Chapter 6 z-Transform

➢ Definition of z-transform
➢ Rational z-transform
➢ z-transform theorems
➢ Inverse z-transform
➢ Transfer function

1
➢ Definition of z-transform
➢ Rational z-transform
➢ z-transform theorems
➢ Inverse z-transform
➢ Transfer function

2
Digital Signals

square wave

triangular wave

？？
{1 , -9 , 33 , -45 , 6 , 36 , 8}
How to find out the characteristics hidden in digital signals
3
Looking at digital signals from another perspective

{1,-9,33,-45,6,36,8}
Polynomial coefficient
x − 9 x + 33x − 45 x + 6 x + 36 x + 8
6 5 4 3 2

= ( x − 3 x + 3 x + 1)( x − 6 x + 12 x + 8)
3 2 3 2

{1,-3,3,1} {1,-6,12,8}
Value of sequence → polynomial coefficient
Benefits: factorization, rooting, residue theory, etc 4
Polynomials (algebraic equations) have perfect analytical theory
 The 18th century BC, Babylon, The Babylonian clay plate records the quadratic
equation
 3rd century AD, China, Zhao Shuang Gives a Root of the Quadratic Equation in
Solving the Area Problem
 7th century AD, India, Brahmagu gave a root formula for a quadratic equation
 9th century AD, Arab, Hualazimi gave a general solution to the quadratic
equation of one variable
 1247, China, In Qin Jiushao's " 数学九章 ", a formula for finding the root of a
univariate cubic equation was published
 sixteenth century, France & Britain,Veda and Newton distributions give the
relationship between roots and coefficients (Veda theorem)
 sixteenth century, Italy, Fontana gave the radical solution of the cubic equation of
one variable → Caldano formula published by Caldano (Cardano formula)
 In 1629, Dutch, Gerald asserted that any n-degree equation has n roots
 In 1799, German, Gauss gave the first substantive proof of Gerald's assertion
 In the 19th century, Norway & France, Abel and Galois have studied that
equations of more than five degrees cannot be solved by formulas → pioneered
group theory
5
Definition

X ( z ) = Z  x  n  =  x  n  z − n （ z为复数）
n =−

( z = re j is a continuous complex variable )

When r = 1 ， z = e j. Then z-transform becomes

discrete-time Fourier transform (DTFT).

X（e j）= 
n =−
x ( n )e − jn

Therefore, DTFT is z-transform on the unit circle.

6
Ex1 Determine z-transform of the causal sequence x[ n ] =  n u[ n ] .
   1
[A]：X ( z ) =   u[ n ] z
n −n
=  z n −n
=  ( z ) =
−1 n

n = − n=0 n=0 1 −  z −1
When |  z −1 | 1 , X(z) converges.

j Im [z ]
Region of convergence
(ROC) is | z | |  | .
Z＝ R e[z ]

7
Ex2 Determine the z-transform of the anticausal sequence
x[ n ] = − n u[ − n − 1] .
[A]： Z＝ j Im [z ]

R e[z ]

When |  −1 z | 1 , X(z) converges.

Region of convergence (ROC) is | z | |  | .

Different sequences may have the same z-transform.

8
Region of Convergence (ROC)

Definition
The set of values of z for which its z-transform converges.

Convergence condition

X(z) converges 
n =−
x(n) z − n = M  

9
Shape of ROC
j
Let z = re
 


n = −
x[ n ] z − n = 
n = −
x[ n ]r − n   X(z) converges

j Im [z ]

ROC depends on |z|, Rx−

therefore the
收敛域与 shape of
z 有关， R e[z ]

ROC is circle.
因此一般是环形 R
x+

10
Finite-length sequences

 n1  n  n2 n2
x[n ]
x[n ] =  X ( z ) =  xn z − n
0 otherwise n = n1

n2
For z  ( 0 ,  ) ,  x[ n
n = n1
] z −n
 X(z) converges

Note：

11
Right-sided sequences
x(n)

 x ( n), n  n1
x (n) =  ...
0, n  n1 ..
n1 0 1 n

−n
X ( z ) =  x[ n ] z j Im [z ]
n = n1

ROC: | z | R x −

R e[z ]
n1≥0 → causal sequence
Rx−
12
Left-sided sequences
x(n)

 x ( n ), n  n2
x(n) = 
0, n  n2 
0 n2 n
n2
−n
X ( z ) =  x[ n ] z j Im [z ]
n = −

ROC: | z | R x +
R e[z ]

Rx +

13
n
Two-sided sequences
  −1
X (z) =  x n z −n =  x n z −n +  x n z −n
n = − n=0 n = −

ROC is |z|>RX- ROC is |z|<RX+

RX-< |z|< RX+ → X(z) converges

j Im [z ]
j Im [z ] j Im [z ]

+
Rx−
=
R e[z ] R e[z ] R e[z ]
Rx− Rx +
Rx +
14
Ex3 Determine the z-transform of the sequence x[ n ] =  n .
 −1
[A]:
解： U (z) =   n z− n +   n z−n
n =0 n =−

收 敛域为 z  
ROC: 收 敛域为 z  
ROC:

No overlap of two ROCs → X(z) does not exist

15
Commonly used z-transform pairs

16
➢ Definition of z-transform
➢ Rational z-transform
➢ z-transform theorems
➢ Inverse z-transform
➢ Transfer function

17
Rational functions: ratios of two polynomials

P ( z ) p0 + p1 z + p 2 z +  + p M z
−1 −2 −M
G (z ) = =
D ( z ) d 0 + d1 z −1 + d 2 z − 2 +  + d N z − N
The factored form：numerator roots - zeros
denominator roots - poles

 (1 −  z )
M M

l
−1
 (z −  )
l
G (z ) =
p0 N −M p0
l =1
=z l =1

 (1 −  z )
N N

 (z −  )
d0 −1 d0
l l
l =1 l =1

Control property of filters

18
Pole-zero plot

In z plane, “×” - poles，“○” - zeros

1 −1
1+ z
1
Ex Develop the pole-zero plot of X ( z ) = 4 , | z | .
1 −1 2
1− z
2

19
➢ Definition of z-transform
➢ Rational z-transform
➢ z-transform theorems
➢ Inverse z-transform
➢ Transfer function

20
Linearity

ax1 [ n ] + bx 2 [ n ]  aX 1 ( z ) + bX 2 ( z )

Is this property always valid？

R OC  R OC 1  R OC 2

21
 1 1
例： x[ n ] = u[ n ] − a u[ − n − 1] 
Ex: n
−1
+ −1
,1  z  a
1− z 1 − az

j Im [z ]
j Im [z ] j Im [z ]

+
Rx−
=
R e[z ] R e[z ] R e[z ]
Rx− Rx +
Rx +

22
Time-shifting
−d
x[n − d]  z X (z)
ROC : ROC x , excluding possibly the point z=0 or z=∞

Special case：d=1 delay unit z-1

Multiplication by an exponential sequence

n
z0 x[n]  X ( z / z0 ) ROC = z 0 ROC x
23
Differentiation
d X( z )
n x[ n ]  − z
dz
ROC : ROC x , excluding possibly the point z=0 or z=∞

Time-reversal
−1
x[−n]  X ( z ) ROC : 1 / ROC x

24
Conjugation
  
x [n]  X (z ) ROC : ROC x

The initial-value
If x[n] is a causal sequence, then

lim X ( z ) = x[0]
z →

25
Convolution

x1[n] x2[n]  X1(z)X2 (z)

RO C  RO C 1  RO C 2

26
 n
 1
Ex: Determine the z-transform of y[ n ] =   u[ − n ] and its ROC.
3
−n
解 ：Let
[A]: 令 x[n ] = 3 u[n ] 
n
y[n ] = x[ − n ] = 3 u[ − n ]

1
x[ n ] ⎯⎯ ⎯
→ X ( z) =
z变换
−1
, | z || 3 |
1 − 3z

Time reversal 1 1 1
y [ n ] ⎯⎯ ⎯ ⎯ ⎯
时间反转性质
→ Y ( z) = X ( ) = ,| z || |
z 1 − 3z 3

27
Ex: Determine the z-transform of y[ n ] = ( n + 1) a n u [ n ] and its ROC.

解 ：Let令 x [ n ] =
[A]: a n u[n ]  y [ n ] = nx [ n ] + x [ n ]
1
x[ n ] ⎯⎯ ⎯
→ X ( z) =
z变换
−1
, | z || a |
1 − az
dX ( z )
differentiation az −1
nx[ n ] ⎯⎯ ⎯ ⎯
→ −z
微分性质
= −1 2
,| z || a |
dz (1 − az )

linearity 1 az −1 1
y [ n ] ⎯⎯ ⎯ ⎯
线性性质
→ Y ( z) = −1
+ −1 2
= −1 2
,| z || a |
1 − az (1 − az ) (1 − az )

28
➢ Definition of z-transform
➢ Rational z-transform
➢ z-transform theorems
➢ Inverse z-transform
➢ Transfer function

29
Direct transform

X ( z ) = Z  x  n  =  x n z −n
（ z为复数）
n =−

How to obtain inverse z-transform?

Principle: Integrate z k counterclockwise over a closed
contour C.

Cauchy’s integral theorem

30
 
X (z) =  x n z −n = + x[0] + x[1] z −1 + x[2] z −2 + + x[ n ] z − n +
n =−

 X ( z ) dz = + 0 + 2 j  x 1 + 0 + +0+
c

在 正 Z 变 换both
Multiply 的 等sides
式两边 同 乘 以 z − 1、 z 0、 z 1
with z n －1

Use Cauchy’s integral theorem

Divide by 2j to obtain

即可得系数 x  0 、 x 1、 x  2  x n

Therefore

31
Method to compute inverse transform

✓ Cauchy’s residue theorem

✓ Partial-fraction expansion

✓ Long division

32
Cauchy’s residue theorem

To solve

x[ n ] =  [ residues of X ( z ) z n −1 at the poles inside c ' ]

33
Partial-fraction expansion

P ( z ) p0 + p1 z + p 2 z +  + p M z
−1 −2 −M
G (z ) = =
D ( z ) d 0 + d1 z + d 2 z +  + d N z
−1 −2 −N

P(z ) M −N
G (z ) =
P1 ( z )
D (z )
= 
l =0
l z +−l

D(z)

 n − l   z −l P1 ( z )
is proper fraction
D( z)

34
 P1 ( z )
For the proper fraction G1 ( z ) =
D(z )
• Simple poles: D(z) has N distinct roots.

D ( z ) = (1 −  0 z − 1 ) (1 − 1 z − 1 )  (1 −  N −1 z − 1 )
P1 ( z )
G1 ( z ) =
(1 −  z )(
0
−1 −1
) (
1 − 1 z  1 −  N −1 z −1
)
0 1  N −1
= + ++
1 − 0 z −1
1 − 1 z −1
1 −  N −1 z −1

 i = (1 − i z −1
) G (z )
1
z = i
35
Inverse z-transform
0 1  N −1
G1 ( z ) = + ++
1 − 0 z −1
1 − 1 z −1
1 −  N −1 z −1
For the right-
sided sequence

g1 n  =  0 (0 ) un  + 1 (1 ) un  +  +  N −1 ( N −1 ) un 

n n n

ROC： z  max l

36
 z
Ex: H ( z ) = , 若 h[n ] 为因果序列，求
compute h[n]. h[n ]
( z − 1)( z − 2)
−1
解： H ( z ) =
[A]: A1
+
A2 −1 A (1 − z )
(1 − z ) H ( z ) = A1 + 2
−1
(1 − z ) (1 − 2 z −1 ) (1 − 2 z −1 )

−1 
−1 z 
A1 = (1 − z ) H ( z ) −1 = (1 − z )   = −1
z =1
 ( z − 1)( z − 2)  z −1 =1
同 理 可 得 A2 = 1
Similarly

When ROC is z  1, h[ n ] = u [ − n − 1] − 2 u [ − n − 1]
n

When ROC is 1  z  2 , h[ n ] = − u [ n ] − 2 n u [ − n − 1]

When ROC is z  2 , h[ n ] = − u [ n ] + ( 2 ) u [ n ]
n

37
 P1 ( z )
For the proper fraction G1 ( z ) =
D(z )
• Multiple poles: D(z) has a root z = v of multiplicity L
P1 ( z ) 1 2 L
G1 ( z ) = = + ++
(1 − vz ) −1 L 1 − vz −1
(1 − vz )
−1 2
(1 − vz )
−1 L

(1 − vz ) G ( z ) =  (1 − vz )
−1 L
1 1
−1 L −1
+  2 (1 − vz )
−1 L − 2
+  + L
Take the derivative
of both sides

d L −i
((1 − vz ) G (z )) = d
−1 L
1
L −i
( (1 − vz ) ) +  + (L − i )!(− v )
1
−1 L −1
L −i
i
( )
d z −1 L − i
( )
d z −1 L − i

38
d L −i
((1 − vz ) G (z )) = d
−1 L
1
L −i
( (1 − vz ) ) +  + (L − i )!(− v )
1
−1 L −1
L −i
i
( )
d z −1 L − i
( )
d z −1 L − i

Set z = v, then right side

leaves only one term

i =
1
(L − i )!(− v ) d z
L −i
d L −i
−1 L − i
( )
−1 L
(
1 − vz G1 ( z ) 1 i  L ) 
z=v

39
 P1 ( z ) 1 2 L
G1 ( z ) = = + ++
(1 − vz )
−1 L 1 − vz −1
(1 − vz )
−1 2
(1 − vz )
−1 L

pi
What is the inverse z-transform of ？
(1 − v z )
i
−1 i

Hint：Use differentiation theorem

40
 P1 ( z )
For the proper fraction G1 ( z ) =
D(z )
• Mix of simple poles and multiple pole:

D ( z ) = (1 −  0 z )  (1 −  N −1 z ) 1 − v z
−1 −1
( )
−1 L

P1 ( z ) 0  N −1 i
G1 ( z ) =
L
++ +
( )
=
D ( z ) 1 − 0 z −1
1 −  N −1 z −1
i =1 1 − vz −1
i

For the right-

sided sequence

g 1 n  =  0 ( 0 ) u n  +  +  ( N −1 ) u n  + g 2 [ n ]
n n
N −1

41
 z −1
Ex: Determine the inverse z-transform of X ( z ) = −1 2
, | z | 1 .
(1 − z )

[A]: 解： z = 1z处有二重根，
The pole X ( z )的部分分式展开为：
= 1 is of multiplicity 2
A1 A2
X ( z) = −1
+
1− z (1 − z −1 ) 2

 d   dz −1 
( )
2
A1 = ( −1) •  −1 1 − z −1 X ( z )  = ( −1) •  −1  = −1
 dz  z =1  dz  z =1

(
A2 =  1 − z −1 ) X ( z) = 1
2

  z =1

42
 1 1
X ( z) = − −1
+
1− z (1 − z −1 ) 2

1
 −1 2
⎯⎯⎯→ ( n + 1)u[ n ]
(1 − z ) 逆 z变换

x [ n ] = − u [ n ] + ( n + 1) u [ n ]

43
Long division

Expand X(z) into a power series in z −1 .

In the series expansion, the coefficient multiplying the

−n
term z is x[n].

The series expansion can be obtained from the

rational fraction by long division.

44
 1 −1
1− z
Ex: Determine the inverse z-transform of X ( z ) = 3 1
, z  .
1 −2 2
1− z
4
1 1 1
1 − z −1 + z −2 − z −3 + 
3 4 12
1 −2 1
1− z 1 − z −1
4 3
1
1 − z −2
4
______________________
 1 1 1 
1 1 x[n ] = 1,− , ,− ,
− z −1 + z −2  3 4 12 
3 4
1 1
− z −1 + z −3
3 12
______________________
1 1
+ z − 2 − z −3
4 12

45
 1
1 − z −1
Ex. 1
X ( z) = 3 , z 
1 −2 2
1− z
4
4
z +4 z 2 +
3
1 1
− z −2 + 1 − z −1 + 1
4 3
1 4
− z −1 z
3 3
______________________
4
1 - z  4 
3 x[n] = ..., 4, − , 0 
1 4z2  3 
______________________
4
- z -4 z 2
3
➢ Definition of z-transform
➢ Rational z-transform
➢ z-transform theorems
➢ Inverse z-transform
➢ Transfer function

47
Transfer function (System function)

X z Y z
H z

Y (z) = X ( z ) H ( z )

Y (z)
H (z) =
X (z)

48
From difference equation to transfer function

d0y[n] + d1y[n-1] + d2y[n-2] + … + dNy[n-N]

= p0x[n] + p1x[n-1] + p2x[n-2] + … + pMx[n-M]

Y ( z) p + p z −1
+ ... + p z −M  k
p z −k

H ( z) = = 0 1
−1
M
−N
= k =0
X ( z ) d 0 + d1 z + ... + d N z N

d
k =0
k z −k

49
Ex: Determine the transfer function of the following
difference equation.
2 y [ n ] + y [ n − 1] + 0 .9 y [ n − 2 ] = x[ n − 1] + x[ n − 4 ]
[A]: z-transform

2Y ( z ) + z − 1Y ( z ) + 0.9 z − 2Y ( z ) = z − 1 X ( z ) + z − 4 X ( z )

(2 + z − 1 + 0.9 z − 2 )Y ( z ) = ( z − 1 + z − 4 ) X ( z )

Y ( z) z −1 + z −4
H ( z) = =
X ( z ) 2 + z −1 + 0.9 z − 2
50
Ex: Determine the difference equation of H ( z ) = z + 4 .
z + 0.3
z+4
[A]： H (z) =
z + 0.3

1 + 4 z −1 Y ( z)
H ( z) = −1
=
1 + 0.3 z X ( z)

 Y ( z ) + 0.3 z − 1Y ( z ) = X ( z ) + 4 z − 1 X ( z )

 y ( n ) + 0.3 y ( n − 1) = x ( n ) + 4 x ( n − 1)
51
Transfer function and impulse response

H ( z ) = Z h[ n ] =  h[ n ] z −n

n = −

Transfer function and frequency response

when r = （
1 z = 1），z-transform = DTFT.

z=e jω represents a unit circle.

Therefore, z-transform on unit circle is DTFT.

52
Geometry explanation of frequency response

 (1 −  z )
M M

Y (z ) p0
l
−1
 (z −  )
l
H (z ) = l =1 N −M p0 l =1
= =z
X (z ) d 0
 (1 −  z )
N d0 N

l =1
l
−1
 (z −  )
l =1
l

z −  l ---- zero vector

z − l ---- pole vector

l =1 −  l )
j
M
j p 0 j ( N − M ) ( e
H (e ) = e
l =1 − l )
j
N
d0 ( e
53
 j −  )

M
p k =1 ( e
H ( e j  ) = 0 e j ( N − M ) N k
d0  ( e j −  k )
k =1

( e j −  e j )

54
Ex：A system has a pole at z = 0, and a zeros at c = 0.9 e j/4 ,
as shown below. The right figure shows the amplitude spectrum
and phase spectrum.

55
（continued）

56
（continued）

57
（continued）

58
（continued）

59
（continued）

60
（continued）

61
Two decomposition methods
Cascade (factor decomposition)

X(z) H1(z) H2(z) Y ( z ) = H1 ( z ) H 2 ( z ) X ( z )

H (z) = H 1(z)H 2 (z)

Parallel (residue decompostion)
H1(z)
X(z) Y ( z ) = ( H 1 ( z ) + H 2 ( z )) X ( z )
H2(z)

H (z) = H 1(z) + H 2 (z)

1 2
H (z) = + + ...
1 +  11 z −1
1 +  12 z −1
62
Transfer function and stability

gain zeros
M

Y (z) p0 (z − )
l
H (z) = =z N −M l =1
X (z) N

(z − )
d0
l
l =1
poles
Pole：influence stability of a system
zero：adjust the characteristics of a system. Its effect
depends on its relative position to poles.
63
Causal stable system all poles in the unit circle

All poles in the unit circle → stable

There is a pole on the unit circle → undecided
There is a pole outside the unit circle → unstable

64
 N2
FIR： H (z) =  h n z −n
n = N1

 N2 
Y ( z ) =   h n z  X ( z )
−n

 n = N1 

All poles are on the origin

ROC：whole z-plane except the origin

When h[n] is bounded, FIR system is stable.

65
IIR：

 (1 −  z )
M M

Y (z ) p0
l
−1
 (z −  )
l
H (z ) = l =1 N −M p0 l =1
= =z
X (z ) d 0
 (1 −  z )
N d0 N

l =1
l
−1
 (z −  )
l =1
l

Could be unstable. Quantizing coefficients

could lead to unstability.

66
 Conclusion
• z-transform and its ROC

• Inverse z-transform using partial-fraction

expansion

• Transfer function, difference function, and

impulse response

• Stability of a system

67
 Homework

6.10（c）、6.13（c）、6.81

68