Digital Signal Processing

ECE651 Digital Signal Processing I
Digital IIR Filter Design

 Introduction
 Some Preliminaries on Analog Filters
 Digital IIR Filter Design (s – z)
 Impulse Invariance Transformation
 Bilinear Transformation
 Frequency Band Transformations
 Analog Domain (s – s )
 Digital Domain (z – z)
Introduction
Analog filter : Infinitely long impulse response
S – Z (complex-valued mapping)
Digital IIR filter : Infinitely long impulse response

Introduction
Introduction
Advantages
• Analog filter design tables available
• Filter transformation (s – z) tables available
• Frequency band transformation (s – s / z – z) available
Disadvantages
• No control over the phase characteristics of the IIR filter
• Magnitude – only design
Introduction
Other Design Approaches

• Simultaneously approximate both the magnitude and the phase
response
• Require advanced optimization tools
• Not covered in the class
Preliminaries On Analog Filters
Analog lowpass filter specifications
 : passband ripple parameter

A: stopband attenuation parameter
 p: passband cutoff frequency (rad/sec)
 s: stopband cutoff frequency (rad/sec)
2 1
H a ( p ) 
1  2
1
H a ( s ) 
2
A2
Analog lowpass filter specifications
R p : passband ripple in dB
As : stopband attenuation in dB
R p  10 log 10 (1   2 )   10
R p / 10
1
As  20 log 10 A A  10 As / 20
Analog lowpass filter system function H a (s )
H a ( s ) H a (  s )  H a ()
2
 s / j
• Poles and zeros of magnitude-squared function are

distributed in a mirror-image symmetry with respect to the
imaginary axis
• For real filters, poles and zeros occur in complex

conjugate pairs ( mirror symmetry with respect to real axis)
Analog lowpass filter system function H a (s )
1. Pick up poles On LHP
2. Pick up zeros on LHP or

Imaginary axis
H a (s )
Stable Causal
Prototype analog filters

1. Butterworth
2. Chebyshev (Type I and II)
3. Elliptic
Butterworth lowpass filters (Magnitude-Squared Response)
1
H a () 
2
2N

1   
 c 
 c The Cutoff frequency (rand/sec)
N The order of the filter

Butterworth lowpass filters (System Function)
c
N
H a ( s) 
 ( s  pk )
LHP poles
 ( 2 k  N 1)
j
pk   c e 2N
, k  0,1,  2 N  1
Butterworth lowpass filters (Design equations)
 log 10 [(10 R p / 10  1) /(10 As / 10  1)] 

N  
 2 log 10 (  p /  s ) 
p
c  1
 1)
R p / 10 2N
(10
p
c  1
(10 As / 10  1) 2N
S - Z transformation
• Complex-valued mappings
• Derived by preserving different aspects of analog filters and digital filters

Impulse Invariance transformation
• Preserve the shape of impulse response
z  e sT
Impulse Invariance transformation (Design Procedure)

(MATLAB function: impinvar)
1. Choose T and determine the analog frequencies
p s
p  s 
T T
2. Design an analog filter H a (s ) using specifications  p ,  s , R p , and As
N
Rk
3. Partial fraction expansion H a ( s)  
k 1 s  pk
4. Transform analog poles { pk } into digital poles {e pk T } to obtain

N
Rk
H ( z)   pk T 1
k 1 1  e z
Impulse Invariance transformation (Aliasing)
>> f=0:0.01:5;T=0.1;
>> z=exp(j*2*pi*f*T);
>> zH=(1-0.8966./z)./(1-1.5595./z+0.6065./z./z);
>> s=j*2*pi*f;
>> sH=(1+s)./(s.^2+5*s+6);
>> plot(f,abs(zH),f,abs(sH)/T);legend('Digitital','Analog')
>>title('Magnitude Response of Analog and Digital IIR Filters')
Impulse Invariance transformation
Advantages:
• Stable design
• Analog frequency and digital frequency are linearly related
Disadvantage
• Aliasing
• Useful only when the analog filter is band-limited (LPF and
BPF)
Bilinear transformation
• Preserve the system function representation
1  ( sT ) / 2 2 1  z 1
z s
1  ( ST ) / 2 T 1  z 1
Bilinear transformation (Design Procedure)

(MATLAB function: bilinear)
1. Choose T (1)and determine the analog frequencies
2  2 
p  tan( p ) s  tan( s )
T 2 T 2
2. Design an analog filter H a (s ) using specifications  p ,  s , R p , and As
3. Bilinear transformation
2 1  z 1
H ( z)  H a ( )
T 1  z 1
Bilinear transformation
Advantages
• Stable design
• No aliasing
• No restriction on the type of filters that can be transformed

Frequency DomainTransformations
Analog Domain
Frequency DomainTransformations
Digital Domain

Digital Signal Processing

Uploaded by

Copyright:

Available Formats

Digital Signal Processing

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Digital Signal Processing

Uploaded by

Copyright:

Available Formats

ECE651 Digital Signal Processing I

Digital IIR Filter Design

Analog filter : Infinitely long impulse response

Digital IIR filter : Infinitely long impulse response

Other Design Approaches

Analog lowpass filter specifications

 : passband ripple parameter

Analog lowpass filter specifications

Analog lowpass filter system function H a (s )

• Poles and zeros of magnitude-squared function are

• For real filters, poles and zeros occur in complex

Analog lowpass filter system function H a (s )

1. Pick up poles On LHP

2. Pick up zeros on LHP or

Prototype analog filters

2. Chebyshev (Type I and II)

Butterworth lowpass filters (Magnitude-Squared Response)

 c The Cutoff frequency (rand/sec)

N The order of the filter

Butterworth lowpass filters (System Function)

Butterworth lowpass filters (Design equations)

 log 10 [(10 R p / 10  1) /(10 As / 10  1)] 

• Derived by preserving different aspects of analog filters and digital filters

Impulse Invariance transformation

• Preserve the shape of impulse response

Impulse Invariance transformation (Design Procedure)

2. Design an analog filter H a (s ) using specifications  p ,  s , R p , and As

4. Transform analog poles { pk } into digital poles {e pk T } to obtain

Impulse Invariance transformation (Aliasing)

Impulse Invariance transformation

• Preserve the system function representation

Bilinear transformation (Design Procedure)

2. Design an analog filter H a (s ) using specifications  p ,  s , R p , and As

• No restriction on the type of filters that can be transformed

You might also like