Chapter I Z-transform

I. Introduction
Z-transform is
→ Mathematical tool (the study of digital systems).
→ Allows the description of discrete-time signals.
→ Generalizes the Fourier transform and allows for overcoming these limitations.
→ The response of linear time-invariant systems..
→ Obtain the impulse response of an LTI system described by a finite difference
equation.
→ Comparable to the Laplace transform used in the case of continuous systems.
→ The convolution operator in the time domain corresponds to the multiplication
operator in the Z-transform domain.
I.1. Definition
The Z-transform : transform sequence 𝑥(𝑛) → 𝑋(𝑧)
𝑛 entire → 𝑧 complex variable
+∞

𝑋 𝑧 = ෍ 𝑥 𝑛 . 𝑧 −𝑛
𝑛=−∞

where 𝑧 is defined wherever 𝑋 𝑧 converge.

➔ 𝑥(𝑛) : time domain ➔ 𝑋 𝑧 ∶ frequency domain by analogy with Fourier transform
➔ 𝑛 : discret time → 𝑧 does not represent anything
I.2. Convergence « Cauchy Criterion »
Z-transform exists if the series 𝑋(𝑧) converges. To determine this, we will use the Cauchy
criterion:
+∞
1
lim 𝑢(𝑛) 𝑛 < 1 => ෍ 𝑢(𝑛) 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒
𝑛→∞
𝑛=0
+∞

𝑅𝑂𝐶 = 𝑧 ∈ 𝐶 / ෍ 𝑥 𝑛 . 𝑧 −𝑛 < +∞
𝑛=−∞

Note: Specifying the Region Of Convergence (ROC) of the transform is just as important
as the transform itself.
Let us express the complex variable 𝑧 in polar form as (𝑧 = 𝑟𝑒 𝑗𝜃 ).

Where 𝑟 = 𝑧 and 𝜃 = arg(𝑧). Then 𝑋(𝑧) can be expressed as

+∞

𝑋 𝑧 = ෍ 𝑥 𝑛 . 𝑧 −𝑛
𝑛=−∞

+∞

≤ ෍ 𝑥 𝑛 . 𝑟 −𝑛
𝑛=−∞

−1 +∞

≤ ෍ 𝑥 𝑛 . 𝑟 −𝑛 + ෍ 𝑥 𝑛 . 𝑟 −𝑛
𝑛=−∞ 𝑛=0
+∞ +∞
𝑥 𝑛
≤ ෍ 𝑥 −𝑛 . 𝑟𝑛 +෍
𝑟𝑛
𝑛=1 𝑛=0
 1
According to the Cauchy criterion: lim 𝑢(𝑛) 𝑛 <1 => σ+∞
𝑛=0 𝑢(𝑛) 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒
𝑛→∞

1 1
𝑥 𝑛 𝑥 𝑛 𝑛 𝑥(𝑛) 𝑛
σ+∞
𝑛=0 converges if lim < 1 ⇒ lim <1
𝑟𝑛 𝑛→+∞ 𝑟 𝑛 𝑛→∞ 𝑟

lim 𝑥(𝑛) 1Τ𝑛 <𝑟

𝑛→+∞

1Τ𝑛
Let’s suppose lim 𝑥(𝑛) = 𝑟2 .
𝑛→+∞

For any 𝑟 ≥ 𝑟2 the series converges

 1
Similarly : σ+∞
𝑛=1 𝑥 −𝑛 . 𝑟 converges if lim
𝑛
𝑥 −𝑛 . 𝑟 𝑛 𝑛
<1
𝑛→+∞

1Τ𝑛 1
lim 𝑥(−𝑛) <
𝑛→+∞ 𝑟

−1Τ𝑛 ; 1 1
Let’s suppose 𝑟1 = lim 𝑥(−𝑛) < ➔ 𝑟 < 𝑟1
𝑛→+∞ 𝑟1 𝑟
In general, the ROC (Region of Convergence) of a signal 𝑥 𝑛 corresponds to the subset
of 𝑪 such that :

0 ≤ 𝑟2 < 𝑧 < 𝑟1 ≤ ∞

Ring of Convergence
I.3. Calculation of some transforms

1 𝑛=0
Dirac impulse : 𝛿 𝑛 =ቊ
0 𝑛≠0
+∞ 0

𝐷 𝑧 = ෍ 𝛿 𝑛 . 𝑧 −𝑛 = ෍ 𝛿 0 . 𝑧 −𝑛 = 𝑧 0 = 1 ∀ 𝑧∈𝑪
𝑛=−∞ 𝑛=0

Region of Convergence is 𝑪.
1 𝑛=𝑘
Translated Dirac impulse : 𝑥 𝑛 =𝛿 𝑛−𝑘 =ቊ
0 𝑛≠𝑘
+∞

𝑋 𝑧 = ෍ 𝛿 𝑛 − 𝑘 . 𝑧 −𝑛 = 𝑧 −𝑘
𝑛=−∞
Si 𝑘 > 0, 𝑋(𝑧) ROC ∶ 𝑪 − 0 .
Si 𝑘 < 0, 𝑋(𝑧) ROC : 𝑪 − ∞ .
 1 𝑛≥0
Unit step signal : 𝑢 𝑛 =ቊ
0 𝑛<0
+∞ +∞

𝑈 𝑧 = ෍ 𝑢 𝑛 . 𝑧 −𝑛 = ෍ 𝑧 −𝑛
𝑛=−∞ 𝑛=0

This sum is that of a geometric series (𝑧 −1 )𝑛 which is finite if 𝑧 −1 < 1.

1 − (𝑧 −1 )𝑛 ȁ𝑛→+∞ 1
𝑈 𝑧 =1 −1
= si 𝑧 −1 < 1
1−𝑧 1 − 𝑧 −1
ROC : 𝑧 > 1
Exponential sequence : 𝑥 𝑛 = 𝑎𝑛 𝑢(𝑛)
+∞ +∞ +∞

𝑋 𝑧 = ෍ 𝑥 𝑛 . 𝑧 −𝑛 = ෍ 𝑎𝑛 . 𝑧 −𝑛 = ෍ (𝑎. 𝑧 −1 )𝑛
𝑛=−∞ 𝑛=0 𝑛=0

This is an infinite geometric series with ratio (𝑎. 𝑧 −1 ) which converges if 𝑎𝑧 −1 < 1.

1 − (𝑎𝑧 −1 )𝑛 ȁ𝑛→+∞ 1
𝑋 𝑧 =1 −1
= si 𝑎𝑧 −1 < 1
1 − 𝑎𝑧 1 − 𝑎𝑧 −1
ROC : 𝑧 > 𝑎
Note
For a finite-duration signal, the ROC is the complex plane 𝑪 with the possible exclusion of
𝑧 = 0 and/or 𝑧 = ∞ .
Exemples
Find the z-transform for each of the following sequences:
1. 𝑥1 𝑛 = 1 7 0 3 2 4
2. 𝑥2 𝑛 = 0 2 3 7 0 1 2 4
3. 𝑥3 𝑛 = 1 0 3 0 3 2 2 1
4. 𝑥4 𝑛 = 0 2 1 2 5 0 1 2 1
5. 𝑥5 𝑛 = 𝛿 𝑛 + 𝑘 𝑘>0
6. 𝑥6 𝑛 = 10 𝑢 𝑛
7. 𝑥7 𝑛 = 𝑏 𝑛 𝑢(−𝑛 − 1)
8. 𝑥8 𝑛 = (3(2)𝑛 +2(3)𝑛 )𝑢(𝑛)
9. 𝑥9 𝑛 = 10 sin(0.25𝜋𝑛)𝑢(𝑛)
10. 𝑥10 𝑛 = (0.5)𝑛 sin(0.25𝜋𝑛)𝑢(𝑛)
11. 𝑥11 𝑛 = 𝑒 −0.1𝑛 cos(0.25𝜋𝑛)𝑢(𝑛)
I.4. Properties of the Z-transform
𝑇𝑍 𝑥 𝑛 =𝑋 𝑧 𝑇𝑍 𝑦 𝑛 =𝑌 𝑧 .

Linearity : 𝑇𝑍 𝑎1 𝑥 𝑛 + 𝑎2 𝑦 𝑛 = 𝑎1 𝑋 𝑧 + 𝑎2 𝑌 𝑧 , ∀ 𝑎1 , 𝑎2 ∈ 𝑅2

ROC intersection of the two ROC

Shift Theorem : 𝑇𝑍 𝑥 𝑛 − 𝑘 = 𝑧 −𝑘 𝑋 𝑧 ROC is the same as 𝑥 𝑛

Advance Theorem : 𝑇𝑍 𝑥 𝑛 + 𝑘 = 𝑧 𝑘 𝑋 𝑧 − σ𝑘−1

𝑛=0 𝑥 𝑛 . 𝑧
𝑘−𝑛 ROC same as 𝑥 𝑛

𝑑𝑋 𝑧
Multiplication by 𝒏 : 𝑇𝑍 𝑛𝑥 𝑛 = −𝑧 ROC is the same as 𝑥 𝑛
𝑑𝑧

𝑧
Multiplication by 𝒂 :
𝒏 𝑛
𝑇𝑍 𝑎 𝑥 𝑛 =𝑋 ROC : 𝑎𝑟1 < 𝑧 < 𝑎𝑟2
𝑎

Time Reversal : 𝑇𝑍 𝑥 −𝑛 = 𝑋 𝑧 −1 ROC : 1Τ𝑟2 < 𝑧 < 1Τ𝑟1

Linear Convolution :

𝑇𝑍 𝑥 𝑛 ∗ 𝑦 𝑛 = 𝑋 𝑧 .𝑌 𝑧 ROC : 𝑚𝑎𝑥 𝑟1𝑥 , 𝑟1𝑦 < 𝑧 < 𝑚𝑖𝑛 𝑟2𝑥 , 𝑟𝑦2

The Initial Value Theorem : if 𝑥 𝑛 is causal then :

𝑥 0 = lim 𝑋(𝑧)
𝑧→+∞

The Final Value Theorem : if 𝑥 𝑛 is causal then :

lim 𝑥(𝑛) = lim(1 − 𝑧 −1 )𝑋(𝑧)
𝑛→+∞ 𝑧→1
 Signal Z-Transform ROC
1 𝛿(𝑛 ) 1 C
1
2 𝑢(𝑛) 𝑧 >1
1−𝑧 −1
1
3 -𝑢(−𝑛 − 1) 𝑧 <1
1−𝑧 −1
1
4 𝑎𝑛 𝑢(𝑛) 𝑧 > 𝑎
1−𝑎𝑧 −1
𝑎𝑧 −1
5 𝑛𝑎𝑛 𝑢(𝑛) 𝑧 > 𝑎
(1−𝑎𝑧 −1 )2

1
6 −𝑎𝑛 𝑢(−𝑛 − 1) 𝑧 < 𝑎
1−𝑎𝑧 −1
𝑎𝑧 −1
7 −𝑛𝑎𝑛 𝑢(−𝑛 − 1) 𝑧 < 𝑎
(1−𝑎𝑧 −1 )2
1−𝑧 −1 cos 𝜔𝑜
8 cos(𝜔𝑜 𝑛) 𝑢(𝑛) 𝑧 >1
1−2 cos 𝜔𝑜 𝑧 −1 +𝑧 −2
𝑧 −1 sin 𝜔𝑜
9 sin(𝜔𝑜 𝑛) 𝑢(𝑛) 𝑧 >1
1−2 cos 𝜔𝑜 𝑧 −1 +𝑧 −2
𝑛 1−𝑎𝑧 −1 cos 𝜔𝑜
10 𝑎 cos(𝜔𝑜 𝑛) 𝑢(𝑛) 𝑧 > 𝑎
1−2𝑎 cos 𝜔𝑜 𝑧 −1 +𝑎2 𝑧 −2
𝑎𝑧 −1 sin 𝜔𝑜
11 𝑎𝑛 sin(𝜔𝑜 𝑛) 𝑢(𝑛) 𝑧 > 𝑎
1−2𝑎 cos 𝜔𝑜 𝑧 −1 +𝑎2 𝑧 −2
I.5. Rational Z-transforms

➔ A rational Z-transform represents the set of Z functions written as the ratio of two
polynomials in 𝑧 −1 or (𝑧).

➔ This class of functions corresponds to linear time-invariant systems described by a

finite difference equation.

𝑥 𝑛 and 𝑦 𝑛 are the input and output of an LTI system (Linear Time-Invariant System).
𝑀 𝑁 𝑀 𝑁
Z−transform
෍ 𝑎𝑖 . 𝑦 𝑛 − 𝑖 = ෍ 𝑏𝑖 . 𝑥(𝑛 − 𝑖) ෍ 𝑎𝑖 . 𝑧 −𝑖 𝑌 𝑧 = ෍ 𝑏𝑖 . 𝑧 −𝑖 𝑋 𝑧
𝑖=0 𝑖=0 𝑖=0 𝑖=0

𝑌 𝑧 σ𝑁
𝑖=0 𝑏𝑖 . 𝑧
−𝑖
𝐻 𝑧 = = 𝑀
𝑋(𝑧) σ𝑖=0 𝑎𝑖 . 𝑧 −𝑖
Poles and Zeros
LTI system is characterized by
➔ impulse response : ℎ(𝑛)
➔ Z-transform : 𝐻(𝑧), (transfer function of the system)
𝑁 𝑏1 𝑁−1 𝑏𝑁
σ𝑁 𝑏 . 𝑧 −𝑖 𝑏 + 𝑏 𝑧 −1 + … + 𝑏 𝑧 −𝑁 𝑏 𝑧 −𝑁 𝑧 + 𝑧 …+
𝑖=0 𝑖 0 1 𝑁 0 𝑏0 𝑏0
𝐻 𝑧 = 𝑀 = −1 −𝑀
= 𝑎
σ𝑖=0 𝑎𝑖 . 𝑧 −𝑖 𝑎0 + 𝑎1 𝑧 + … + 𝑎𝑀 𝑧 𝑎0 𝑧 −𝑀 𝑧𝑀 + … + 𝑀
𝑎0
𝑏0 𝑀−𝑁 𝑁(𝑧)
= 𝑧
𝑎0 𝐷(𝑧)

𝑁(𝑧) has N simple or multiple roots : 𝑧1 … 𝑧𝑁 . (zeros)

𝐷(𝑧) has M simple or multiple roots : 𝑝1 … 𝑝𝑀 . (poles)

𝑁
𝑏0 𝑧 −𝑁 𝑧 − 𝑧1 … (𝑧 − 𝑧𝑁 ) 𝑀−𝑁
ς𝑖=1(𝑧 − 𝑧𝑖 )
𝐻 𝑧 = = 𝐾𝑧
𝑎0 𝑧 −𝑀 𝑧 − 𝑝1 … (𝑧 − 𝑝𝑀 ) ς𝑀
𝑖=1(𝑧 − 𝑝𝑖 )
 𝑁
𝑏0 𝑧 −𝑁 𝑧 − 𝑧1 … (𝑧 − 𝑧𝑁 ) 𝑀−𝑁
ς𝑖=1(𝑧 − 𝑧𝑖 )
𝐻 𝑧 = = 𝐾𝑧
𝑎0 𝑧 −𝑀 𝑧 − 𝑝1 … (𝑧 − 𝑝𝑀 ) ς𝑀
𝑖=1(𝑧 − 𝑝𝑖 )

if 𝑀 > 𝑁, 𝐻(𝑧) has 𝑀 − 𝑁 zeros at z = 0

Si 𝑀 < 𝑁, 𝐻(𝑧) has 𝑁 − 𝑀 poles at z = 0

The zeros of 𝐻(𝑧) are the values of 𝑧 for which 𝐻 𝑧 = 0.

The poles of 𝐻(𝑧) are the values of 𝑧 for which 𝐻 𝑧 = ∞.

𝐻(𝑧) is completely determined by the position of its poles and zeros, as well as by the
amplitude factor 𝐾 .
Remarks
➔ The poles and zeros indicate the behavior of the system, while the factor 𝐾 only
affects the amplitude.

➔ The position of the poles and zeros of a system (plus the amplitude factor 𝐾=𝑏₀/𝑎₀)
provides a complete description of 𝐻(𝑧) (ℎ(𝑛) and 𝐻(𝑓)), and thus of the system's
behavior

➔ 𝐻 𝑧 represented as a circle modeling (indicating) the position of the poles and

zeros in the complex plane.

➔In most systems, the 𝑎𝑖 and 𝑏𝑖 are real, while the poles and zeros are either real or
pairs of complex conjugates.

Example
4𝑧−3 (𝑧−1)2
𝐻 𝑧 = 𝐻 𝑧 =
(𝑧−1)(𝑧+0,7) (𝑧−0.5)(𝑧+1)2
All-Zero System

𝑀 = 0 → 𝑎1 = 𝑎2 = ⋯ = 𝑎𝑀 = 0.

transfer function
𝑁 𝑁
1
𝐻 𝑧 = ෍ 𝑏𝑖 . 𝑧 −𝑖 = 𝑁 ෍ 𝑏𝑖 . 𝑧 𝑁−𝑖
𝑧
𝑖=0 𝑖=0

The numerator polynomial has 𝑁 roots.

The impulse response of this system is finite type (FIR).
Example1

ℎ 𝑛 = 𝛿 𝑛 + 2𝛿 𝑛 − 1 + 𝛿 𝑛 − 3
+∞ +∞

=> 𝐻 𝑧 = ෍ ℎ 𝑛 . 𝑧 −𝑛 = ෍ (𝛿 𝑛 + 2𝛿 𝑛 − 1 + 𝛿 𝑛 − 3 ). 𝑧 −𝑛
𝑛=−∞ 𝑛=−∞

−1 −3 1 3
𝐻 𝑧 = 1 + 2𝑧 +𝑧 = (𝑧 + 2𝑧 2 + 1).
𝑧3

Representation of poles and zeros

All-Pole System

𝑁 = 0 → 𝑏1 = 𝑏2 = ⋯ = 𝑏𝑁 = 0.

transfer function :

𝑏0 𝑏0 𝑧 𝑀
𝐻 𝑧 = = 𝑀
1 + σ𝑀
𝑖=1 𝑎𝑖 . 𝑧 −𝑖 σ𝑖=0 𝑎𝑖 . 𝑧 𝑀−𝑖

The denominator polynomial has 𝑀 roots.

The impulse response of this system is infinite type (IIR).
Example2

ℎ 𝑛 = 0.2𝑛 𝑛𝑢 𝑛 − 1 => 𝐻 𝑧 = σ+∞

𝑛=−∞ ℎ 𝑛 . 𝑧 −𝑛

𝑎𝑧 −1
We use the property TZ 𝑛𝑎𝑛 𝑢(𝑛) =
(1−𝑎𝑧 −1 )2

0.2𝑧 2
=> 𝐻 𝑧 =
(𝑧−0.2)2

Representation of poles and zeros of 𝐻 𝑧

I.6. Causality Condition of H(z)

LTI system: causal if and the ROC (gray part) of its Z-transform is the outside of a circle of
radius 𝑟 < ∞ including 𝑧 = ∞.

ROC : 𝑟1 < ȁ𝑧ȁ < 𝑟2 ROC : ȁ𝑧ȁ < 𝑟 ROC : ȁ𝑧ȁ > 𝑟
Mixted System Noncausal System Causal System
I.7. Stability Condition of H(z)

For a linear time-invariant system to be stable, it is necessary and sufficient for its
impulse response to satisfy σ+∞
𝑛=−∞ ℎ 𝑛 < ∞ .

Since 𝐻 𝑧 = σ+∞𝑛=−∞ ℎ 𝑛 . 𝑧 −𝑛 it is therefore sufficient that 𝑧 = 1 is part of the

region of convergence (ROC)

LTI system : stable ➔ If the ROC of its Z-transform includes the unit circle.

➔ The poles of its transfer function are inside the unit circle.
II. Analysis of LTI Systems using the Z-transform
Study the behavior of linear time-invariant systems by analyzing their transfer function
described by a Z-function.
The transfer function (the position of poles and zeros) allows the determination of :

➔ The impulse response.

➔ The frequency response.
➔ Specify the stability conditions.
II.1. Determination of the impulse response

To a simple or multiple pole 𝑝𝑖 ➔ The impulse response is convergent if 𝑝𝑖 < 1.

➔ The response is divergent if 𝑝𝑖 > 1.

For each complex pole (conjugate pole)➔ ℎ 𝑛 Oscillatory (cosine or sine)

Damped if 𝑝𝑖=1,2 < 1, Divergent if 𝑝𝑖=1,2 > 1.

In a minimum-phase system ➔ All the zeros and poles are inside the unit circle ( 𝑧𝑖 <
1 𝑎𝑛𝑑 𝑝𝑖 < 1, ∀𝑖).
Determination of the impulse response from the plot of poles and
zeros
𝑛
𝑧
ℎ 𝑛 = 𝑎 𝑢 𝑛 => 𝐻 𝑧 =
(𝑧 − 𝑎)
If 0 < 𝑎 < 1
Determination of the impulse response from the plot of poles and zeros
𝑛
𝑧
ℎ 𝑛 = 𝑎 𝑢 𝑛 => 𝐻 𝑧 =
(𝑧 − 𝑎)
If 𝑎 = 1
Determination of the impulse response from the plot of poles and zeros
𝑛
𝑧
ℎ 𝑛 = 𝑎 𝑢 𝑛 => 𝐻 𝑧 =
(𝑧 − 𝑎)
If 𝑎 > 1
Determination of the impulse response from the plot of poles and zeros
𝑛
𝑧
ℎ 𝑛 = 𝑎 𝑢 𝑛 => 𝐻 𝑧 =
(𝑧 − 𝑎)
If −1 < 𝑎 < 0
Determination of the impulse response from the plot of poles and zeros
𝑛
𝑧
ℎ 𝑛 = 𝑎 𝑢 𝑛 => 𝐻 𝑧 =
(𝑧 − 𝑎)
If 𝑎 = −1
Determination of the impulse response from the plot of poles and zeros
𝑛
𝑧
ℎ 𝑛 = 𝑎 𝑢 𝑛 => 𝐻 𝑧 =
(𝑧 − 𝑎)
If 𝑎 < −1
Determination of the impulse response from the plot of poles and zeros
ℎ 𝑛 = 𝑎𝑛 cos(2𝜋𝑓0 𝑛)𝑢(𝑛)
𝑧 2 − 𝑎𝑧 cos 𝜔𝑜
𝑢 𝑛 => 𝐻 𝑧 =
(𝑧 − 𝑎𝑒 𝑗2𝜋𝑓0 )(𝑧 − 𝑎𝑒 −𝑗2𝜋𝑓0 )
If 𝑎 = 1
Determination of the impulse response from the plot of poles and zeros
ℎ 𝑛 = 𝑎𝑛 cos(2𝜋𝑓0 𝑛)𝑢(𝑛)
𝑧 2 − 𝑎𝑧 cos 𝜔𝑜
𝑢 𝑛 => 𝐻 𝑧 =
(𝑧 − 𝑎𝑒 𝑗2𝜋𝑓0 )(𝑧 − 𝑎𝑒 −𝑗2𝜋𝑓0 )
If 0 < 𝑎 < 1
Determination of the impulse response from the plot of poles and zeros
ℎ 𝑛 = 𝑎𝑛 cos(2𝜋𝑓0 𝑛)𝑢(𝑛)
𝑧 2 − 𝑎𝑧 cos 𝜔𝑜
𝑢 𝑛 => 𝐻 𝑧 =
(𝑧 − 𝑎𝑒 𝑗2𝜋𝑓0 )(𝑧 − 𝑎𝑒 −𝑗2𝜋𝑓0 )
If 𝑎 < −1
II.2. Determination of the frequency response
If 𝑧 = 1 ∈ 𝑅𝑂𝐶 of 𝐻 𝑧 . The calculation of 𝐻 𝑧 is limited to the unit circle :
𝑧 = 𝑒 𝑗2𝜋𝑓𝑇𝑒 => if 𝑧 = 1 => 𝑗2𝜋𝑓𝑇𝑒 = 0 => 𝑓 = 0

1 𝑓𝑒
if 𝑧 = −1 => 𝑗2𝜋𝑓𝑇𝑒 = 𝑗𝜋 => 𝑓 = =
2𝑇𝑒 2

➔ A zero on the unit circle introduces a cancellation of ȁ𝐻 𝑓 ȁ (for the corresponding

frequency).

➔ A zero close to (in the vicinity of) the unit circle introduces an attenuation ofȁ𝐻 𝑓 ȁ.

➔ The closer the zero is to the unit circle, the greater the attenuation..
➔ A pole on the unit circle introduces infinite resonance at 𝐻 𝑓 .

➔ A pole near the unit circle introduces a resonance that is more significant the closer
the pole is to the unit circle, in terms of ȁ𝐻 𝑓 ȁ.

➔ This provides an idea of the general behavior of a system, such as its cutoff
frequencies and its nature.

➔ A zero or a pole at the origin does not affect ȁ𝐻 𝑓 ȁ

➔ The closer the poles or zeros are to the unit circle, the more they influence the
frequency response.
Examples :
1. Let’s 𝑝1 = 0 and 𝑧1 = 0.8. To obtain the shape of 𝐻 𝑓 , At each frequency (point of
contact on the circle between blue and red), we divide the numerator vector (in
blue) by the denominator vector (in red).

𝒛 − 𝟎. 𝟖𝟖
𝑯 𝒛 =
𝒛
1. 𝑝1 = 0.7 and 𝑧1 = −0.6. To obtain the shape of 𝐻 𝑓 , At each frequency (point of
contact on the circle between blue and red), we divide the numerator vector (in blue)
by the denominator vector (in red).

𝑧 + 0.6
𝑯 𝒛 =
𝑧 − 0.7
 𝑧 2 −1
3. 𝐻 𝑧 = 3
, We have 2 complex conjugate poles, 𝑝𝑖=1,2 = 0.433 ± 𝑗0.8201,
𝑧 2 − 2 𝑧+0.86
and 2 real zeros, 𝑧𝑖=1,2 = ±1
 1−𝑧 −2
4. 𝐻 𝑧 = , 2 complex conjugate poles, 𝑝𝑖=1,2 = ±𝑗0.9306 and 2 real zeros,
1+ 0.871𝑧 −2
𝑧𝑖=1,2 = ±1
 1−0.94𝑧 −1 +𝑧 −2
5. 𝐻 𝑧 = ,2 complex conjugate poles,
0.86− 0.871𝑧 −1 +𝑧 −2
𝑝𝑖=1,2 = 0.433 ± 𝑗0.8201 2 complex conjugate zeros, 𝑧𝑖=1,2 = 0.47 ± 𝑗0.8827
One pole and one zero (First-order system)
Examples : 1. FIR system
𝑦 𝑛 = 𝛼𝑥 𝑛 − 1 + 𝑥 𝑛 => 𝑌 𝑧 = 𝛼𝑧 −1 𝑋 𝑧 + 𝑋(𝑧)
𝑥 𝑛
𝑧 −1

=> 𝐻 𝑧 = 1 + 𝛼𝑧 −1 => 𝐻 𝑓 = 1 + 𝛼𝑒 −𝑗2𝜋𝑓

+ 𝛼

𝑦 𝑛
2. IIR system
𝑦 𝑛 = 𝛼𝑦 𝑛 − 1 + 𝑥 𝑛
𝑧 1 1
𝐻 𝑧 = = −1
=> 𝐻 𝑓 =
𝑧−𝛼 1 − 𝛼𝑧 1 − 𝛼𝑒 −𝑗2𝜋𝑓
𝑥 𝑛 𝑦 𝑛
+

𝛼 𝑧 −1
A pair of zeros or poles (second order system)
Examples : 1. FIR system
𝑦 𝑛 = 𝑏0 𝑥 𝑛 + 𝑏1 𝑥 𝑛 − 1 + 𝑏2 𝑥 𝑛 − 2 => 𝑌 𝑧 = (𝑏0 + 𝑏1 𝑧 −1 + 𝑏2 𝑧 −2 )𝑋(𝑧)
𝐻 𝑧 = 1 − 𝑧1 𝑧 −1 1 − 𝑧2 𝑧 −1 = 1 − 𝑟𝑒 𝑗𝜑 𝑧 −1 1 − 𝑟𝑒 −𝑗𝜑 𝑧 −1

𝑥 𝑛
𝑧 −1 𝑧 −1

𝑏0 𝑏1 𝑏3

+
𝑦 𝑛
 𝐻 𝑧 = 1 − 𝑟𝑒 𝑗𝜑 𝑧 −1 1 − 𝑟𝑒 −𝑗𝜑 𝑧 −1 = 1 − 2𝑟 cos 𝜑 𝑧 −1 +𝑟 2 𝑧 −2

Frequency response for a pair of complex conjugate zeros of a second-order FIR

system, depending on the values of φ and the radius r
A pair of zeros or poles (second order system)
Examples : 1. IIR system
𝑦 𝑛 = 𝑏0 𝑥 𝑛 − 𝑎1 𝑦 𝑛 − 1 − 𝑎2 𝑦 𝑛 − 2
=> 𝑌 𝑧 = 𝑏0 𝑋 𝑧 − 𝑎1 𝑧 −1 𝑌 𝑧 − 𝑎2 𝑧 −2 𝑌(𝑧)
(1 + 𝑎1 𝑧 −1 + 𝑎2 𝑧 −2 )𝑌(𝑧) = 𝑏0 𝑋 𝑧
𝑏0 𝑏0
=> 𝐻 𝑧 = =
𝑗𝜑
1 − 𝑟𝑒 𝑧 −1 1 − 𝑟𝑒 −𝑗𝜑 𝑧 −1 1 − 2𝑟 cos 𝜑 𝑧 −1 +𝑟 2 𝑧 −2

𝑥 𝑛 𝑦 𝑛
𝑏0 +

𝑎1 𝑧 −1

𝑎2 𝑧 −1
A pair of zeros or poles (second order system)

𝑏0 𝑏0
𝐻 𝑧 = =
𝑗𝜑
1 − 𝑟𝑒 𝑧 −1 1 − 𝑟𝑒 −𝑗𝜑 𝑧 −1 1 − 2𝑟 cos 𝜑 𝑧 −1 +𝑟 2 𝑧 −2

Frequency response for a pair of complex conjugate poles of a second-order IIR

system, depending on the values of φ and the radius 𝑟
III. inverse Z-Transform

➢ General techniques for inverse transformation :

➔ Residue method (Integration over a closed contour).

➔ Power series expansion of 𝑧 or/and 𝑧 −1
➔ Partial fraction expansion and look-up table.
III.1 Residue method (Integration over a closed contour)

The relationship determining 𝑥(𝑛) from 𝑋(𝑧) by applying Cauchy's integral :

1
𝑥 𝑛 = ර 𝑋 𝑧 . 𝑧 𝑛−1 . 𝑑𝑧
2𝜋𝑗
C
Cauchy's theorem specifies that :

𝑥 𝑛 = ෍ résidus of [𝑋 𝑧 𝑧 𝑛−1 ]z=𝑝𝑖

𝑝𝑖 𝑝ô𝑙𝑒𝑠 𝑑𝑒 𝑋 𝑧

The residue at a pole 𝑧 = 𝑝𝑖 ​ of order 𝑞 of the function 𝑋 𝑧 𝑧 𝑛−1 is the quantity

1 𝑑 𝑞−1
Res[𝑋 𝑧 𝑧 𝑛−1 ]𝑧=𝑝𝑖 = 𝑧 − 𝑝𝑖
𝑞 𝑋 𝑧 𝑧 𝑛−1
𝑧=𝑝𝑖
𝑞 − 1 ! 𝑑𝑧 𝑞−1
For a simple pole : Res[𝑋 𝑧 𝑧 𝑛−1 ]𝑧=𝑝𝑖 = 𝑧 − 𝑝𝑖 𝑋 𝑧 𝑧 𝑛−1 𝑧=𝑝𝑖
Examples
𝑧
1. Let’s the z-transform : 𝑋 𝑧 = .
𝑧−1

One simple pole 𝑝1 = 1 and the residu will :

𝑧 − 1 𝑧 𝑛−1
Res1 = lim (𝑧 − 𝑝1 ) 𝑋(𝑧) 𝑧 𝑛−1 = lim 𝑧 =1
z→𝑝1 =1 z→1 (𝑧 − 1)

If the ROC is outside of a circle ( 𝑧 > 1) => the time signal 𝑥(𝑛) is causal, thus :
𝑥(𝑛) = 1. 𝑢 𝑛

If the ROC is inside a circle ( 𝑧 < 1) => the time signal 𝑥(𝑛) is noncausal, thus :
𝑥 𝑛 = −𝑢 −𝑛 − 1
 𝑧
2. Let’s the z-transform : 𝑋 𝑧 = avec 𝑧 > 𝑒 𝑎 .
𝑧−𝑒 𝑎

One simple pole 𝑝1 = 𝑒 𝑎 and the residu will :

𝑧 − 𝑒 𝑎 𝑧 𝑛−1
Res[𝑋 𝑧 𝑧 𝑛−1 ]𝑧=𝑒 𝑎 = lim 𝑎 (𝑧 − 𝑝1 ) 𝑋(𝑧) 𝑧 𝑛−1 = lim𝑎 𝑧 = 𝑧 𝑛
𝑧=𝑒 𝑎 = 𝑒 𝑎𝑛
z→𝑝1 =𝑒 z→𝑒 (𝑧 − 𝑒 𝑎 )

Since the ROC is outside of a circle ( 𝑧 > 𝑒 𝑎 ) => the time signal 𝑥(𝑛) is causal, thus :
𝑥 𝑛 = 𝑒 𝑎𝑛 . 𝑢 𝑛
III.2 Power series expansion of 𝒛 or/and 𝒛−𝟏

Write 𝑋 𝑧 as a power series in 𝑧 −1 (𝑧)

➔ Take the coefficients of the series for the time-domain signal

+∞

𝑋 𝑧 = ෍ 𝑐𝑛 𝑧 −𝑛
𝑛=−∞

𝑥 𝑛 = 𝑐𝑛
Example
1
𝑋(𝑧) =
1 − 𝑧 −3
Using the limit of geometric series :
+∞
1 −3 𝑛
= ෍ 𝑧
1 − 𝑧 −3
𝑛=0

+∞

𝑋(𝑧) = ෍ 𝑧 −3𝑛 = 1 + 𝑧 −3 + 𝑧 −6 + ⋯
𝑛=0

+∞

𝑥 𝑛 = ෍ 𝛿(𝑛 − 3𝑖)
𝑖=0
If the complex function 𝑋(𝑧) is in the form of a ratio of polynomials, we can perform an
approximate expansion by carrying out the division. This division can be done either
using negative or positive powers of 𝑧.

➢If the ROC is 𝑧 > 𝑎, The division must be done using negative powers of 𝑧.

➢If the ROC is 𝑧 < 𝑎, The division must be done using positive powers of 𝑧.

➢If the ROC is 𝑎1 < 𝑧 < 𝑎2 , The method cannot be used directly.
Examples
1
1. Soit 𝑋(𝑧) = pour 𝑧 > 𝑎
1−𝑎𝑧 −1

ROC Outside a circle => causal signal => division to obtain a series in 𝑧 −1 .
1 1 − 𝑎. 𝑧 −1
−1 + 𝑎. 𝑧 −1 1 + 𝑎. 𝑧 −1 + 𝑎2 . 𝑧 −2 + ⋯
0 + 𝑎. 𝑧 −1
−𝑎. 𝑧 −1 + 𝑎2 . 𝑧 −2
0 + 𝑎2 . 𝑧 −2
𝑥 0 𝑥 1 𝑥 2

1
𝑋 𝑧 = = 1 + 𝑎. 𝑧 −1 + 𝑎2 . 𝑧 −2 + ⋯ = σ+∞
𝑛=0 𝑎 𝑛 𝑧 −𝑛 => 𝑥 𝑛 = 𝑎 𝑛 . 𝑢(𝑛)
1−𝑎𝑧 −1
 1
2. Soit 𝑋(𝑧) = pour 𝑧 < 𝑎
1−𝑎𝑧 −1

ROC Inside a circle => Anti-causal signal=> Division to obtain a series in 𝑧.

𝑧 −𝑎 + 𝑧

−𝑧 + 𝑎 −1 . 𝑧 2 −𝑎−1 𝑧 − 𝑎−2 . 𝑧 2 − 𝑎−3 . 𝑧 3 + ⋯

0 + 𝑎 −1 . 𝑧 2

−𝑎 −1 . 𝑧 2 + 𝑎−2 . 𝑧 3

0 + 𝑎−2 . 𝑧 3 𝑥 −1 𝑥 −2 𝑥 −3

1
𝑋 𝑧 = = −𝑎−1 𝑧 − 𝑎−2 . 𝑧 2 − 𝑎−3 . 𝑧 3 + ⋯ => 𝑥 𝑛 = −𝑎𝑛 . 𝑢(−𝑛 − 1)
1−𝑎𝑧 −1
III.3. Partial fraction expansion

Expand 𝑋 𝑧 complex into a set of simple Z-functions (known inverse transform).

Apply the principle of linearity of the transform

𝑋 𝑧 = 𝛼1 𝑋1 𝑧 + 𝛼2 𝑋2 𝑧 + ⋯ + 𝛼𝐿 𝑋𝐿
𝑇𝑍 −1
𝑥 𝑛 = 𝛼1 𝑥1 𝑛 + 𝛼2 𝑥2 𝑛 + ⋯ + 𝛼𝐿 𝑥𝐿 (𝑛)
Example
1
𝑋(𝑧) =
1 − 3𝑧 −1 + 2𝑧 −2
This function has two simple poles, 𝑝1 = 1 and 𝑝2 = 2.
1 1 2 1
=> 𝑋 𝑧 = = −1 −1
= −1
−
𝑧−1 𝑧−2 𝑧 − 1 (2𝑧 − 1) 1 − 2𝑧 1 − 𝑧 −1

According to the table of Z-transform, there will be three cases depending on ROC:

If ROC 𝑧 > 2 => 𝑥 𝑛 = (2𝑛+1 + 1). 𝑢(𝑛).

If ROC 𝑧 < 1 => 𝑥 𝑛 = −(2𝑛+1 − 1). 𝑢(−𝑛 − 1).

If ROC 1 < 𝑧 < 2 => 𝑥 𝑛 = −2𝑛+1 𝑢(−𝑛 − 1) − 𝑢(𝑛).

 𝑃(𝑥)
➔ We will start by trying to determine the partial fraction decomposition of, where
𝑄(𝑥)
both 𝑃(𝑥) and Q(𝑥) are polynomials and the degree of 𝑃(𝑥) is smaller than the degree
of Q(𝑥). Partial fractions can only be done if the degree of 𝑃(𝑥) is strictly less than the
degree of Q(𝑥). That is important to remember.
➔So, once we’ve determined that partial fractions can be done we factor the denominator
as completely as possible. Then for each factor in the denominator we can use the
following table to determine the term(s) we pick up in the partial fraction decomposition.

For examples of factorization, please refer to

the following links
https://www.purplemath.com/modules/partfrac3.htm
http://tutorial.math.lamar.edu/Classes/Alg/PartialFractions.aspx