Signals And Systems

(Spring 2024)

z-Transform (z-T)

Geng-Shi Jeng (鄭耿璽), Ph.D.

Institute of Electronics
National Yang Ming Chiao Tung University
 Content

• Introduction
• The z-Transform
• Properties of the Region of Convergence (ROC)
• Properties of z-T
• Inverse of z-T
• The Transfer Function
• Causality and Stability
• The Unilateral z-T
• Determining Frequency Response from Poles and Zeros
• Computational Structure for Implementing DT LTI Systems (Skip)
 3

All Transforms
 4

Eigenfunction
 5

z-Transform (z-T)
<Continuous-Time> <Discrete-Time>
Fourier Series DTFS (DFT)
Fourier Transform DTFT
Laplace Transform z-Transform

• z-T is a generalization of DTFT.

• Types of z-Ts: Bilateral (two-sided) and Unilateral (one-sided)
• we use almost exclusively the bilateral z-T only
• Why study z-T?
• Powerful tool for signals and systems analysis – similar to FT
• Unilateral z-T (one-sided) for solving difference Eq.
 6

z-T Basics
• Importance of z-T in system analysis: Eigenfunctions of
discrete-time LTI systems
zn Discrete-Time H ( z) z n
LTI DT Transfer function

y[n ] = h[n ]  x[n ] =  h[k ]x[n − k ] .
k = −


  
Let x[n ] = z , y[n ] = k = z   h[k ]z −k  = H ( z ) z n
n n −k n
h[k ]z
= −  k =− 

➢ Note: H(z) depends only on z, but not on n.

j ( z )
➢ Polar form of H(z): H ( z ) = H ( z ) e . The system modifies the amplitude of the
input by | H ( z ) | and shifts its phase by  (z ) .
 7

z-T Basics
• z= 𝒓𝒆𝒋𝜴
• If r = 1, then z= 𝑒 𝑗Ω (i.e., DTFT)
 8

z-T

Z {x[n]} =  x[ n
n = −
] z −n
 X ( z)
j 1
Let z = re x[n ] =  X ( z ) z n −1
dz  Z −1
[ X ( z )]
2j r
Forward z-T
 
Z {x[n ]} =  x[n]z
n = −
−n
=  x[n ][re j ]− n
n = −

= r
n = −
−n
x[n ]e − jn = DTFT {r − n x[n ]}

Inverse z-T For a given r, as Ω goes from –π to π,

1 
x[n ]r −n =
2 
−
X ( re j )e jn d z = re j traverses the z-plane circle
of radius r in a counterclockwise
r +n direction.
x[n] =
2


−
X (re )e j jn
d =
1
2


−
( ) n
X (re j ) re j d

dz = d [re j ] = jre j d 1

n −1
x[n ] = X ( z ) z dz
d = 1j z −1dz 2j r
 9

DTFT is a special case of z-T

j
➢ DTFT may be viewed as a special case of z-T with z = e , i.e., with z on the
unit circle.
X (e j ) = X ( z ) | z =e j ; that is, r=1, the unit circle.
 10

Poles and zeros

• z-plane: The two-dimensional complex number plane of all
possible values of z
• Region inside/outside the unit circle
• Poles and zeros: Given a rational function in z as
~ M
−1
b + b z + ... + bM z −M
b k =1 (1 − ck z −1 ) x: pole
X ( z ) = 0 1 −1 = o: zero
k =1
−N
a0 + a1 z + ... + a N z N
(1 − d k z −1
)
~
b = b0 a0
zeros of X(z): ck
poles of X(z): d k

• The locations of poles and zeros completely specify X(z), except b~

for the gain factor
 11

Region of Convergence (ROC)

• ROC: The set of z values for which the forward z-T transform
converges, forming a region on the z-plane.

Z [ x[n]] = DTFT [ x[n]r − n ]

−n
A necessary condition for convergence is absolute summability of x[n ]r



n = −
x[n]r − n  

Example
x[n] = 0.5 n u[n]
 
1
X ( z ) =  0.5 z
n −n
=  (0.5 z −1 ) n =
n =0 n =0 1 − 0.5 z −1 , if the infinite sum converges.
−1
Condition for convergence of infinite sum: | 0.5 z | 1  | z | 0.5
 12

ROC: An example

DTFT {𝒂𝒏 𝒖 𝒏 }does not

exist when 𝒂 > 1
ZT{𝒂𝒏 𝒖 𝒏 } exists
if 𝒓 > 𝒂
Multiplied by 𝒓−𝒏
𝒙 𝒏 𝒓−𝒏 = 𝒂𝒏 𝒓−𝒏 𝒖 𝒏
𝒂
= ( )𝒏 𝒖[𝒏]
𝒓
 13

Example 1: Causal

<Sol.>  
1
X ( z ) =  a z =  (az −1 ) n =
n −n

n =0 n =0 1 − az −1
−1
ROC: | az | 1  | z | a
 14

Example 2: Anti-causal

<Sol.> i = -n
−1  
a −1 z 1
X ( z) = − a
n = −
n
z −n
= − a z = − ( a z ) = −
i =1
−i i

i =1
−1 i
=
1 − a −1 z 1 − az −1

ROC: | a −1 z | 1  | z | a

Note:
• Different DT signals can have the same
expression for the (bilateral) z-T but with
different ROC’s.
• To completely and uniquely specify a z-T,
the ROC must be given.
 15

Example 3: Bounded ROC

<Sol.>

k = -n
 16

Properties of the ROC

 17

Properties of the ROC

• The ROC cannot contain any poles
• Same as LT
• The ROC consists of a “ring” centered about the origin.
Proof



−n
Recall that z = re
j
, if x[n ] r   , x[n]r − n has DTFT that converges.
n = −

Thus, any z that has the same |r| satisfies the condition.

Remarks:
➢ Thus, if a specific value of z is in the ROC, then all values of z on the same circle
will be in ROC.
➢ Inner boundary may extend inward to the origin.
➢ Outer boundary may extend outward to infinity.
 18

Properties of the ROC

• If x[n] is finite-duration, then the ROC is the entire z-plane
except possibly z = 0 and/or z = 
Proof
A finite-duration sequence has only a finite number of nonzero values:
N2
X ( z) =  x[n]z
n = N1
−n
, where N1 and N2 are finite.
For z≠0,∞, each term will be finite.
=> X(z) converges.

N1≧0 => only negative powers of z (causal)

=> ROC includes z =∞
N2≦0 => only positive powers of z (anti-causal)
=> ROC includes z=0
 19

ROC (Exponential bounded signals)

If 𝑰− 𝒛 < ∞ 𝒂𝒏𝒅 𝑰+ 𝒛 < ∞, 𝒕𝒉𝒆𝒏 𝑿(𝒛) < ∞

 20

ROC (Exponential bounded signals)

= NULL

ROC of a right-sided signals is z  r+ except possibly z = ∞

ROC of a left-sided signals is z  r− except possibly z = 0
ROC of a two-sided signals is r+  z  r−
 21

ROC (Exponential bounded signals)

(a) A right-sided signal has an

ROC of the form |z| > r+.

(b)A left-sided signal has an

ROC of the form |z| < r–.

(c) A two-sided signal has an

ROC of the form r+ < |z| < r–.
 22

Example 1
 23

Example 1
 24

Example 2
• Find all possible ROCs of
z 2z
X ( z) = +
z + 12 z − 14

<Sol.>
(a) x[n] = −( −1/ 2)n u[−n − 1] + 2(1/ 4) n u[n]

(b) x[n] = ( −1 / 2) n u[n] + 2(1 / 4) n u[n]

(c) x[n] = −( −1/ 2)n u[−n − 1] − 2(1/ 4) n u[ −n − 1]

 25

Properties of Bilateral z-T

 26

Properties of Bilateral z-T

z z
Let x[n]  X ( z ) and y[n]  Y ( z ) with ROC’s Rx and Ry, respectively
z
⚫ Linearity: ax[n] + by[n]  aX ( z ) + bY ( z ) , ROC  Rx  Ry (At least as
large as their intersection; larger if pole/zero cancellation occurs)
z
−1
⚫ Time reversal: x[ − n ]  X ( z ) , ROC  1/Rx ( z 0  ROC for x[n] ➔
1
 ROC
z0 for x[-n])
z
−n0
⚫ Time shift: x[ n − n 0 ]  z X ( z ) , ROC = Rx, except possibly z = 0 or |z| = ∞.
z
z
 x[n ]  X 
n

⚫ z-scale (modulation):  ,
ROC = |α| Rx (Expands or contracts) ∞ ∞
𝒛
Proof ෍ 𝜶𝒏 𝒙[𝒏]𝒛−𝒏 = ෍ 𝒙[𝒏](𝜶−𝟏 𝒛)−𝒏 = 𝐗( )
𝜶
𝒏=−∞ 𝒏=−∞
 27

More on multiplication → z-scale

 28

Properties of Bilateral z-T

z
⚫ Convolution: x[n] * y[n]  X ( z )Y ( z ) ,
ROC  Rx  Ry (At least as large as their intersection; larger if pole/zero
cancellation occurs)
z d
⚫ Differentiation in z-domain: nx[n ] − z dz X ( z ) , ROC = Rx
−1
⚫ Time differencing: x[n] − x[n − 1] ⎯→(1 − z ) X ( z ) ,
z
ROC = Rx −{0}.
⚫ Summation:
n
X ( z)
 x[i ] ⎯→
i =0
z

1 − z −1 where n  0 ,
ROC  {RX  { z  1}}
 29

Properties of Bilateral z-T

Will show later in

Example 3

Why?
 30
Example 1: Pole-zero cancellation to
extend ROC

Why?

<Sol.>

The pole z = ½ is
cancelled. ROC is
enlarged.
 31

Example 2: Many properties

<Sol.>
 32
Example 3: Damped complex
exponential

<Sol.>
 33

Inverse of Bilateral z-T

 34

Inverse of Bilateral z-T

• Consider inverse z-T (bilateral) of

b0 + b1 z −1 + ... + bM z − M
X ( z) =
a0 + a1 z −1 + ... + a N z − N

If M  N , then we use long division to express

M −N ~
~ B ( z)

~
X ( z) = f k z −k + X ( z ) where X ( z) =
k =0 A( z )

so that X~ ( z ) is strictly proper.

 35
Inverse of Bilateral z-T
(Distinct poles)
M −N ~
f −k ~ ~ B( z)
X ( z) = k z + X ( z) X ( z) =
k =0 A( z )
M −N ~
For f
k =0
k z −k
, we have For
~
X ( z) =
B( z)
A( z ) , Partial-fraction expansion (PFE)
◼ No repeated poles:
M −N z M −N N
Ak
X ( s) = 
~

k =0
f k [n − k ]  f z
k =0
k
−k

k =1 1 − d k z −1 .

Depending on the ROC:

 [n − m]  z , m  0, z  0
−m
z Ak
Ak (d k ) u[n] 
n
ROC  ( z  d k )
1 − d k z −1 , if
z Ak
− Ak (d k ) u[−n − 1] 
n
ROC  ( z  d k )
1 − d k z −1 , if
 36
Example 1: Inverse z-T with distinct
poles
−1 −2
Given X ( z ) = 1 − z + z 1 2 2
= + −
(1 − 12 z −1 )(1 − 2 z −1 )(1 − z −1 ) (1 − 12 z −1 ) (1 − 2 z −1 ) (1 − z −1 )
Determine DT signals according to the following possible ROCs

If ROC = (1  z  2) , then
n

x[n ] =   u[n ] − 2(2 n )u[−n − 1] − 2u[n ] .

1
2
If ROC = ( 2  z  1) , then
1

x[n ] =   u[n ] − 2(2 n )u[−n − 1] + 2u[ −n − 1]

1
2 .

If ROC = ( z  2 ) , then
1

x[n ] = −  u[−n − 1] − 2(2 n )u[−n − 1] + 2u[−n − 1] .

1
2
 37
Inverse of Bilateral z-T
(Repeated poles)
◼ If a pole d i is repeated r times, then there are r terms in the PFE associated with
the pole:
Ai1 Ai2 Air
1 − d i z −1 , (1 − d i z −1 ) 2 , …, (1 − d i z −1 ) r .

Then,

A(n + 1) (n + m − 1) z A
( di ) u[n] 
n

( m − 1)! (1 − d i z −1 ) m , if ROC  ( z  d i )
A( n + 1) ( n + m − 1) z A
( i)
n
− − − 
(1 − d i z −1 ) m , if ROC  ( z  d i )
d u[ n 1]
( m − 1)!
 38

Example

<Sol.>

Use long division

 39

Remarks
• If we know causality/stability of a signal’s DTFT, then we can
determine the inverse z-T.
 40

Inverse z-T by power series expansion

>

<
 41

Inverse z-T by power series expansion

<
 42

Example 2: NON z (z -1 ) polynomials

 43

The Transfer Function of DT

LTI Systems
 44

Transfer function
• The transfer function of a DT LTI system is the z-T of the
impulse response. Therefore, if x[n] and y[n] are system input
and output, respectively, then transfer function is
H ( z) = Y ( z) / X ( z)

• N-order DT LTI system described by the difference equation

N M

a k y[n − k ] =  bk x[n − k ]
k =0 k
k =0 k =0 z M −k
x[n − n0 ]  z −n0 X ( z ) b z
z-T H ( z) =
k =0 k −k
N M N

 k  k X ( z)
−k −k a z
a z Y ( z ) = b z
k =0 k =0
 45

Transfer function

~
 −k
bM  k =1 (1 − ck z −1 )
M M
bk z
H ( z) = k =0 factor H ( z) =
  k =1 k )
−1
−
N −k N
ak z (1 d z
k =0
~
b = b0 / a0
 46

Example

<Sol.>
 47

Causality and Stability of

DT LTI Systems
 48

Recall: Example 1: Causal

<Sol.>  
1
X ( z ) =  a z =  (az −1 ) n =
n −n

n =0 n =0 1 − az −1
−1
ROC: | az | 1  | z | a
 49

Recall: Example 2: Anti-causal

<Sol.>
−1  
a −1 z 1
X ( z) = − a
n = −
n
z −n
= − a z = − ( a z ) = −
i =1
−i i

i =1
−1 i
=
1 − a −1 z 1 − az −1

ROC: | a −1 z | 1  | z | a

Note:
• Different DT signals can have the same
expression for the (bilateral) z-T but with
different ROC’s.
• To completely and uniquely specify a z-T,
the ROC must be given.
 50

Causality
ROC includes unit circle

Causal &
decaying
exponential
(STABLE)

Causal &
increasing
exponential
(UNSTABLE)
 51

Stability

ROC includes unit circle

Causal &
decaying
exponential
(STABLE)

NONcausal &
decaying
exponential
(STABLE)
 52

Causal + Stable

all of its poles are inside the

unit circle in z-plane
 53

Example 1
 54

Remarks: Causality and Stability

• Given the pole-zero plot and the ROC of a DT LTI
system,
• The system is causal iff the ROC is exterior of a circle including infinity.
• The system is anticausal iff ROC is inside a circle including 0.
• The system is stable (impulse response is absolutely summable) iff the
ROC includes the unit circle.
• The system is stable and causal iff all poles are inside the unit circle
 55

Inverse system
• We know that the impulse responses of a system and its
inverse are related by
h inv [n] * h[n] =  [n]
z-T
H inv ( z ) H ( z ) = 1 H inv ( z ) = 1 / H ( z )

~
k =1 k )
−1
− 
M
−1
−
N
bM (1 c z (1 d z )
H ( z) = H ( z) = ~
inv k =1 k

k =1 bM k =1 (1 − ck z −1 )
−1
−
N M
(1 d k z )

Hence poles  zeros and zeros  poles in system inversion

 56

Remarks: Inverse system

• Hinv(z) in the last slide only specifies the algebraic form but
not its ROC. Any ROC allowed by the pole-zero plot can be
chosen, if we do not care about the stability or causality of
the inverse system

• Poles of Hinv(z) = Zeros of H(z)

• Hinv(z) is stable and causal iff all zeros of H(z) are inside the unit circle
• A stable, causal system that has all its poles and zeros inside
the unit circle is said to have minimum phase
• A minimum-phase system has a stable, causal inverse
 57

Unilateral z-Transform &

Its Properties
 58

Unilateral z-T
• In causal system analysis, we often are interested in obtaining
system response for n>=0 under some initial conditions
• Useful for studying LTI by the difference equation with ICs

X ( z ) =  x[n ]z −n
n =0

1

n −1 −1
x[n ] = X ( z ) z dz  Z [ X ( z )]
2j r

• We do not have to specify the ROC, because it is effectively

bilateral z-T for a right-sided signal, which can converge only if
the ROC is outside a DISK in the z-plane
 59

Properties of u z-T (time-shift)

• Major difference between bilateral and unilateral z-T time-shift
z
Bilateral z-T x[n − n0 ]  z −n0 X ( z )
Unilateral z-T x[n − 1]  z −1 X ( z ) + x[−1]
<Proof>
Let w[n ] = x[n − 1]

W ( z ) =  w[n ]z −n
n =0
 
=  x[n − 1]z −n
= x[−1] +  x[n − 1]z −n
n =0 n =1
 
= x[ −1] +  x[m]z z −m −1
= x[−1] + z −1
 x[m]z −m
= x[−1] + z −1 X ( z )
m =0 m =0

<Generalization>
x[n − k ]  z − k X ( z ) + x[−1]z − k +1 +  + x[−k + 1]z −1 + x[−k ]
 60

Solving DE with ICs

Find y[n] of the difference eqn.
y[n] − 0.1 y[n − 1] − 0.02 y[n − 2] = 2 x[n] − x[n − 1]
with y[−1] = −10 and y[−2] = 20 and x[n ] = u[n ] .
<Sol.>
Take single-sided z-T:
   
Y ( z ) − 0.1 z −1Y ( z ) + y[−1] − 0.02 z −2Y ( z ) + z −1 y[−1] + y[−2]
= 2 X ( z ) − z −1 X ( z )
1
where X ( z ) =
1 − z −1 .
−1 −2 −1 2 − z −1
Thus Y ( z ) − 0.1[ z Y ( z ) − 10] − 0.02[ z Y ( z ) − 10 z + 20] =
1 − z −1

−1 −2 2 − z −1 −1
[1 − 0.1z − 0.02 z ]Y ( z ) = −1
− 0.2 z − 0.6
1− z
 61

Solving DE with ICs

To find PFE of Y(z): (Here shows a trick of pre -multiply Y(z) by 1/z. Then, you can
work on z (rather than z-1).
Y ( z) 1.4 z 2 − 0.6 z + 0.2
=
z ( z − 1)( z − 0.2)( z + 0.1)
A A2 A3
= 1 + +
z − 1 z − 0.2 z + 0.1
1.4 − 0.6 + 0.2 1.0
A1 = = = 1.136
0.8  1.1 0.88
1.4  0.2 2 − 0.6  0.2 + 0.2
A2 = = −0.567
− 0.8  0.3
1.4  (−0.1) 2 − 0.6(−0.1) + 0.2
A3 = = 0.83
− 1.1  (−0.3)
1 1 1
=> Y ( z ) = 1.136 − 0.567 + 0.83 
1 − z −1 1 − 0.2 z −1 1 + 0.1z −1
=> y[n] = 1.136 u[n] − 0.567 (0.2) u[n] + 0.83  ( −0.1) u[n]
n n
 62

Initial value theorem

If x[n]=0, n<0, then

x[ 0] = lim X ( z )
z →

<Proof>

X ( z ) =  x[n]z − n
n =0

As z→∞ , z-n → 0
Only when n=0, z-n =0
 x[0] = lim X ( z )
z →
 63

Final value theorem

If (1) x[n]=0, n<0, and
−1
(2) All singularities of (1 − z ) X ( z ) are inside the unit circle,
−1
then
x[] = lim (1 − z ) X ( z)
z →1

Remarks: (1) If all poles of X(z) are inside unit circle, x[n] → 0 as n → 
(2) If there are multiple poles at “1”, x[n] →  as n → 
(3) If poles are on the unit circle but not at “1”, x[n] → cos  0 n
 64

Final value theorem

 N
<Proof>
X ( z ) =  x[n ]z −n
= lim  x[n ]z −n
n =0
N →
n =0
(*1)

Z x[n − 1] = z −1 X ( z ) − x[ −1]

 N
=  x[n − 1]z −n
= lim  x[n − 1]z −n (*2)
N →
n =0 n =0

(*1) – (*2) ➔
(1 − z −1 ) X ( z ) − x[−1]
N 
( )
= lim  x[n] z − n − z −(n −1) + x[ N ]z − N −
1
x[−1]
N →
 n =0 
As z→1,
lim (1 − z −1 ) X ( z ) − x[−1] = x[] − x[−1]
z →1

(
 x[] = lim 1 − z −1 X ( z )
z →1
)
 65

Determining the Frequency

Response from Poles and
Zeros
 66

Frequency response
• The frequency response corresponds to the transfer function
evaluated on the unit circle in the z-plane
~
k =1 k )
−1
−
M
bM (1 c z
H ( z) =
k =1
−1
−
N
(1 d k z )

z = e j (unit circle)
~ ~
k =1 k ) bM e ( N − M ) j k =1 (e j − ck )
− j
−
M M
bM (1 c e
H (e j ) = =
k =1 (1 − d k e k =1 − d k )
N − j N j
) ( e

Magnitude
~
k =1 − ck )
Mj
bM ( e j
H ( e j ) = A ratio of products of e − a
k =1 − d k )
Nj
( e , where a is either a pole or a zero.
 67

Frequency response
~
k =1 − ck )
M j
bM ( e
H ( e j ) =
k =1 − d k )
N j
( e
• The location of g (poles or zeros) affects the shape of the
frequency response. A peak (notch) is produced when a pole
(zero) is close to the unit circle as shown below
 68

Example
𝑯 𝒛 = 𝟏 − 𝒂𝒛−𝟏 𝟏/𝑯 𝒛 = 𝟏/(𝟏 − 𝒂𝒛−𝟏 )

a = 0.5ej/4

a = 0.8ej/4

a = 0.95ej/4
Close to 1 →
notch/peak
 69

Appendix: ZT Table
 70

Basic z-Transform
 71

Bilateral ZT (anticausality)
 72

ZT Properties
 73

Other properties
 All proof for Properties of 74

Bilateral z-T
• https://zh.wikipedia.org/wiki/Z%E8%BD%89%E6%8F%9B