DESIGN OF DIGITAL FILTERS

residue() Vs residuez()

2
residuez() :

3
residue() :

4
Contents:
 Design of iir filters from analog filters
o IIR Filter Design by Approximation of Derivatives
o IIR Filter Design by Impulse Invariance
o IIR Filter Design by the Bilinear Transformation
o Characteristics of Commonly Used Analog Filters
o Some Examples of Digital Filter Designs Based on the
Bilinear Transformation

5
§10.3.2 IIR Filter Design by
Impulse Invariance

6
Impulse Invariance Summary [3]:

Recall Sampling Theorem

@ 𝑧 = 𝑒 𝑠𝑇 , LHS also 𝑠 = 𝑗Ω
becomes 𝑓(𝑠).

7
Impulse Invariance Summary [3]:

8
Impulse Invariance Summary [3]: 𝜔 = Ω𝑇 =
Ω
Fs

9
IIR Filter Design by Impulse Invariance :

10
IIR Filter Design by Impulse Invariance :

11
RECALL : 𝐹-axis: Periodic with 𝐹𝑝 = 𝐹𝑆
𝐹
If we  𝐹-axis by 𝐹𝑆 , = 𝑓-axis
𝐹𝑠
becomes periodic with 𝑓𝑝 = 1.

1
𝐹𝑠 = 𝐹𝑠 /2 > 𝐵
𝑇

1
𝐹-axis: Periodic with 𝐹𝑝 = 𝐹𝑆 =
𝑇
If we  𝐹-axis by 2𝜋, 2𝜋𝐹 = Ω-axis
𝐹𝑠 /2 < 𝐵 becomes periodic with Ω𝑝 = 2𝜋/𝑇.

12
RECALL :

Normalized Frequencies:
𝐹 Ω
𝑓= ; 𝜔=
𝐹𝑠 𝐹𝑠

13
IIR Filter Design by Impulse Invariance :

𝐹
𝑓=
𝐹𝑠

14
IIR Filter Design by Impulse Invariance : 𝜔 = 2𝜋𝑓
Ω = 2𝜋𝐹

15
IIR Filter Design by Impulse Invariance :
𝐹
𝜔 = 2𝜋𝑓 = 2𝜋
𝐹𝑠
Ω = 2𝜋𝐹
Ω 2𝜋𝐹
 Ω𝑇 = =
Fs 𝐹𝑠
= 2𝜋𝑓 = 𝜔

Ω-axis: Periodic with

Ω𝑝 = 2𝜋/𝑇
Ω
If we  Ω-axis by 𝐹𝑆 , =
𝐹𝑠
Ω𝑇 = 𝜔-axis becomes
periodic with 𝜔𝑝 = 2𝜋.
16
IIR Filter Design by Impulse Invariance :

See graphs on
previous slide

17
IIR Filter Design by Impulse Invariance :

Figure 10.3.3 depicts the

frequency response of a
lowpass analog filter and
the frequency response of
the corresponding digital
filter. Ω
= Ω𝑇 = 𝜔
Fs

18
IIR Filter Design by Impulse Invariance :
o It is clear that the digital filter with frequency response
𝑯(𝝎) has the frequency response characteristics of
the corresponding analog filter if the sampling interval
T is selected sufficiently small to completely avoid or
at least MINIMIZE THE EFFECTS OF ALIASING.
o It is also clear that the impulse invariance method is
INAPPROPRIATE FOR DESIGNING HIGHPASS FILTERS
due to the spectrum aliasing that results from the
sampling process.
19
IIR Filter Design by Impulse Invariance :

20
Relation between 𝒛 & 𝒔 variables
for Impulse Invariance Method

21
RECALL – Laplace T/F and z-T/F :

𝑠 = 𝜎 + 𝑗Ω with −∞ ≤ Ω ≤ ∞;
𝑗𝜔
𝑧 = 𝑟𝑒 with −𝜋 ≤ 𝜔 ≤ 𝜋
𝑠𝑡 𝜎𝑡 𝑗Ω𝑡
BFs of Laplace T/F : 𝑒 = 𝑒 𝑒
𝑛 𝑛 𝑗𝜔𝑛
BFs of z T/F : {𝑧 = 𝑟 𝑒 }
22
Mapping of points between s-plane and z-plane :
 Let ℎ𝑎 (𝑡) ↔ 𝐻𝑎 (𝑠) and ℎ𝑎 𝑛𝑇 = ℎ(𝑛) ↔ 𝐻 𝑧
 Now, consider the Periodic Sampling of ℎ𝑎 (𝑡).
 As ℎ𝑎 𝑡 ℎ𝑎 𝑛𝑇 = ℎ(𝑛); the corresponding
spectrum becomes Periodic Repetition of the
Scaled Spectrum 𝐹𝑆 × 𝐻𝑎 (𝑠).
 The resultant spectrum can be thought as a
transform whose BFs are DT version of BFs of
𝑠𝑛𝑇
Laplace T/F i.e., {𝑒 }. Recall: DT frequencies are 2𝜋 periodic.
23
Mapping of points between s-plane and z-plane :
 But we know that, for DT signal, the z-T/F is used whose BFs
are {𝑧 𝑛 }.
 This implies : 𝑧 𝑛 = 𝑒 𝑠𝑛𝑇 which further implies: 𝑧 = 𝑒 𝑠𝑇 .
 This establishes the following RELATION between the the
Laplace transform of ℎ𝑎 (𝑡) and the z-transform of
ℎ𝑎 𝑛𝑇 = ℎ(𝑛).
∞
1 2𝜋
𝐻 𝑧 ቚ 𝑠𝑇 = ෍ 𝐻𝑎 𝑠 − 𝑗 𝑘 … (10.3.22)
𝑧=𝑒 𝑇 𝑇
𝑘=−∞
∞
SPECIAL CASE: If we restrict ourself on imaginary 1 2𝜋𝑘
𝐻 Ω𝑇 = ෍ 𝐻𝑎 𝑗 Ω − … (10.3.21)
axis 𝑠 = 𝑗Ω, then (10.3.22) becomes (10.3.21). 𝑇
𝑘=−∞
𝑇 24
Mapping of points between s-plane and z-plane :

The MAPPING corresponding

to Impulse Invariance is:
𝑠𝑇
𝑧=𝑒 = 𝑓(𝑠; 𝑇)
where 𝑠 is input variable, 𝑧 is output variable, and 𝑇 is parameter.
25
Mapping of points between s-plane and z-plane :

𝐻 𝑒 𝑠𝑇 =

Both sides of (10.3.22) are 𝑓 𝑠; 𝑇 .

26
Mapping of points between s-plane and z-plane :

𝑧= 𝑒 𝑠𝑇

 𝜔 = Ω𝑇

27
Mapping of points between s-plane and z-plane :

𝑧=𝑒 𝑠𝑇
 𝜔 = Ω𝑇
clearvars; close all; clc;
LW = 'LineWidth'; FS = 'FontSize';
Om = [-100:0.01:100];
T = 1;
om = Om*T;
plot(Om,om,LW,2); grid;
xlabel('\Omega',FS,12);
ylabel('\omega',FS,12);
title('\omega = \Omega \times T',FS,18);

28
Mapping of points between s-plane and z-plane :

29
Mapping of points between s-plane and z-plane :

Moreover, 𝜎 = 0  𝑟 = 1  𝑗Ω axis is mapped

onto the unit circle.
30
Mapping of points between s-plane and z-plane :
𝜔 Ω
𝜔 = Ω𝑇  Ω = 𝜔 = Ω𝑇 =
𝑇 Fs
𝜋 𝜋
− ≤Ω≤ maps to −𝜋 ≤ 𝜔 ≤ 𝜋
𝑇 𝑇
𝜋 3𝜋
≤Ω≤ maps to 𝜋 ≤ 𝜔 ≤ 3𝜋 ≜ −𝜋 ≤ 𝜔 ≤ 𝜋
𝑇 𝑇

[Recall: 𝝎 is periodic with unique values within (−𝝅, 𝝅)]

Hence the mapping Ω → 𝜔 (analog freq  digital freq.) is

NOT UNIQUE i.e., this mapping is a Many-to-One Mapping. 31
Mapping of points between s-plane and z-plane :

32
Mapping of points between s-plane and z-plane :

33
 STEPS
for Impulse Invariance Method

34
IIR Filter Design by Impulse Invariance :

35
IIR Filter Design by Impulse Invariance :

−𝑎𝑡
1 ℒ
𝑥 𝑡 =𝑒 𝑢 𝑡 𝑋 𝑠 =
𝑠+𝑎
ROC: ℛ𝑒 𝑠 ≥ −𝑎

𝑛
𝑧 1
𝑥 𝑛 =𝑎 𝑢 𝑛 𝑋 𝑧 =
1 − 𝑎𝑧 −1
ROC: |𝑧| > |𝑎| 36
IIR Filter Design by Impulse Invariance :

37
IIR Filter Design by Impulse Invariance :

38
IIR Filter Design by Impulse Invariance :

39
IIR Filter Design by Impulse Invariance :

𝒉𝒂 𝒕 is NOT required in order to design the digital IIR filter based on the
method of impulse invariance. Rather 𝐻𝑎 (𝑠) can be directly converted to
𝐻(𝑧) using (10.3.31). [INDIRECT WAY: 𝐻𝑎 (𝑠) → ℎ𝑎 𝑡 → ℎ 𝑛𝑇 → 𝐻(𝑧)] 40
EX. 10.3.3

41
Ex. No. 10.3.3 :

42
Ex. No. 10.3.3 :

b = [1 0.1];
a = [1 0.2 9.01]
roots(b), roots(a)
freqs(b,a)

a= 1.0000 0.2000 9.0100

ans = -0.1000
ans =
-0.1000 + 3.0000i
-0.1000 - 3.0000i
43
Ex. No. 10.3.3 :

b = [1 0.1]; a = [1 0.2 9.01]; [r p c] = residue(b,a)

r= p=
c= []
0.5000 -0.1000 + 3.0000i
44
0.5000 -0.1000 - 3.0000i
Ex. No. 10.3.3 :

45
Ex. No. 10.3.3 :

46
Ex. No. 10.3.3 :

47
Ex. No. 10.3.3 :

HOMEWORK
 Compute ℎ𝑐 (𝑡) from 𝐻𝑐 (𝑠).
 Compute ℎ(𝑛) from 𝐻(𝑧).
 Verify that ℎ 𝑛 = ℎ𝑐 (𝑛𝑇)

𝑛
𝑧 1 −𝑎𝑡
ℒ 1
𝑥 𝑛 =𝑎 𝑢 𝑛 𝑋 𝑧 = 𝑥 𝑡 =𝑒 𝑢 𝑡 𝑋 𝑠 =
1 − 𝑎𝑧 −1 𝑠+𝑎
ROC: |𝑧| > |𝑎| ROC: ℛ𝑒 𝑠 ≥ −𝑎 48
Ex. No. 10.3.3 :

𝜔𝑟 = Ωr 𝑇 = 3 × 0.1 = 0.3 = 0.047(2𝜋)

𝜔𝑟 = Ωr 𝑇 = 3 × 0.5 = 1.5 = 0.238(2𝜋) Resonance at pole location i.e., at 𝛀𝐫 = 𝟑 49
Ex. No. 10.3.3 : 1
∞
2𝜋
𝐻 𝑧 ቚ = ෍ 𝐻𝑎 𝑠 − 𝑗 𝑘 … (10.3.22)
𝑧=𝑒 𝑠𝑇 𝑇 𝑇
𝑘=−∞

 (10.3.22) explains the effect of aliasing!

 Also check: Impulse Invariance from [9].
50
Ex. No. 10.3.3 :

51
Ex. No. 10.3.3 :

52
Ex. No. 10.3.3 :

The preceding example illustrates the

importance of selecting a small value for T to
minimize the effect of aliasing.
Due to the presence of aliasing, the impulse
invariance method is appropriate for the design
of lowpass and bandpass filters only.

53
impinvar()

54
IIR Filter Design by Impulse Invariance :

55
IIR Filter Design by Impulse Invariance : 𝑒 −𝑎𝑡 𝑢 𝑡
ℒ 1
𝑠+𝑎

56
IIR Filter Design by Impulse Invariance :
1
Matlab’s impinvar() function introduces a gain of to the
𝐹𝑠
digital filter. Hence we must multiply the analog impulse
response ℎ𝑎 𝑡 by this gain to get meaningful comparison.
Matlab uses
1
ℎ 𝑛 ≡ 𝑇ℎ𝑎 𝑛𝑇 = ℎ𝑎 𝑛𝑇
𝐹𝑠
in place of
ℎ 𝑛 ≡ ℎ𝑎 𝑛𝑇
in order to compensate for scaling factor
introduced by TD sampling. 57
§10.3.3 IIR Filter Design by
the Bilinear Transformation

58
DISCLAIMER
59
These power point slides are NOT
SUBSTITUTE of reading TEXT
BOOK(S).
You’re ALWAYS DIRECTED to
CAREFULLY READ the relevant
book chapter and SOLVE ALL
Examples and End Problems.
60
REFERENCES :
[1] [Proakis-2007] DSP 4th Ed

[2] [Proakis-2007] DSP 4th Ed (CH 11: Solved Examples and MATLAB Solutions)

[3 [Proakis-2004] A Self-Study Guide for DSP

[4 [Inge-2017] Digital Signal Processing Using Matlab - A Problem Solving Companion

(4th Ed)

[5] [Kronenburger-2008] Analog and Digital Signal Processing

[6] [Manolakis-2011] Applied Digital Signal Processing

[7] [Sierra-2017] Digital Signal Processing with Matlab Examples, Volume 1
[8] [Rabiner-1975] Theory and Application of Digital Signal Processing

[9] [Kim-2021] Conceptual Digital Signal Processing with MATLAB 61