Digital Signal Processing

Chapter 3

The z-Transform

Wasif Latif

z-transform
▪ Introduction
▪ Definition
▪ Region of convergence
▪ z-transform properties
▪ Rational z-transform and its implication on ROC
▪ Examples

2
 Limitation of the DTFT

Closed Form Solution

▪ h[n] can only be easily computed for 0 < α < 1,

because the summation has a closed-form solution

4
 Z-Transform
▪ We would like to be able to use the same frequency-
domain methods to analyze the system defined by
LCCDEs regardless of our particular choice of
parameter α
▪ The solution to this problem is to modify the Fourier
transform by providing an additional degree of
freedom to handle cases of divergent sequences
▪ The result is the z-transform

Z-Transform

6
 Laplace Transform

z-transform

8
 The z-Transform

▪ We can always write zn = rnejωn (polar coordinates)

▪ The eigenfunction ejωn, used in the DTFT, is just a
particular case (r = 1) of the eigenfunction used in
the z-transform
▪ The factor rn helps the z-transform sum converge to
a broader class of signals
9

Few Terms
▪ z-Plane: Since z is just a complex number, every
value of z can be placed on the complex plane,
which we call the z-plane
▪ ROC: A contiguous locus of values of z for which H(z)
exists is called the region of convergence (ROC) of
H(z).
▪ Zeros: The values of z in the finite z-plane for which
H(z) = 0 are called the zeros of H(z).
▪ Poles: The values of z in the finite z-plane for which
H(z) → ∞ are called the poles of H(z).
▪ The poles and zeros are collectively termed the
singularities of H(z).
▪ These singularities of H(z) can be indicated graphically on
the z-plane as a pole-zero plot, denoting the poles with an
x and zeros with an o
10

10
 ROC – Region of Convergence
▪ The z-transform has a region of convergence
(ROC), which is the value of z for which the
infinite sum in the direct transform is finite

▪ Without the ROC, the z-transform is an

ambiguous representation of a signal

11

11

Laplace Transform vs. z-Transform

x(t) Analog y(t) x[n] Digital y[n]

LTI
e st
H ( s )e st zn LTI
H ( z) z n


H ( s) =  h(t )e dt
−
− st H ( z) =  h[n]z
n = −
−n

s = j
⎯⎯⎯ → H ( j), (−, ) z = e j
⎯⎯⎯→ H (e j ),  (− ,  )

s-plane z-plane

Stable ⇒ RoC includes j-axis Stable ⇒ RoC includes unit circle

Causal ⇒ RoC rightward Causal ⇒ RoC outward

12

12
 z-trasform and DTFT

Background-Definition

j
- Fourier transform X (e ) =  x[n]e
n = −
− jn
extracts the essence of x[n]

but is limited in the sense that it can handle stable systems only.
X (e j ) converges if  | x[n] | 
i.e., stable system → Fourier Transform converges

- So, we want to extend it such that it can be used as a tool to

analyze digital systems in general.

13

13

z-trasform and DTFT

14

14
 z-trasform and DTFT - II

15

15

For what will we use the z-transform?

1. Representing LTI systems
2. Determining the stability of LTI systems
3. Solving difference equations

16

16
 Exponential weighting

17

17

Examples: Finite Duration Signals

18

18
 Right Sided Sequence of Infinite Duration

The ROC is shown by green coloured area

19

19

Monotonic Right-Sided Sequences

The ROC is shown by blue coloured area

20

20
 Alternating Right-Sided Sequences

The ROC is shown by blue coloured area

21

21

Right-Sided Sequence with Multiple Poles

XA[n] = (1/2)nu[n] + 2.2nu[n]

22

22
 Left Sided Sequence of Infinite Duration

The ROC is shown by green

coloured area
23

23

Monotonic Left-Sided Sequences

The ROC is shown by blue coloured area

24

24
 Alternating Left-Sided Sequences

The ROC is shown by blue coloured area

25

25

Left-Sided Sequence with Multiple Poles

XB[n] = (-1/2)nu[-n-1] – 2.2nu[-n-1]

26

26
 Double-Sided Convergent Sequence

Two-sided sequence
1 1
x[n] = (− ) n u[n] − ( ) n u[−n − 1]
3 2
 −1
1 1
X ( z ) =  (− ) n z −n −  ( ) n z −n
n =0 3 n = − 2

1 1
= −
1 −1 1 −1
1+ z 1− z
3 2
1 1
z , z 1/3 1/2
3 2

27

27

Double-Sided Convergent Sequence

XC[n] = (1/2)nu[n] - 2.2nu[-n-1]

28

28
 Double-Sided Non-Convergent Sequence

XD[n] = (2)nu[n] - 2.(1/2)nu[-n-1]

29

29

Properties of Region of Convergence

1. In general, 0 ≤ rR < ROC < rL ≤ ∞

2. ROC is always a ring
3. X(ejw) absolutely converges if ROC includes Unit Circle
4. A causal and stable LTI system has all poles inside the unit circle.
5. ROC cannot contain a pole
6. FIR sequence → entire z plane, may be except for 0 or ∞
7. Right-sided sequence → outward of the outermost pole to ∞,
8. Left-sided sequence → inward from the innermost pole to 0
9. Two-sided sequence → a ring in between two adjacent rings
10. ROC is a connected region
30

30
 Properties of ROC

31

31

Example: ROC

32

32
 Example: ROC

33

33

Example: Sum of Two Exponential Sequences

34

34
 Example: Sum of Two Causal Exponential Sequences

+ =

35

35

Ex: ROC

36

36
 Causal System

37

37

Stability, Causality – Illustration - 1

1
x[ n] = ( ) n u[ n]
2

① Causal
② Stable
 1
X ( z ) =  ( ) n z −n =
1
1
n =0 2 1 − z −1
2
1
RoC : z 
2 1
1 1
2 1
2
① Outward
② UC  RoC
38

38
 Stability, Causality – Illustration - 2
1
x[n] = −( ) n u[−n − 1]
2
① Anti Causal
② Unstable

−1
1 1
X ( z) =  − ( 2) n
z −n =
1 −1
n = −
1− z
2
1
RoC : z 
1
1
2
2 2

① Inward
② UC  RoC 39

39

Stability, Causality – Illustration - 3

x[ n] = ( 2) n u[ n]

① Causal
② Unstable


1
X ( z ) =  2n z −n =
n =0 1 − 2 z −1

RoC : z  2
1 2
1 2
① Outward
② UC  RoC
40

40
 Stability, Causality – Illustration - 4
x[ n] = −2 n u[ − n − 1]
① Anti Causal
② Stable

−1
1
X ( z) = − 2 z
n = −
n −n
=
1 − 2 z −1

RoC : z  2
1 2

① Inward
② UC  RoC
Stable
41

41

ROC of Finite Length Sequence

42

42
 Complex Poles and Zeros
h[n] = αn cos ωon u[n]

43

43

2nd Order System at a Fixed Value of ω0=π/4

44

44
 2nd Order System at a Fixed Value of α=0.75

45

45

ROC

46

46
 DTFT from z-transform

47

47

Common z-Transform Pairs

48

48
 Properties of the z-transform

49

49

z-Trasform Properties

50

50
 Example: Linearity

51

51

z-Trasform Properties

𝛧
𝑑𝑒𝑙𝑎𝑦: 𝑥[𝑛 − 𝑛𝑑 ] ՞ 𝑧 −𝑛𝑑 𝑋(𝑧)

52

52
 z-Trasform Properties

53

53

Ex: Exp Multiplication

54

54
 Ex: z-Trasform Properties

55

55

z-Trasform Properties

56

56
 Proof: Convolution

57

57

Ex: Convolution

58

58
 Ex: Convolution

59

59

Ex: Convolution (c)

60

60
 Applications of Convolution
The convolution property of z-transforms allows us to
solve three basic problems of linear systems:
1. Filtering: If you know x[n] and h[n], you can obtain
y[n] by multiplication of the transforms,
Y(z) = X(z)H(z),
from which y[n] = Z-1{Y(z)}
1. System identification: If you know x[n] and y[n], you
can obtain h[n] by division of transforms,
H(z) = Y(z)/X(z) ,
from which h[n] = Z-1{H(z)}
3. Deconvolution: If you know y[n] and h[n], you can
obtain x[n] by division of transforms,
X(z) = Y(z)/H(z) ,
from which x[n] = Z-1{X(z)}. 61

61

Linear Constant-Coefficient Difference Equation

▪ Many practical problems appear in the form of

difference equations

▪ This difference equation defines an LTI system only if:

▪ All coefficients ak, bk are constant
▪ The initial conditions (or rest conditions) are zero
y[−N] = y[−N + 1] = . . . = y[−1] = 0
▪ Applying the linearity and time-shift properties of the
z-transform:

62

62
 Rational z-transform

63

63

Example: First-Order system

▪ First-order system: y[n] − ay[n − 1] = x[n]
▪ Calculating the z-transform:
Y (z) − aY(z)z−1 = X(z)

Y (z)(1 − az−1) = X(z)

𝑌(𝑧) 1
𝐻(𝑧) = =
𝑋(𝑧) 1 − 𝑎𝑧 −1
▪ This corresponds to the exponential: h[n] = anu[n].
▪ Questions: Why the causal exponential? For what values of a is this
system stable?
▪ This system is autoregressive i.e., the present output depends
on previous outputs
▪ Autoregressive systems have infinite impulse response (IIR)
▪ Systems with rational z-transforms with non-zero poles are IIR

64

64
 Example: First-Order system (contd.)

65

65

Example

66

66
 Example: Moving Average System

▪ This system has impulse response

h[n] = 1/M(δ[n] + δ[n − 1] + . . . + δ[n − M + 1])
▪ The impulse response only depends on a finite number of
previous inputs. Hence, this system has a finite impulse
response (FIR)

67

67

Example: Moving Average System (contd.)

68

68
 Example: Output of a Moving Average Filter
▪ Suppose the input signal frequency is ω0 = 0.2π

69

69

Example: Output of a Moving Average Filter

▪ Suppose the input signal frequency is ω0 = 0.5π

70

70
 Poles and Zeros

71

71

Ex: Poles and Zeros

72

72
 Poles and Zeros – In MATLAB
a=[1 -0.8741 0.9217 -0.2672];
b=[0.1866 0.2036 0.2036 0.1866];
zplane(b, a); % plots the zeros Z and poles P using row vectors of
b and a of the transfer function
[z, p, k] = tf2zp(b, a); % Transfer function to zero-pole
conversion
zplane(z, p); grid; % plots zeros and poles using column vectors of
z and p

polezero
73

73

Poles and Zeros – In MATLAB

▪ Given the transfer function, find pole and residue form
0.1866 + 0.2036𝑧 −1 + 0.2036𝑧 −2 + 0.1866𝑧 −3
𝐻 𝑧 =
1 − 0.8741𝑧 −1 + 0.9217𝑧 −2 − 0.2672𝑧 −3

a=[1 -0.8741 0.9217 -0.2672];

b=[0.1866 0.2036 0.2036 0.1866];
[r,p,k]=residuez(b,a)
r_mag = abs(r); r_phi = angle(r);
p_mag = abs(p); p_phi = angle(p);

▪ So
0.16𝑒 −𝑗3.09 𝑧 −1 0.16𝑒 𝑗3.09 𝑧 −1 1.20
𝐻 𝑧 = −0.70 + + +
1 − 0.86𝑒 𝑗1.27 𝑧 −1 1 − 0.86𝑒 −𝑗1.27 𝑧 −1 1 − 0.36𝑧 −1

poleresiduez

74

74