Unit-4

IIR DIGITAL FILTER DESIGN TECHNIQUES

Introduction
Infinite impulse response (IIR) digital filters: The present output sample depends on
the present input, past input and past output samples.
[Link] Analog Filters Digital Filters
1. Analog Filters are consented from 1. Digital Filter Consists of elements
active and passive element R, L, C. like adder, multiplier and delay unit.
2. It is described by differential 2. It is described by difference
equations. equations.
3. The frequency response of analog
filter can be modified by changing 3. The frequency response can be
the components. changed easily by changing the filter
4. It is a difficult process. coefficients.
4. It is more flexible.

Advantages of Digital Filters are Analog Filters:

1. The digital filter performance is not influenced by component aging, temperature
and power supply variations.
2. It is highly immune to noise.
3. More stability.
4. It can be operated over a wide range of frequencies.
5. The coefficients of Digital filters can be programmed and altered to obtain the
desired characteristics.
Basic Concept of Filters
1. A filter which allows certain range of input frequencies and rejects all other
unwanted frequencies.
2. There are 2 bands.
a) Pass band.
b) Stop band or Attenuation band.
Pass band: - The range of frequencies which is allowed by the filter is called pass band.
1. Low Pass Filter:-
Pass band  0  w  Wc
Stop band  w  Wc

2. High Pass Filter:-

Pass band w  Wc
Stop band  0  w  Wc
3. Band Pass Filter:-

Pass band w1  W  W2

Stop band 0  w  W1 W  W2

4. Band Reject Filter:-

Design of Digital Filters

There are 2 methods to design to design digital filters

(i) Direct Method
(ii) Indirect Method i.e. design of digital from analog filter
Indirect method (Procedure)
1. Map the desired digital filter Specifications in to those for an equivalent analog
filter.
2. Derive the analog filter transfer function.
3. Transform the analog filter transfer function into a digital filter transfer function.

Analog filter approximations:

Types of Analog Filter Design:-
There are 2 approaches is which you can design analog filter.
1. Butterworth.
2. Chebyshev
Analog Butterworth Low pass Filter Design:-
The Magnitude Square function of Butterworth functions is
1
H ( jw) 2  2N
W 
1   
 WC 
N  Order of filter.
WC  Cutoff frequency.
 1. As N increases the graph approaches to Ideal Response. The Magnitude function
is monolithically decreasing where the maximum .response it unity where
W=0
2. When W  WC h( Jw)  1 is monotonicallydecreas es.

3. When W  WC H ( Jw) is decreases rapidly.

4. When W  WC H ( jW ) Cure passes through 3dB point.

Design Steps :-
1. Use normalization procedure in which WC  1 rods which is to be assumed

1
H ( jW ) 
2

1  W 2N
H ( S )  H ( jW )
H ( jW )  H ( S )
2 2

1
 H ( S ).H ( S ) 
1  (W 2 ) N
1

1  ( S 2 ) N
2. Due to H(S) the poles will be available on left half of S-Plane due to H(-S) the
poles will be available on right half of S-Plane.
1
H N (S ) 
DN (S )
1
H N ( S ) 
D N ( S )
3. The poles can be obtaines by equating the denominator to Zero.
DN (S ) indirectly the left half of S-Plane.

DN (S ) indirectly the Right of S-Plane.

 1  ( S 2 ) N  0
j ( 2 K 1)
( S 2 ) N   1  e , K  1,2,   
j ( 2 K 1)
( S 2 )  e /N
 2 K 1 
j 
S2  e  N  
j ( 2 K 1)
S2  e 2
j
e 2N

  2 K 1  
j   / 2    
SK  e   2N   , K  1,2,      2 N

  2 K 1  
j   
Expression for poles S K  e 2  2 N  K  1, 2,    2 N

Butterworth Filter
jK (2 K  1) 
SK  e L  
2 2N
 L  The angle at which the Butterworth poles are available.
The Butterworth poles are calculated on a circle in S-plane.
Ex :- Let us Compute the poles of the Normalised Butterworth filter for N=3 and also
compute angle at which the poles are available.

    2  2
j    j    0.5  0.866 j ,  , 
S1  e  2 23 
e  3  3


3 j  
  e   1;  2  
j
S2 e 2
23
 5  
4
e 8 / 3
j  
   0.5  0.866 j , 3 
j
S3 e 2 23  2
3
 5  5
j  
  0.5  0.866 j
J
S4 e 2 23 
e 3

J 2
S5 e 1
J 7 / 3
S6  e  0.5  0.866 j.
As per as the stability is on armed the poles on LPF plane are only to be considered.
Therefore it Concerned system junction of Butterworth LPF cases obtained by
considering the poles on LHS plane.
1
H N (S ) 
( S  S1 ) ( S  S 2 ) (S  S 3 )
1

( S  S1 ) ( S  0.5  j 0.866) . ( S  0.5  S 0.866 )
1

( S  1) ( S  0.25  S  0.7 S )
2

1

( S  1) ( S 2  S  1)

360 0
Angular Separation b/w poles 
2N
1
1St Pole Start is at  90 0  ( Angular Separation).
2
1. Derive the normalized Butterworth polynomial for N=1, 2,4
  
J  
S1  e
J   
2 2
e  1 1   
  
J   
S2  e 2 2 1,
1
H N (S ) 
( S  1) ( S  1)
1

( S  1) 2
  
J   (1)   J 3 / 4
S1  e 2 4 
e   0.707  J 0.707
  
J   ( 3)   J 5 / 4
S2  e 2 4 
e   0.707  J 0.707
  5 
J    J 7 / 4
S3  e 2 4 
e   0.707  J 0.707
J  / 2  7 / 4  J 9 / 4
S4  e e   0.707  J 0.707
Magnitude Response of a Practical Butterworth Low Pass Filter

ωp = Pass band Frequency.

ωs = Stop band Frequency
ωc = Cutoff band Frequency
As = Stop band attenuation it is because of the change it gain due to stop band
frequency ωs.
Ap = Pass band attenuation.
 = The Parameter specifying the allowable pass band.
 = The Parameter specifying the allowable Stop band.
 1
H ( Jw)  2N
______(1)
W 
(1   2   ]1 / 2
 Wp 
1
H ( Jw) 2
 2N
W 
(1    
2

 Wp 
log arithm on both sides
 2 W 
2N

20 log H ( Jw)  10 log 1  10 log 1      _____ (2)
  Wp  
When W  Wp
20 log 10 H ( JwP )   10 log 1   2  
20 log 10 1  ( JWP )   Ap
Ap   20 log 10 H ( Jw p )

Ap  10 log 10 1   2 
  10 0.1 AP
1
1
H ( JW ) 
1   W / Wc 
2 2N

  Ws 
2N

As  10 log 10 1   2   
  Wp  
 
2N
 Ws 
 
2
  10 0.1 As
1
 Wp 
2N
 Ws  10 0.1 As
1
  
 Wp   2

2N
 Ws  10 0.1 As
1
  
 Wp  10 0.1 Ap
1
Take log on both sides
 Ws   10 0.1 As
1
2 N log    log  
 Wp   10
0.1 Ap
 1 
 Ws  10 0.1 As
1
N log    log
 Wp  10 0.1 Ap
1
 10 1
0.1 As
N  log
10 1
0.1 Ap

 Ws 
log  
 Wp 
log  / 
N
Ws
log
Wp

Since the Result of above expression is may be all integer (a) fraction value of it is a
fraction it is either integer.
Wp Wp
[Link] that Wc  
( )1 / N 1

( 10 0.1 Ap
 1) 2N

The magnitude Response of ideal Butterworth LPF which is given by

1
      (1)
2
H ( jw) 2N
 W 
1  Wc 
 Wp 
The magnitude Response of Practical Butterworth LPF which
1
      (2)
2
H ( jw) 2N
W 
1   2  
 Wp 
By equations (1) and (2)
2N
W 
2N
W 
1    1    2

 Wc   Wp 
2N
WP

2N
WC
2
WP
Wc 
 1/ N
W
Wc  P

(10 0.1 Ap
 1)1/ N
Design of Analog Low pass Butterworth Filter

1. From the given specifications, find out the order of the filter round off the order if
it C function.
2. Write the normalized transfer function H N (S ) Wc 1Rad / Sec for the above
value of N.
3. Find out the transfer function H(S) from H N (S ) [Denormalisation] by using the
following formula.
 S / Wc
H (S )  H N (S ) S

Wp
Where Wc 
 1/ N

Problems
1. Design Analog Low pass Butterworth filter 2dB Pass band attenuation at a
frequency of at a frequency 20 rad/Sec and 10dB stop band attenuation at a
frequency of 30rad/Sec.

Pass band Attenuation Ap = 2dB

 Pass band frequency Wp = 20rad/Sec
Stop band Attenuation As = 10dB
Stop band frequency Ws = 30rad/Sec.
Step 1:
Order of filter

10 0.1 As  1
N  log
10 0.1 Ap  1
 Ws 
log  
 Wp 
10  1
 log
1
10 0.2

 30 
log  
 20 
3.922

0.176
Round of to next higher integer.
N=4
Step 2:-
Normalized transfer function for N = 4
1
H N (S ) 
S 
 0.765S  1 S 2  1.847 S  1
2

for Wc  1 rad / Sec
Step 3:-
Denormalised transfer function
 S
H (S )  H N (S ) S 
Wc
Wp Wp
Wc  
 1 / N 10 0.1 Ap
1 1/ 2N

20

10 0.2
1 1/ 8

20

0.935
 21.386 Rad / Sec.
S
H (S )  H N (S ) S 
Wc
1 S
 2 S
 
S  0.765 S  1 S  1.877 S  1
2
21.386 
1
 2
S S S2 S
 0.765  1  1.847 1
21.386 2
21.386 21.386 2
21.386


21.386  4

S  16.36S  457.36S  39.5S  457.36

2 2

2091.79  10 3
H (S )  2

S  16.36 S  457.36 S 2  39.5S  
2. Determine the order and the poles of Butterworth LPF that has 3dB pass band
attenuation at 500Hz and attenuation of 40dB at 1000hZ

10 0.1 As
1
N  log
10 1 0.1 Ap

 Ws 
log
 Wp 

 
10 4
1
 log
10 1 0.3

 1000 
log  
 500 
2.001
1) order of filter 
0.301
N  6.647
Since it is a practical value to round off
 2)
N 7

J 
 2 K 1 

S1  e 2 2N  K  1, 2, 3,          14
    4 
J   J 
S1  e 2 4  e  7    0.222  0.974 j
J 5 / 7 
S2  e   0.623  0.78 iJ
J  6 / 7 
S3  e   0.9  0.433 j
J 7 / 7 
S4  e  1
J 8 / 7 
S5  e   0.9  0.433 j
J 9 / 7
S6  e   0.623  0.781 j
J 10 / 7
S7  e   0.222  0.974 j
Transfer function
1

S  1 S  0.222  0.974 j  S  0.222  0.974 j 
 S  0.623  0.781J  S  0.623  0.781J 
1

   
( S  1) S  1.8S  1 S  1.24 S  1 S 2  0.445S  1
2 2


Analog Low pass Chebyshev Filter:-

There are 2 types of Chebyshev filter.

Type-1 Chebyshev
Type-2 Chebyshev
1. Type-1 Chebyshev filter has equirripple behavior in pass band and monolithic
characteristic in stop-band.
2. Type-2 Chebyshev filter has monolithic behavior in pass band and equiripple in
stop band.
Type-1 Chebyshev low pass filter design:-
Magnitude square function of Type-1 Chebyshev LPF

1
H ( j ) 2  N 1, 2,      (1)
1  2
N   p 
2

 
C N  is Chebyshew' s Polynomial
  Parameter related to pass band
C N ( x)  Cos ( N Cos 1 x), x  1 for pass band
 Cos h ( N Cos h 1 x) x .  1 for stop band _________ (2)
The Chebyshev’s polynomial can be obtained by recursive formula.
N ( x)  2 xC N 1 ( x)  C N  2 ( x) . N  1 _______ (3)
C 2 ( x)  2 xC1 ( x)  C 0 ( x)  2 x( x)  2 x  1
2

C3 ( x)  2 xC2 ( x)  C1 ( x)  2 x(2 x 2  1)  x  4 x 3  3x
C 4 ( x)  2 xC3 ( x)  C 2 ( x) 2 X ( 4 x 3  3x)  (2 x 2  1)  8 x  8 x 2  1
For odd values of N, the Oscillations curve starts from unity,
For even values of N, the oscillations curve starts ‘N’ odd, type-1 chebyshev.

Order of Chebyshev Filter:-

1
H ( j ) 
2

2   
1   2 C N  
 p 
Taking logarithm 10 log on both sides
   
H ( J )  10 log 1  10 log 1   2 C N        (4)
2

20log   p 

At ω = ωp
   W 
20 log H ( JWp )   10 log 1   2 C N   
2

  Wp  

 20 log H ( JWp )  10 log 1   2 C N (1)
2

C N (1)  1
Ap  10 log 1   2 
  10 0.1 Ap
1

At W  Ws, in equation  (4)

  Ws  
20 log H ( jW )   10 log 1   2 C N    _________ (5)
2

  Wp 
But C N ( x)  Cos h C N Cos h 1 x) for stop band
2
 Ws  Ws
As  20 log H ( jW )  10 log ( 1   Cos h ( N Cos h 
2
 1
 1 ______ (6)
 Wp  Wp
 1  Ws 
2

10 0.1 As

 1  10 0.1 Ap

 1 Cos h ( N Cos h )    ______ (6)
  Wp  
10 0.1 AS
1  Ws 
 Cos h( NCos h 1  
10 0.1 AP
1  wp 
  Ws 
 Cos h ( N Cos h 1  
  Wp 

N 1
Cos h 1 
  
 Ws 
Cos h 1  
 Wp 

Steps to design Analog Chebyshev Filter:-

1. From the given Specifications, find out the offer of chebyshev filter. Round off to
next higher integer if it sir a fractional value.
2. Using the following formula, find out the value of ‘a’ and ‘b’
‘a’ is minor axis of ellipse
 1 N   1 N 
a  Wp  
 2 
 ‘b’ is major axis of ellipse.
 1 N   1 N 
b  Wp  
 2 
Where    1    2  1
  10 0.1 Ap
1

3. Calculate the poles of chebyshev filter which lies on the ellipse using the formula.
S L  a Cos  L  J bSin  L , l  1,2,       N
() nly Poles on LHS )

L 


2K  1 , K  1, 2,       N
2 2N

4. Find the denominator of the stranger function for above poles.

5. The numerator of the transfer function depends on value of N lie odd or even
(a) For N is Odd substitute S=0 in the denominator find out the value, then it
is numerator.
(b) For N is even substitute S=0 in the denominator and divide the result by

1   2 and this value is the numerator.

Comparison between Butterworth And Chebyshev Filter :-
Butterworth Chebyshev
1..The magnitude Response of Butterworth [Link] magnitude Response of Chebyshev
filter decreases monotonically as freque- Filter has Ripples in the pass band, stop
ncy increases ( flat filter) band depending on type of Filter ( ripple
[Link] Bandwidth is more filter)
3. The Poles of Butterworth filter lies on 2. Transition Bandwidth is less.
the Circle. 3. The Poles of Chebyshev filter lies on
4. For the same given specifications the The ellipse.
Order of the Butterworth filter is more 4. For the same given specifications the
The no. of compared to Chebyshev Order Chebyshev filter is less is
Filter. No. of poles are less when compared
To Butterworth filter.
1. Design a Chebyshev LPF Type-1 LPF for the following specifications
Ap  2.5dB, Wp  20 rad / Sec
As  30dB, Ws  50 rad / Sec
(1) Order
 
Cos h 1  
N  
 Ws 
Cos h 1  
 Wp 
 30
  10 0.1 As
 1  10 0.1 1  10 31

 31.606
 25 4.2716
  10 0.1 Ap  1  10 0.1  1   2.74 j
1.506
N 3
2. Minor axis
 1 
 N  1 
a  Wp  N
 2 
 
 
   1   2  1  0.8821  0.882 2 1
 2.645 1/ 3
 2.643
1 / 3

a  20 



 2 
a  6.6
 1 N   1 N 
b  Wp  
 2 
 2.6451 / 3  2.6451 / 3 
 20 



 2 
b  21.06

3. Poles
 S1  a Cos1  j b Sin 
S 2  a Cos 2  j b Sin  2
S 3  a Cos3  j b Sin 3

 


2 K  1 ; K  1,2,3
2 2N
 
1    120 0 ;
2 6
 
2    180 0
22
 5
3    240 0
2 10
S1  6.6Cos 120  j 21.06 Sil120
S1  3.3  J 18.238
S 2   6.6
S 3   3.3  18.238 j
4. Denominator.
( S  3.3  j18.238) ( S  3.3  j18.238) ( S  6.6)
 S 2  10.89  6.6S  332.60
 S 2  6.65  343.514

Frequency transformation in Analog Domain:-

These frequency transformation are used to convert a normalized Low Pass
( Wc = 1rad/Sec)
LPF and normalized LPF to HPF

1. Normalised LPF to HPF :-

S
S
Wc
H ( S )  H N ( S ) S  S Wc

2. Normalised LPF to HPF :-

Wc
S
S
Wc
H (S )  H N (S ) S 
S
Design of IIR digital Filters from Analog Filters
These techniques are used to convert analog filter into digital filter. The conversion
technique should have the following properties.
1. The j  axis in the S-plane should map into the unit circle in the Z-plane.
2. The left half of S-plane should map into the inside of unit circle in Z-plane.
Conversion Techniques:
1. Impulse Invariant transformation.
2. Bilinear Transformation.
Impulse Invariant Transformation
1. In this method, the IIR filter is designed such that the unit impulse response h (n)
of digital filter is the sampled version of the impulse response of the Analog filter.

Z Transform of infinite impulse response.


H ( Z )   h(n) Z  n      (1)
n 0

H ( Z ) Z  e   h( n) e
T Tn

n 0

  Ana log frequency

T
W  Digital frequency let Z  e ________ (2)
S    J
  j  T
re  e
jw

If we substitute T jT
re  e
jw
e ________ (3)

T
r e
Which gives
W T
The Real Part of the analog Pole determines the Radius of Z-Plane pole and
imaginary past of Analog pole dictates the angle of digital pole.

Case 1: Consider a pole on imaginary axis   0

Stable it maps poles on j axis into Z-plane circle.
Case 2: Poles on the left half of S-plane   0 these poles maps inside the unit
circle.
Case 3:- all poles the right half of S-plane   0 these poles maps outside the unit circle.
Note :-
But this method is not one-to-ne mapping rather it is many to –one mapping. ( many
point in S-plane are mapped to a single point it Z-plane.
Case (4):- The above note can be justified as follows consider 2 poles in S-plane with
identical real points, but the imaginary directly.
2
T

S 1    j
 2 
S 2    j   
 T 
  j T T jT
then Z 1  e e e _______ 5 (a)
  j (   2 T ) T
Z2  e
T J 2T J 2
 e .e e
T J T
 e e
These 2 poles map to the same.

Case 1:- Consider a

 There are infinite no. of S-plane poles that map to the same reflects the effect due to
sampling of impulse Response.
Drawback: - alias:
Ha(S) is the system function of Analog filter. This can be expressed as partial
function form as.

N
CK
Ha ( S )  S P
K 1
________ (6)
L

The inverse Laplace transform equation (6)

N
ha(t )   C L e PLT
K 1

If we sample ha(t) periodically at T=nT

 h(n)  ha (nt )
N
  CK
PKnT

K 1
e _________ (8)

We know H ( Z )   h(n) Z  n ________ (9)
n 0

e 
 N
H (Z )   
n
Z 1
PKT
CK
n 0 K 1

e 
N 
  CK 
n
Z 1
PKT

K 1 n 0
N
CK
  CK
1 e Z 1
PKT
K 1

if Ha ( S )
N
CK
Ha ( S )  
K 1
CK
5  Pk
then
n
CL
H (Z )  
1 e Z 1
PLT
L 1

Steps to design digital filter using Impulse invariant Method

1. For the given specifications, filed Ha(S) transfer function of analog filter.
2. Select the Sampling Rate of digital filter, T Seconds/Sample.
3. Express Analog filter as
N
CK
Ha (S )  
K 1 S  PL

4. Compute Z-transform of digital filter.

N
CK
H (Z )  
1 e Z 1
PLT
K 1
Problems:

1. Design Chebyshev HPF with

f P  1000 Hz AP  3dB
f S  500 Hz AS  40dB
From the above given that HPF Specifications but 1st for LPF
f P  500 Hz , f S  1000hZ Design 1st LPF

Chebyshev LPF :-
(1) Order
  
Cos h 1 
C  
N   0 
 Wc 
Cos h 1 
 Wp 

 
  
Cos h 1 
C 

N   0 
 fS 
Cos h 1 
 f 

 P 
  10 0.1 As
1  10 4
 1  99.994
  10 0.1 Ap
1  10 3
 1  0.997
 99.994 
Cos h 1  
N   0.997   4.025
 1000 
Cos h 1  
 500 
N 5
2) Minor axis
 1 
 N  1 
a  WP  N
 2 
 
 
   1   2  1
  0.997 1  0.997 2  1
  2.419
 2.419 1 / 5  2.419 1 / 5 
a  2f P  

 2 
a  2  500 0.177 
 177.6
a  557.97

Major axis
  1   1 
b  WP  N N

 2
 

 2  500
2.419  2.419
0.2  0.2

2
b  3190.748
(3)Poles
S L  a Cos L  jb Sin  L k  1,2,3,4,5

L 


2 K  1
2 2N
 
1    108
2 10
 3
2    144
2 10
 5
3    180
2 10
 7
4    216
2 10
 9
5    252
2 10
S1  557.97Cos108  j 3190.748Sin 108
S1   172.422  3034.581 j
S 2  557.97 Cos 144  j 3190.74  Sin 144
S 2   451.407  1875.469 j
S 3  557.97 Cos 180
S 3   557.97
S 4  557.97 Cos 216  j Sin 216 03 190.748
S 4   451.407  1.875.47 j
S 5  557.97 Cos 252  j Sin 252 03 190.748
S 5   172.42  3034.581 j

(4) denominator
 ( S  172.422  3034.581 j ) ( S  172.422  3034.581 j )
( S  451.407  1875.469 j ) ( S  451.407  1875.47 j )
( S  557.97)
 ( S  557.97) ( S 2  344.844 S  9.2384  10 6 )
( S 2  902.814S  3.721  10 6
Numerator
N is Odd
Substitute S=0 is denominator then

 (557.97) (9.2384  10 6 ) (3.721  10 6 )

 1.918  10 6
1.918  10 6
H N (S ) 
( S  557.97) ( S 2  344.844  9.234  10 6 ) ( S 2  902.814S  3721  10 6 )
S
SS S
Wc Wc
WP
WC 
Replace 10 1
0.1 Ap

1/ 2N

2  500

0.9971 / 5
 3143.48

Normalised LPF to HPF :-

Wc
H (S )  H N (S ) S 
S
Wc  3143.48
1.918  1016
H (S ) 
 3143.48 
  557.97  344.825
 5 
 3143.482 344.844 3143.48 
   9 .2383  10 6

 2 
 5 5 
 3143.48

2

902.814 3143.48  3721  10 6 
 52 
 5 
1.918  10 S
16 5
H (S ) 
557.97 S  3143.48 9.238  10 6 S 2  1.084  10 6 S  9.881  10 6
1.918  10 4 S 5
H (S ) 
557.97 S  3143.48 9.238  10 6 S 2  1.084  10 6 S  9.881  10 6
 Design examples: Analog-Digital transformations
Problems:
3. Determine H(Z) using impulse invariant transformation.
2
H (S ) 
( S  1) ( S  2)

2
1) H (S ) 
( S  1) ( S  2)

2) Sampling Rate T = 1 Sec.

N
CL
H (S )  
K 1 S  PL
2
CL C1 C2
  
K 1 S  PL S  P1 S  P2
C1 C
  2
S 1 S  2
 2 2 
  
 S 1 S  2 
 1 1 
3)  2  
 S 1 S  2 
C1  2
C2   2
P1   1
P2   2.
(4)
 2
CL
H (Z )  
1 e Z 1
PLT
K 1

C1 C2
 
1 e Z 1 1 e Z 1
P1T P 2T

2 2
H (Z )  1 1
 2
1 e Z 1  e Z 1
2 2
H (Z )  

1  0.367 Z 1  1  0.135Z 1 
1 1
1  0.135Z  1  0.367 Z
H ( Z )  ( 2)

1  0.367 Z 1 1  0.135 Z 1  
0.464


1  0.367 Z 1

1  0.135 Z 1 

3. Design IIR Digital Butterworth LPF using impulse invariant Transformation.

0.707  H e   1, 0  W   / 2
Jw

H e  0.2, 3 / 4  W  
Jw
 
WP  ;   1 Ap  3dB
2
WS  3 ;
1
 0.2; 1  2  25;
4 1  2
2  24
  4.898
  10 0.1 As
1
2  1  10 0.1 As

25  10 0.1 As

log 25  0.1As log 10 1

AS  13.979dB

WP 
2
WS  3   4.898
4
 1
1. Given set of in digital so we have to convert Digital Specifications.
into analog Specification
W  T assume T  1Sec; W  
2. Derive Analog filter transfer function
3
So S  ;   4.898
4

P   1
2
Analog Filter transfer function:-
(1) Order of filter
 
log
N 
S
log
P
log 4.898


log
4
2
N  3.918
Since it is a fractional value
N 4
2) Normalized transfer function N=4
1
H N (S ) 
( S  0.765S  1)
2
S 2
 1.847 S  1 
for WC  load / Sec
3) . Demoralized transfer function
P
C 
 1 / N
Since   1;
so
C  
2
H (S )  H N (S ) S  S
C

1
H (S ) 
 S 2
 S  
  0. 765   1
 1.57 2  1.57  

 S2 1.847 S 
   1
 1.57 2   
 1 .57 


1.57 4
S 
 1.201S  2.465 S 2  2.9S  2.46
2

6.075
H (S ) 

S  1.201S  2.46S S 2  2.95  2.465
2
 
Design of IIR digital Filter Using Bilinear Transformation

Conversions of Analog transfer function into digital frequency.

1. Bilinear Transformation technique is one of-one mapping is corresponding pole
location in Z plane.
2. There is no aliasing.
3. Let us consider an analog linear with a system function.
b
H (S )   (1)
S a
y(S ) b

X (S ) S a
Sy ( S )  ay ( S )  b  ( S )    (2).
4. The equation (2) can be characterized in a differential equation.
dy (t )
 a y (t ) bx(t ) _____(3)
dt
5. The y(t) can be approximated by trapezoidal formula where y(t) is given by
t
y(t )   y1 ( ) d  y(0)     (4)
0

6. The approximation of the integral in equation (4) by using trapezoidal formula at

t  nT and T 0  n   equation (4) can be written by

y (n ) 
2
y (n  y
1 1

(n   )  y (n   )      (5)

7. From equation (3)

y 1 (t )  bx(t )  ay(t )
y 1 (nT )  bx(nT )  ay(nT )      (6)
8. substitute equation (6) in equation (5)

T
y (nT )  ( bx(nT )  ay (nT )  bx(nT  T )  ay (nT  T )  y (nT  T )
2
 aT 
 bx(nT )  bx(nT  T )  ay (nT  T )  y (nT  T )
T
y (nT ) 1  
 2  2
 aT   aT  bT
y (nT ) 1    y (nT  T ) 1    x(nT )  x(nT  T )
 2   2  2
 aT   aT  bT
y (nT ) 1    y (nT  T ) 1   x(nT )  x(nT  T )       (7)
 2   2  2
Assume y (nT )  y (n)
 aT   aT  bT
y (n) 1    y (n  1) 1   x(nT )  x(nT  T )  __________ (8)
 2   2  2
9. Taking Z-transform on both sides of equation (8)
 aT 
1  
2 
y ( Z ) 
 aT  1
 1   Z y ( Z ) 
bT
 
1  Z 1 X ( Z )
 2  2
  aT    aT  1
y ( Z )  1     1  Z
  2    2 


bT
2

X ( Z ) 1  Z 1 
bT
2
 1  Z 1 
H (Z ) 
 aT  aT  aT  1 
1  2   2 1  2  Z 
    
bT
2

1  Z 1 
H (Z )        (9)

 1
 aT
 1  Z  2 1  Z  1 

b
H (Z )            (10)

 2 1  Z 1 
 a



T 1  Z
1
 
10. Compare equation (10) with (1)

S

2 1  Z 1 

T 1  Z 1 
2  Z  1
S  Z  1
T
11. Bilinear Transformation technique is easy to impulse element when compared to
2  Z  1
impulse invariant formation technique because ‘S’ is Replaced by
T  Z  1 
Conversion of Digital frequency to Analog frequency in Bilinear Transformation
Technique

1. Let us consider

Z  re
eJw
(digital )
S  j  J ( ana log)
2  Z 1
S  
T  Z  1
 r eJw  1
 eeJw 
2
S
T  r e  1

S


eJw

2  r e  1  J r SinW 

T  rCosW  1  JrSinW 
 
2 CosW  1   J r SinW rCosW  1  JrSin
S
T rCosW  12  r 2 Sin 2W
 2 Cos 2 w  1  j r 2 Sin WCos W  J r Sin W  J r 2 SinW  r 2 SinW
2 
S  r 
T r 2  1  2rCosW 
2  r 2  2 j r Sinw  1
S  2 
T  r  1  2r Cos W 
2  r 2 1 2 jr SinW 
S  2  2 
T  r  2r CosW  1 r  2r Cos W  1
S  T  j
2  r 2 1 
  2 
T  r  2r CosW  1 
2  2rSinW 
  2 
T  r  2r CosW  1 

1. If r  1 then  is- ve that means left half of SD-plane of poles lies with in unit
circle then the system is stable.
2. If r  1 ,  is +ve that means poles are available on the right half of S-plane poles
are exterior of the unit circle so system is unstable.
3. If r ,1  is 0 that the poles on the unit circle imaginary axis so system is
magically stable.
 2 2SinW 
   2  2CosW 
T
2 2SinW
 
T2 2CosW / 2
2 W
  tan
T 2
Steps to design IIR digital filter using Bilinear Transformation :-
1. Convert the given digital filter Specification to Analog filter Specifications by
using
2 W
 Tan .
T 2
2. Derive the Analog Transfer function for Analog Specifications.
3. Convert the Analog filter transfer H(S) function into Digital filter transfer
function H(Z) by Replacing.
2  Z  1
S  Z  1
T