
Example 1: Right-Sided Exponential Sequence
|
| az
n
1 n
Thus, the ROC is the range of values of z for which |az-1| < 1 or,
equivalently, |z| > |a|. Inside the ROC, the infinite series converges
to:
X z (az 1 ) n
1 z
1
, | z || a |
n 0 1 az za
Recall this
1
k
k 0 1
20
Polezero plot
and ROC for
Example 1.
For |a| >1, the ROC does not include the unit circle, and thus the
DTFT of x[n] does not converge.
21
Example 2: Left-Sided Exponential Sequence
n 1 n 0
If |a -1z| < 1 or, equivalently, |z| < |a|, the infinite series converges
and
X z 1
1
1 a 1 z
1 z
1
, | z || a |
1 az za
Same as Example1 !
22
A left-handed series
has as ROC the
Polezero plot interior of a circle
and ROC for
Example 2.
For |a| <1, the ROC does not include the unit circle, and thus the
DTFT of x[n] does not exist.
Note that the z-transforms of the two sequences anu[n] and
-anu[-n-1] are identical even though the sequences are different!
26
Summary
27
4.5 Properties of Z-Transform
Linearity. If the sequences x1[n] and x2[n] have z-transforms X1[z] and
X2[z], the z-transform of their linear combination is
ax1[n] bx2 [n] aX1[ z] bX 2 [ z]
Delays or shifts. If the z-transform of a sequences x[n], is X[z] then the
z-transforms of the sequence delayed by m samples is z-mX[z].
x[n] X [ z ]
x[n m] z m X [ z ]
31
z-Transform: Table of Properties
32
z-Transform Pairs
33
34
4.6 Inverse z-Transform
One of the important roles of the z-transform is in the analysis of
discrete-time linear systems.
Often, this analysis involves finding the z-transform of
sequences and, after some manipulation of the algebraic
expressions, finding the inverse z-transform.
The IZT is particularly useful in DSP, for example in finding the
impulse response of digital filters.
Possible methods for determining the inverse z-transform are:
Contour Integration (formal method, difficult to use)
Inspection Method (informal methods,
Partial Fraction Expansion sufficient and
preferable)
Power Series
35
Inspection Method
The inspection method relies on using known transform pairs.
Example 1 Example 2
Find the inverse for Find the inverse for
1 1 1 1
X ( z) , | z|| | X ( z) , | z|| |
1 2 1 2
1 z 1 1 z 1
2 2
From table: From table:
1 1
a n u[ n] 1
, |z||a | a n u[ n 1] , | z||a |
1 az 1 az 1
Answer: Answer:
n
1
n
1
x[ n] u[ n] x[ n] u[ n 1]
2 2
37
If the poles of H(z) are of first order and N=M, then H(z) can be
expanded as
C1 C2 CN
H ( z ) B0 1
1
...
1 p1 z 1 p2 z 1 pN z 1
N
C1 z C2 z CN z Ck z
B0 ... B0
z p1 z p2 z pN k 1 z pk
where pk are the poles of H(z) (assumed distinct), Ck are the partial
fraction coefficients and B0 bM aM . The Ck are also known as the
residues of H(z).
If M<N, then B0 will be zero. If M>N then H(z) must be reduced
first, to make MN, by long division with the numerator and
denominator polynomials written in descending powers of z-1.
38
The coefficient, Ck, associated with the pole pk may be obtained
by multiplying both sides of equation earlier by ( z pk ) z and then
letting z = pk.
z pk
Ck H ( z ) k 0,1,2,3...N
z z pk
If H(z) contains one or more multiple-order poles (that is poles
that are coincident) then extra terms are required in equation to
take this into account. For example, if H(z) contains an m-th order
pole at z=pk the partial fraction expansion must include terms of the
form m
Di
(z p
i ) i
k
39
Example 1
X(z) contains simple, first-order poles. Find the inverse z-transform of
the following: z 1
X ( z)
1 0.25z 1 0.375z 2
Solution:
First, express the z-transform in positive power of z by multiplying the
numerator and denominator by z2 (the highest power of z):
z z
X ( z) 2
z 0.25z 0.375 ( z 0.75)( z 0.5)
X(z) contains first-order poles at z=0.75 and at z=-0.5 (that is, only one
pole occurs at each pole position). Since the order of the numerator is less
that the order of the denominator (N<M), the partial fraction expansion
has the form z C1 z C2 z
X ( z)
( z 0.75)( z 0.5) z 0.75 z 0.5
Divide both sides by z:
X ( z) z C1 C
2
z z ( z 0.75)( z 0.5) z 0.75 z 0.5
40
To obtain C1, we simply multiply both sides of the equation by z-0.75
and let z=0.75: ( z 0.75) X ( z ) ( z 0.75) C2 ( z 0.75)
C1
z ( z 0.75)( z 0.5) z 0.5
1 1 4
C1
z 0.5 z 0.75 0.75 0.5 5
C2 is obtained by: ( z 0.5) X ( z ) 1 4
C2
z z 0.5 0.75 0.5 5
4 5z 4(0.75)
4 5z 4 5z
n
Thus we have 1
Z
X ( z) z 0.75 5
z 0.75 z 0.5 4 5z 4(0.5)
1
n
Z
z 0.5 5
The desired inverse z-transform, x(n), is the sum of the two inverse z-
transforms:
4
x(n) [(0.75) (0.5) ], n 0
n n
5
41
Example 2
X(z) contains a second-order poles. Find the discrete-time sequence, x[n],
with the following z-transform: z2
X ( z)
( z 0.5)( z 1) 2
Solution:
X(z) has a first-order pole at z=0.5 and a second-order pole at z=1. The
partial fraction expansion has the form: C D1 D2
X ( z)
z 0.5 z 1 ( z 1) 2
To obtain C, multiply both sides of the equation by z-0.5 and let z=0.5:
( z 0.5) z 2 0.5
C 2
z ( z 0.5)( z 1) z 0.5 (0.5 1)
2 2
To obtain D1, use the equation (slide 39), with i=1 and m=2. Thus
d z 1 X ( z ) d z 1 z 2
2 2
D1 2
dz z z 1 dz z ( z 0.5)( z 1) z 1
d z z 0.5 z
2
dz z 0.5 z 1 ( z 0.5) z 1
2
42
To obtain D2, use the equation (slide 34), with i=2 and m=2:
D2
z 1 X ( z)
2
z 12 z 2
1
2
z z 1 z( z 0.5)( z 1) z 1 (1 0.5)
2
43
Example 3
Since the poles are both first order, X(z) can be expressed as:
A1 A2
X ( z)
1 1 1 1
1 z 1 z
4 2
44
cont
1 1
A1 [(1 z 1 ) X ( z )]z 1/ 4 1 A2 [(1 z 1 ) X ( z )]z 1/ 2 2
4 2
Therefore, 1 2
X ( z)
1 1 1 1
1 z 1 z
4 2
Since x[n] is right sided, the ROC for each term extends
outward from the outermost pole. From the z-transform table:
n n
1 1
x ( n) 2 u[ n] u[ n]
2 4
45
Example 4
1 z 1
Find the inverse for X ( z)
1 z 1 6 z 2
X(z) can be expressed as:
1 z 1 A1 A2
X ( z)
(1 2 z 1 )(1 3z 1 ) (1 2 z 1 ) (1 3z 1 )
Therefore 15 45
X ( z)
(1 2 z 1 ) (1 3 z 1 )
46
cont
z2-5z+4 = (z-1)(z-4)
z -5z+6|z4-10z3+35z2-50z+24
2
48
cont
Therefore
1/ 6 4 27 / 2 32 / 3
X ( z) 1
1
1
ROC : 2 | z | 3
1 z 1 2z 1 3z 1 4 z 1
From the table:
1
X ( z) 1
, | z || a | x[ n ] a n
u[ n]
1 az
1
X ( z) 1
, | z || a | x[ n ] a n
u[ n 1]
1 az
The sequence x[n] that satisfies the ROC is:
1 27 32
x[ n] ( )u[n] (4)2 n u[ n] ( )( 1)3n u[ n 1] ( )( 1)4 n u[ n 1]
6 2 3
n2 1 3n 3 2 2 n 5
x[ n] ( 2 )u[n] ( )u[ n 1]
6 2 3
50
Example 6
1 2 z 1 z 2
X ( z) , | z | 1
Find the inverse for 3 1 1 2
1 z z
2 2
1 5 z 1 A1 A2 1 5 z 1 9
X ( z) 2 2 A1 1
9
1 1 1 1 1 1 z 1 1 z z 1/ 2 1
(1 z )(1 z ) 1 z
2 2
1 5z
1
4
A2 8
1 1 1/ 2
From the table: 1 z
2 z 1
1
x[ n ] 2 [ n ] 9( ) n u[ n ] 8u[ n ]
2
51
Example 7
Find the inverse z-transform of H(z) given the ROC
i) -0.6<|z|<0.2 ii) |z|>0.2 z 2 2z 1
H ( z) 2
z 0.4 z 0.12
52
Power Series
53
Example 8
Find the inverse for X z z 2 (1 z 1 )(1 z 1 )(1 z 1 )
1
2
By inspection, 1, n 2
1 / 2, n 1
x[n] 1, n0
1 / 2, n 1
0, otherwise
Equivalently, 1 1
x[n] [n 2] [n 1] [n] [n 1]
2 2
54
In MATLAB
55
4.7 System Function
y[ n ] h[ n ] * x[ n ] Y (e j ) H (e j ) X (e j ) Y ( z ) H ( z ) X ( z )
h[ n ] : impulse response, H (e j ) : freq. response, H ( z ) : system function
Causality
Recall that for a system to be causal, its impulse response must
satisfy h[n]=0, n<0, that is for a causal system, the impulse
response is right sided.
A pole inside the unit circle contributes an exponentially decaying
term to the impulse response.
A pole outside the unit circle contributes an exponentially
increasing term to the impulse response.
The ROC of a causal system extends outside of the outermost
pole circle.
56
causal +
stable
e.g., 0.6 u[n]
n
causal +
unstable
e.g., 2 u[n]
n
The relationship between the location of a pole and the impulse response characteristics for
a causal system. (a) A pole inside the unit circle contributes an exponentially decaying term
to the impulse response. (b) A pole outside the unit circle contributes an exponentially
57
increasing term to the impulse response.
Stability
If a system is stable, then the impulse response h[n] is
absolutely summable and the DTFT of the impulse response
exists.
An LTI system is stable, if and only if the ROC of its transfer
function H(z) includes the unit circle!
A pole inside the unit circle contributes a right-sided
decaying term to the impulse response.
A pole outside the unit circle contributes a left-sided
decaying term to the impulse response.
Note that a stable impulse response cannot contain any
increasing exponential or sinusoidal terms, since then the
impulse response is not absolutely summable.
An FIR Filter is always stable, why?
58
causal +
stable
e.g., 0.6 u[n]
n
anticausal
+ stable
e.g., 2 u[n 1]
n
The relationship between the location of a pole and the impulse response characteristics for
a stable system. (a) A pole inside the unit circle contributes a right-sided decaying term to
the impulse response. (b) A pole outside the unit circle contributes a left-sided decaying term
to the
59
impulse response.
Causality and Stability
A causal systems ROC lies outside of a pole circle. If that
system is also stable, its ROC must include unit circle.
So, for a LTI system to be both stable and causal, the ROC
must be outside the outermost pole and include the unit
circle.
Clearly, a realizable LTI system that is both stable and
causal must have all their poles inside the unit circle.
A pole inside the unit circle a right-sided decaying term.
Similarly, an anticausal system is stable, if and only if its
poles lie outside the unit circle.
Poles on the unit circle complex sinusoidal terms
unstable.
60
cont
A system that is both stable and causal must have all its poles inside the
unit circle in the z-plane, as illustrated above.
61
Example 1
An LTI system has the transfer function:
X z
2 2 3
j
1
j 1 2 z 1
1 0.9e z 4
1 0.9e 4
z 1
Find the impulse response if the system is (a) stable or (b) causal.
Can this system be both stable and causal?
Solution: The poles are z = 0.9e j/4, z = 0.9e-j/4, and z = 2
62
If the system is stable, then the ROC includes the unit circle. The two poles
inside the unit circle contribute right-sided terms to the impulse response, while
the pole outside the unit circle contributes a left-sided term. Hence for case (a)
n n
j j
h[n] 2 0.9e u[n] 2 0.9e 4 u[n] 3 2 u[n 1]
4 n
40.9 cos n u[n] 3 2 u[n 1]
n n
If the system is causal, then all poles contribute right-sided terms to the impulse
response, so for case (b), we have:
n n
j j
h[n] 2 0.9e 4 u[n] 2 0.9e 4 u[n] 32 u[n]
n
Note that this LTI system cannot be both stable and causal, since there is a pole
outside the unit circle.
63
A Summary
ROC
For causal sequences of finite duration the z-transform converges
everywhere except at z=0. For causal infinite duration sequences the z-
transform converges everywhere outside a circle bounded by the radius of
the pole with the largest radius. For stable causal system the ROC always
encloses the unit circle which is important for the system to have a
frequency response.
Pole-zero plot
The pole-zero diagram provides an insight into the properties of a given
discrete-time system. For example, from the locations of the poles and
zeros we can infer the frequency response of the system as well as its
degree of stability. For a stable system, all the poles must lie inside the
unit circle (or be coincident with zeros on the unit circle).
64
Frequency response
Why frequency response is important? For example, in the design of
discrete filters, it is often necessary to examine the spectrum of the filter
to endure that the desired specifications are satisfied.
Do you know that the frequency response of a discrete-time system is
simply the Fourier Transform of its impulse response?
For example, if we set z e jT , that is evaluate the z-transform around
the unit circle, we obtain the Fourier transform of the system:
H ( z) h[n]z
n
n
z e jT
H (e jT
) h[ n
n
]e jnT
65