DSP Unit 4 Iir Filters

UNIT-V
IIR DIGITAL FILTERS

Contents
• Introduction
– Filtering of signals
– Classification of filters
• Analog and digital
• Based on frequency response
• Practical analog filter specifications
– LPF. HPF, BPF and BSF.
• Analog filters approximation
– Butterworth and Chebyshev
– Finding Order and normalized Stable filter
– Design examples
• Analog to analog transformations
• Design of IIR Digital filters from analog filters
– Impulse invariance, step invariance and bilinear
transformations
– Design examples
• Analog to digital transformations
May 31, 2024 Prof K.Venkat Reddy 2
Infinite Impulse Response (IIR) Filters
• The basic technique is called the A/D filter

transformation.
• However, the AFD tables are available only for low-
pass filters. We also want design other frequency-
selective filters (high-pass, band-pass, band-stop,
etc.)
• To do this, we need to apply frequency-band
transformations to low pass filters. These
transformations are also complex-valued
mappings, and they are also available in the
literature.

MAIN PROBLEM IN IIR FILTERS is

• We have no control over the phase characteristics of
the IIR filter.
• Hence IIR filter designs will be treated as
magnitude-only designs.
In literature
• There are several methods to design IIR filters.
• All are based on converting stable analog filter into
stable Infinite duration Impulse Response Digital
filters.
Stability in Both Domains
Z-plane S-plane
Im Im
Marginally
Stable Marginally
Unstable
Stable
-1 1
Re Re
Stable
Stable Unstable
Discrete-Time Systems Continuous-Time Systems

• If the conversion technique is to be effective it

should posses the following desirable properties
1. The jΩ-axis in the S-plane should map into the unit
circle in the Z-plane.
2. The left half of the S-plane should map in to the
inside of the unit circle in the Z-plane.
Thus a stable Analog filter must be converted to
Stable Digital Filter.
some of the methods are
1. Impulse Invariance Transformation
2. Step Invariance Transformation and
3. Bilinear Transformation
Topics:
1. IIR Filter design: Impulse-Invariant Method
2. IIR Filter design: Step-Invariance method
3. IIR Filter design: Bilinear Transformation

Method
4. Frequency Transformations in digital domain

1. IIR Filter design: Impulse-Invariant Method

• Basic principle: Sampling of impulse response
of an analog filter.
2. IIR Filter design: Step-Invariance method
• Basic principle: Sampling of step response of
an analog filter.
3. IIR Filter design: Bilinear Transformation
Method
• Basic principle: application of the trapezoidal
formula for numerical integration of
differential equation
Impulse-Invariant Method
(Impulse Invariant Transformation)
Objective: To design an IIR filter having an impulse
response h(n) as the sampled version of the
impulse response of the analog filter h(t ):
h(t )  h(nT )  h(n) for n  0,1,2,...

where T is the sampling interval.
In consequence of this result, the frequency response of
the digital filter is an aliased version of the frequency
response of the corresponding analog filter.

Impulse Invariance Transformation method-
Behind Concept
h[t] h[n]
34567 n
t
-8-7-6-5-4-3 -2 -1 1 2 8 9 10
Impulse response of the discrete-time system be the

same as the corresponding impulse response of the
reference analog filter at sampling points.
H ( z )  Z .T h(t )
Impulse Invariance Transformation
• The Z-Transform of an IIR filter is

 
H ( Z )   h[n] Z n
 H ( Z ) Z e   h[n] e
sT
 sTn
n 0 n 0
Considering mapping of points from S-plane to Z-Plane

implied by the relation
sT
Z  e
by substituting
s    j and Z  re j
 re  e j (   j ) T
r e T
and   T
• Consider any pole
On jΩ-
axis
  0  r  e 1 0.T
 UNIT CIRCLE
Left-half S-plane
  0  r  eT  1  INSIDE UNIT CIRCLE
Right-half S-plane
  0  r  eT  1  OUTSIDE UNIT CIRCLE

UNIT CIRCLE Im jΩT

Z-plane
S-plane
ΩT ΩT
Re σT
Mapping of jΩ-axis, Stable and Unstable poles from

S-plane to Z-plane
• Therefore the impulse invariance method maps
1. Poles from the jΩ-axis of S-plane to Unit circle

of Z-plane.
2. Poles from Left-half (Negative real part) of
S-plane to inside unit circle of Z-plane.
3. Poles from Right-half (Positive real part)
S-plane to outside unit circle of Z-plane.
• Disadvantage:
It is not one-to-one mapping it is many-to-one
mapping.
jΩ
UNIT CIRCLE Im Z-plane 2π/T
S-plane
π/T
ΩT Ω
Re 2π/T
σ
-π/T
Impulse Invariance pole many-to-one mapping

• Impulse invariance pole mapping
s1    j  z1  e(  j )T  eT e jT
  j   2T T
s2    j   2
T
  z2  e  eT e jT e j 2
s2    j   2T   z2  eT e jT (because e j 2  1)
•Here z1=z2, there are infinite number of S-plane
poles that map to the same location in the Z-plane.
– They have same real part, but imaginary parts differ by
2π/T.
The S-plane poles having imaginary parts between

–π/T to π/T causes aliasing, when sampling analog
signals.
Derivation of steps
• Let Ha(s) is the system function of an analog filter

In partial Fraction form it is
N
Ck
H a (s)  
k 1 s  pk
Where {pk} are the poles of the analog filter and {Ck}
are the coefficients in the partial fraction expansion.
It’s inverse Laplace Transform is
N
ha (t )   Ck e p t
k
t0
k 1
if we sample ha(t) periodically at t=nT we have

N
h[n]  ha (nT )   Ck e p nT k
k 1
Derivation of steps
• We know that

H ( z )   h[n]z n
n 0
N
by substituting h[ n ]  ha ( nT )   k
C e p nT
k 1
k
 1 
z   Ck  e z 
 N N  N
H ( z )   Ck e pk nT n pk T 1 n
  Ck  p T 1 
n  0 k 1 k 1 n 0 k 1 1 e z 
k
.i.e. if
Ck
N
 CkN

H a (s)   then H ( z)    p T 1 
k 1 s  pk k 1  1  e z 
k

Steps in solving problems
1. From the given specifications, find Ha(s), the
transfer function of an analog filter.
2. Take / Select the sampling rate of the digital filter.
3. Express the analog filter transfer function as the
sum of single pole filters.
N
Ck
H a (s)  
k 1 s  pk
4. Compute the Z-transform of the digital filter by

using the formula
 Ck
N

H ( z)    p T 1 
k 1  1  e z 
k

5. With the previous given expressions for the
transfer function H(z), the IIR filter is easily
realized as a parallel bank of single-pole
filters.
6. If some of poles are complex-valued, they may
be paired together and combined to form two-
pole filter sections with real-valued
coefficients:
Ck Ck Ck Ck
 pk T 1
and 
s  pk 1 e z s  pk 1  e pk T z 1
Ck Ck Ck Ck
  pk T 1

s  pk s  pk 1 e z 1  e pk T z 1
7. In addition, two factors containing real-valued poles

may be combined to form two-pole filters with
real-valued coefficients.
8. Consequently, the resulting filter may be
realized as a parallel bank of two-pole filters
with real-valued coefficients.

Example 8:
For an analog system whose transfer function is given by
2
H (s) 
( s  1)( s  2)
determine the digital transfer function using Impulse

invariance
Solutionmethod,
H ( s )ifthe sampling
2 frequency is 1Hz.
( s 1)( s  2 )
:
Using partial fractions  H (s)  A
 s B2
s 1
We can get A  2, B  2
 H (s)  2
s 1  s 2 2  2
s  ( 1)  s (22 )
Using Impulse invariance technique we have
N
Ck N
Ck
If H (s)   then H ( z)  
k 1 s  pk
pk T 1
k 1 1  e z
Example 8:Cont.d
So
2 2
H ( z)  T 1

1 e z 1  e  2T z 1
For fs=1Hz  T=1sec
2 2 2 2
H ( z)  1 1
  2 1
 1

1 e z 1 e z 1  0.3678z 1  0.1353z 1
0.465 z 1
H ( z) 
1  0.503z  0.04976 z
1 2

Example 9:
Apply Impulse invariance method and find H(z) for H ( s )  sa
( s  a) 2  b 2
Solution
e  at cos bt for t  0
: The inverse Laplace h(t )  
transform of the given  0 otherwise
function is
e  anT cos(bnT ) for n  0
After sampling it is h(nT )  
 0 otherwise
 
 e jbnT
 e  jbnT

H ( z )   e cos(bnT ) z  e z 
 anT n  anT  n

n 0 n 0  2 
1  ( a  jb )T 1 n 1  ( a  jb )T 1 n
H ( z )   e z    e z 
2 n 0 2 n 0

Example 9:
sa
Apply Impulse invariance method and find H(z) for H ( s ) 
( s  a) 2  b 2
Solution 1  ( a  jb )T 1 n 1  ( a  jb )T 1 n
: H ( z )   e z    e z 
2 n 0 2 n 0
1 1 1 
H ( z)  
2 1  e ( a  jb )T z 1 1  e ( a  jb )T z 1 
1  e  aT cos(bT ) z 1
H ( z) 
1  2e  aT cos(bT ) z 1  e  2 aT z  2

Example 10:
An analog filter has a transfer function 10
H (s) 
s 2  7 s  10
Design a digital filter equivalent to this using impulse invariant
method with T=1 sec
0.4287 z 1
Answer : H ( z ) 
(1  0.1421z 1  0.0009 z  2 )
Example 11:
10
An analog filter has a transfer function H (s) 
s 2  2s  1
Design a digital filter equivalent to this using impulse invariant

method with T=1 sec
0.453z 1
(1  0.7497 z 1  0.2432 z  2 )

Step Invariance Transformation method-
Behind Concept
hs[t] hs[n]
34567 n
t
-8-7-6-5-4-3 -2 -1 1 2 8 9 10
Step response of the discrete-time system be the

same as the corresponding step response of the
reference analog filter at sampling points.
 z 1 
H ( z)   
 Z .T I .L.T  
H (s)
s
 z 
Steps derivation
As shown in the figure it is required that hs [n]  hs (t ) t  nT
H (s)
We know that hs (t )  h(t ) * u (t )  H s ( s ) 
s
or hs (t )  I .L.T  
H (s)
s
H s ( z )  Z .T of hs (t ) after sampling 
H s ( z )  Z .T I .L.T    I
H (s)
s
We also know that hs (n)  u[n] * h(n)

 z 
H s ( z)    H ( z )  II
then  z 1 
From I and II  z  H ( z )  Z .T  I .L.T H s( s )  

 z 1 
 z 1   H (s)  
So we can write H ( z)    Z .T  I .L.T   
 z    s 
Example.12:
2
H (s) 
( s  1)( s  2)
determine the digital transfer function using step

invariance
Solution method,
H ( s )  if the
2 sampling frequency
H (s) is2 10Hz.
( s 1)( s  2 )  
s s ( s 1)( s  2 )
:
Using partial fractions  H s (s)  H (s)
s  As  sB1  sC 2
We can get A  1, B  2, C  1
 H s ( s )  1s  s 21  s 1 2  hs (t )  u[t ]  2e  t  e 2t
 nT 2 nT
It’s sampled version hs (n)  u[n]  2e e
By using Z .T u[n] 
z
z 1
 
Z .T e t 
z
z  e T
 
Z .T e  2t 
z
z  e  2T
Example.12:Cont.d
Z-transform of the sampled version with T=0.1 sec is
z 2z z z 2z z
H s ( z)      
z 1 z  e T
ze  2T
z  1 z  0.9048 z  0.8187
z ( z 2  1.7235z  0.7408  2 z 2  3.637 z  1.637)

H s ( z) 
( z  1)( z 2  1.7235z  0.7408)
10 3 z (9.055 z  8.194)
H s ( z) 
( z  1)( z 2  1.7235 z  0.7408)
 z 1  9.055x103 ( z 1  0.9048z 2 )
H ( z)   H s ( z) 
 z  (1  1.7235z 1  0.7408z  2 )
Bilinear Transformation Method
The IIR filter design techniques described in the
previous sections have severe limitations in that
they are appropriate only for low-pass filter
design and limited class of band-pass filter
design.
In this section we describe a mapping from the s-
plane to the z-plane, called the bilinear
transformation, that overcomes the limitation of
the other design methods described previously.
Basic principle: application of the trapezoidal

formula for numerical integration of differential
equation.
Bilinear Transformation method
• The Bilinear transformation is a conformal mapping

that transforms the
1. That transforms the jΩ-axis into the Unit circle

in the Z-plane only once.
2. Thus avoiding aliasing of frequency components
3. All points in the L.H.P of ‘s’ are mapped inside

the unit circle in the Z-plane and
4. All points in the R.H.P of ‘s’ are mapped into

corresponding points outside the unit circle in the
Z-plane.

Bilinear Transformation method-
Derivation
Let us consider an analog linear filter of system function
b Y (s) b
H (s)   (1)    ( 2)
sa X (s) s  a
So
sY ( s )  aY ( s )  bX ( s )  (3)
It’s inverse Laplace transform (differential equation)
is equal to dy (t )
 ay (t )  bx(t )  (4)
dt
y(t) can be approximated by the trapezoidal formula
t
y (t )   y ' ( )d  y (t0 )  (5)
t0
Where y’(t) denotes the derivative of y(t).

Bilinear Transformation method-
Derivation
Instead of substituting a finite difference for the
derivative, suppose that we integrate the derivative
and approximate the integral by the trapezoidal
formula:x2
1
 f ( x)dx   x2  x1  f ( x2 )  f ( x1 ) 
2
Thus: x1
t nT
y (t )   y '( )d  y (t0 ) 
t0

nT T
y '( )d  y (nT  T )
The approximation of the previous integral by

trapezoidal formula is
T
y (nT )   y '(nT )  y '(nT  T )   y (nT  T )
2
Derivation
The approximation of above integral by the trapezoidal
formula at t=nT and t0=nT-T yields
 
y (nT )  T2 y ' (nT )  y ' (nT  T )  y (nT  T )  (6)
From the differential equ(4) we obtain
y ' (nT )  ay (nT )  bx(nT )  (7)
By Substituting equ.7 in equ.6, we get
y (nT )  T2  ay (nT )  bx(nT )  ay (nT  T )  bx(nT  T ) y (nT  T )
1  aT2 y(nT )  1  aT2 y (nT  T )  bT2 x(nT )  x(nT  T )  (8)
The Z-transform of this differential equation is
1  aT2 Y ( z )  1  aT2 z 1Y ( z )  bT2 1  z 1 X ( z )
Derivation
Transfer (System) function H(z) of the equivalent
digital filter is is
H ( z) 
Y ( z)

bT
2 1  z1

X ( z ) 1  aT2   1  aT2 z 1

bT
1  z  
1 bT
2
 1  z   
2
1  z 1 aT
2
1 1 z 1
1 z 1
aT
2
b
H ( z)   (9)
  a
2 1 z
T 1 z 1
1

Derivation
Transfer function of Transfer function of
the analog filter: the digital filter:
b b
H ( z) 
H (s) 
sa   a
2 1 z 1
T 1 z 1
By Comparing H(s) and H(z), it can be seen that

H(z) can be obtained from H(s) by using the
mapping relation
S    (10)
2 1 z 1
T 1 z 1
This relationship between s and Z is known as
Bilinear transformation.
Bilinear Transformation
one-to-one mapping
S  T2  
1 z1
1 z 1
z  1 sT
1 sT2
2

Investigation of the properties of the bilinear transformation:
1 sT2 2 s
z  re j
s    j z  1 sT  T
2 s
2 T
If s  j
2
2 2
 j     2
j arctan
T
T
T
  e 2
T
j 2 arctan
z  re j  T  e 2
2 2  j arctan
 j 2
     
2 e 2
T
T 
From this equation it can be found
T 2 
z  r  1   2arctan   tan
2 T 2
T
  2arctan  z  1   0 z 1
2
Between these limits the angle of z varies from 0 to  .
Conclusion 1: j -axis unit circle
Conclusion 2: The entire range in  is mapped only

once into range      , the aliasing errors are
eliminated. However, the mapping is highly nonlinear.
We observe a frequency compression or frequency
warping, as it is usually called, due to these nonlinearity.


  2arctan 2T
T

• Nonlinear mapping introduces a distortion in
the frequency axis called frequency warping
• Effect of warping shown below

If s    j we obtain for z:
2
2 
2 s 2         2
j arctan

j 2 / T 
z T  e
T
2 s  T
2   j
 2 
2
2 e
j arctan

2 / T 
T T
     
T 
2
2 
      2
 T 
z 
2
2 
      2
 T 
Conclusion 3:
If   0 (left-half s-plane), we find z  1(inside
unit circle)
Conclusion 4:
If   0 (right-half s-plane), we find z  1 (outside
unit circle)

2 s
Mapping: z T
2 s
T
j  j Im[s ]
Im[ z ]
0 1
  Re[s ] 0 Re[ z ]
s-plane z-plane unit circle

Another way of giving specifications
Then calculate
 p  20 log10 1
 s  20 log10  2

Summary on digital filter design
using bilinear transformation:
1. Digital filter specification:  P , S ,  1 and  2
2. Transformation of requirements of the digital filter
to the analog filter( to avoid Frequency Warping
effect)
2 
  tan P   p S   S  1 and  2
T 2
3. The analog filter design: H (s )
4. The analog filter conversion to the digital filter:
H ( z )  H ( s ) s  2  1z1 
T  1 z 1 
Example.13
2
H (s) 
( s  1)( s  2)
determine the digital transfer
function using Bilinear Transformation method, if the
sampling frequency is 1Hz.
Solution Given H ( s )  2
( s 1)( s  2 ) and T  1sec
: so
H ( z )  H ( s ) s  2  1z1   2
( s 1)( s  2 ) s  2  1 z 1 
T  1 z 1   1 
 1 z 
We can get
H ( z)  2
 2  1 z 1  1  2  1 z 1   2 
  1 z 1     1 z 1  
Example.13 Cont.d
So 2 (1 z 1 ) 2
H ( z )  2 (1 z 1

)  (1 z 1 ) 2 (1 z 1 )  2 (1 z 1 ) 
2 (1 z 1 ) 2
H ( z )  2 2 z 1

1 z 1 2  2 z 1  2  2 z 1 ) 
2 (1 z 1 ) 2 0.5 (1 z 1 ) 2
H ( z)  4 ( 3 z ) 1  ( 3 z 1 )
0.166 (1 z 1 ) 2
H ( z)  (1 0.333 z 1 )

Example.14
Design a digital Chebyshev filter to meet the following
specifications.
0.8  H (e j )  1 for 0    0.2
H (e j )  0.2 for 0.6    
Using a) Bilinear Transformation method and
b) Impulse invariance method
Solution
: 1
From  0.8    0.75
1  2
 4.899
  6.5318
1  0.75
From  0.2    4.899
1 2

Example.14 Cont.d
a) Bilinear Transformation method
 p  T2 tan( p / 2)  2 tan(0.2 / 2)  0.6498  s 2.7528
  4.236
 s  T2 tan( s / 2)  2 tan(0.6 / 2)  2.7528  p 0.6498
We can get cosh 1 ( /  ) cosh 1 (6.5318)

N 1
 1
 1.208
cosh ( s /  p ) cosh (4.2363)
So select N=2 S k   k  j k for k  1,2
 k   sin 2   2 k 1
2
sinh 
1
2
1 1
sinh (  )   k  cos2  2 k21 cosh 12 sinh 1 ( 1 ) 
1
2
1 1
sinh (  )  0.5493 sinh 1
2 sinh 1 1

(  )  0.5773   
cosh 12 sinh 1 ( 1 )  1.1547
 1, 2   sin 2 12 (0.5773)  0.4082
1, 2  cos2 12 (1.1547)  0.8165

Example.14 Cont.d
Left half plane Poles are given by S1, 2  0.4082  j 0.8165
Normalized Transfer function is
H0 H0
H (S )  

S  0.4082  j 0.8165S  0.4082  j 0.8165 ( S  0.4082) 2  (0.8165) 2 
1
H ( S ) S 0   H 0  (0.8)(0.83)  0.64
1
C   P /  1/ N
 0.6498 / 0.75  0.75
2
1  2
0.64
H (S ) 
S 2  0.816 S  0.83
Denormalized Transfer function is
0.64 0.64
H (s)  H (S ) S  s  
c S 2  0.816S  0.83 S  0.s75 ( 0.s75 ) 2  0.816( 0.s75 )  0.83
0.36
H (s)  2
s  0.612 s  0.467
Example.14 Cont.d
0.36
H ( z )  H ( s ) s  2  1z1   2
T  1 z 1  s  0.612s  0.467 s  2 1z11 
 1 z 
0.36
H ( z) 
4    0.6122  0.467
1 z 1
1 z 1
2
1 z 1
1 z 1
0.36(1  z 1 ) 2
H ( z) 
4(1  z 1 ) 2  1.224(1  z 1 )(1  z 1 )  0.467(1  z 1 ) 2

Example.14 Cont.d
0.36( z 2  2 z 1  1)
H ( z) 
4( z  2  2 z 1  1)  1.224(1  z  2 )  0.467( z  2  2 z 1  1)
1 2
0.36(1  2 z  z )
H ( z)  1 2
5.691  7.066 z  3.243z
1 2
0.063(1  2 z  z )
H ( z)  1 2
1  1.242 z  0.57 z

Example.14 Cont.d
 s s 0.6
b) Impulse invariance method   3
 p  p 0.2
  0.75   4.899   /   4.899 / 0.75  6.5318
cosh 1 ( /  ) cosh 1 (6.5318)
N 1
 1
 1.45
cosh ( s /  p ) cosh (3)
0.64
Normalized Transfer function is H (S ) 
S 2  0.816S  0.83
Denormalized Transfer function is

0.64 0.64
H (s)  H (S ) S  s  2  s 2
c S  0.816S  0.83 S  0.6283
s ( 0.6283 )  0.816( 0.6283
s
)  0.83
0.24
H (s)  2
s  0.51s  0.32

Example.14 Cont.d
(0.48)(0.5)
H (s) 
( s  0.255) 2  (0.5) 2
0.255t
h(t )  0.48e sin(0.5t )u (t )
0.255 nT
h(nT )  0.48e sin(0.5nT )u (nT )
 aT 1
0.48e sin(bT ) z
H ( z)   aT 1  2 aT  2
1  2e cos(bT ) z  e z

Example.14 Cont.d
0.255T 1
0.48e sin(0.5T ) z
H ( z)   0.255T 1  0.51T  2
1  2e cos(0.5T ) z  e z
T=1 Sec
0.255 1
0.48e sin(0.5) z
H ( z)   0.255 1  0.51  2
1  2e cos(0.5) z  e z
1
0.178 z
H ( z)  1 2
1  1.36 z  0.6 z
Example.15
An analog filter has a transfer function s 2  4.525
H (s)  2
s  0.6925s  0.504
Design a digital filter equivalent to this using Bilinear
transformation method with T=1 sec
1.447  0.178 z 1  1.447 z 2
1  1.187 z 1  0.529 z 2
Example 16:
For the given specifications αp=3dB, αs=15dB, Ωp=1000 rad/s
and Ωs=500 rad/s design a butter worth approximated digital
filter using Bilinear transformation.
0.453z 1
(1  0.7497 z 1  0.2432 z 2

Frequency Transformations
in Digital Domain

Frequency Transformations in
Digital Domain
As in the analog domain, frequency
transformations can be performed on a digital
low-pass filter to convert it to either a band-pass,
band-stop, or band-stop filter. The
1
transformation involves replacing the variable z
 
g z 1
by a rational function , which must
satisfy the following properties:
1. The mapping z  g  z  must map points inside

1 1
the unit circle in the z-plane into itself.

2. The unit circle must also be mapped into itself.

Digital Domain
The second condition implies that for r=1,
e  j  g  e  j   g ( ) e
j arg g ( ) 
It is clear, that we must have

g ( )  1 for all 
That is, the mapping must be all-pass. Hence it is of the
form n
z 1
 k
g z   
1
k 1 1   k z 1
where  k  1 ensures that a stable filter is transformed

into another stable filter (i.e. satisfies condition 1).
Digital Domain
1. Low-Pass to Low-Pass Transformation
z 1

z 1 
1  z 1


sin  p   'p  2 
sin  p   'p  2
 p : passband frequency of lowpass filter
 'p : passband frequency of new filter
 z 1   
 
H z 1  H  
1 
 1  z 
Digital Domain
2. Low-Pass to High-Pass Transformation
1 z 1

z 
1  z 1
 

cos  'p   p  2 
cos '
p p  2
 p : passband frequency of lowpass filter
 'p : passband frequency of highpass filter
 z 1   
 
H z 1  H   
1 
 1  z 
Digital Domain
3. Low-Pass to Band-Pass Transformation
2k 1 k  1 cosu  l  2
z  2
z  
1 k  1 k  1 cosu  l  2
z 
k  1  2 2k 1
z  z 1 p
k 1 k 1 k  cotu  l  2 tan
u : upper passband frequency
2
l : lower passband frequency
 2 2k 1 k  1 
 z  z  
H z  
1
 H k 1 k 1 
 k  1 z  2  2k z 1  1 
 
 k 1 k 1 
Digital Domain
4. Low-Pass to Band-Stop Transformation
2 1 1  k cosu  l  2
z 2
z  
1 k 1 k cosu  l  2
z 1 
1  k  2 2 1
z  z 1 p
1 k 1 k k  tanu  l  2 tan
2
u : upper passband frequency
l : lower passband frequency
  2 2 1 1  k 
 z  z  
 
H z 1
 H 1 k 1 k 
 1  k z  2  2 z 1  1 
 
 1 k 1 k 
Comments on frequency
transformations in digital domain
Since, the frequency transformations may be
performed either in the analog domain or in the
digital domain, a filter designer has a choice which
approach will take.
We know that the impulse invariance method
and the step invariance method are
inappropriate to use in designing high-pass
and many band-pass filters due to the aliasing
problem.
Consequently, one would not employ analog
frequency transformations followed by conversion of
the result into digital domain by use of these two
mappings.
Comments on frequency
transformations in digital domain
Instead, it is much better to perform the mapping

from an analog low-pass filter into a digital low-
pass filter by either of these mappings and then to
perform the frequency transformations in the digital
domain. Thus the problem of aliasing is avoided.
In the case of bilinear transformation, where

aliasing is not a problem, it does not matter the
frequency transformation is performed in the
analog domain or in the digital domain. In fact, in
this case only, the two approaches result in
identical digital filters.

Filter Specifications
• Specifications
– Passband
0.99  Heff j   1.01 0    22000 

– Stopband
Heff j   0.001 23000   
• Parameters
1  0.01
2  0.001
p  22000 
 s  23000 
• Specs in dB
– Ideal passband gain =20log(1) = 0 dB
– Max passband gain = 20log(1.01) = 0.086dB
– Max stopband gain = 20log(0.001) = -60 dB
Butterworth Lowpass Filters
• Passband is designed to be maximally flat
• The magnitude-squared function is of the form
1 1
Hc j   Hc s 
2 2
1  j / j c  1  s / j c 
2N 2N
sk   1 jc   cej / 2N2k N1

1 / 2N
for k  0,1,...,2N - 1
Chebyshev Filters
• Equiripple in the passband and monotonic in the stopband
• Or equiripple in the stopband and monotonic in the passband
Hc j  
2 1
1   V  /  c 
2 2

VN x   cos N cos 1 x 
N

Filter Design by Impulse Invariance
• Remember impulse invariance
– Mapping a continuous-time impulse response to discrete-time
– Mapping a continuous-time frequency response to discrete-time
hn  Tdhc nTd 
  2 
 

He j
  Hc  j
 j k 
k    Td Td 
• If the continuous-time filter is bandlimited to
Hc j   0    / Td
 
 
H e j  Hc  j   
 Td 
• If we start from discrete-time specifications Td cancels out
– Start with discrete-time spec in terms of 
– Go to continuous-time =  /T and design continuous-time filter
– Use impulse invariance to map it back to discrete-time = T
• Works best for bandlimited filters due to possible aliasing

Impulse Invariance of System Functions
• Develop impulse invariance relation between system functions
• Partial fraction expansion of transfer function
N
Ak
Hc s  
k 1 s  sk
• Corresponding impulse response

N
 Ak esk t t  0
hc t   k 1
 0 t0
• Impulse response of discrete-time filter

 
N N
hn  Tdhc nTd   un  TdAk esk Td un
n
 TdAk e
k 1
sknTd
k 1
• System function
N
TdAk
Hz    sk Td 1
k 1 1  e z
esk Td
• Pole s=sk in s-domain transform into pole at
Example
• Impulse invariance applied to Butterworth
 
0.89125  H e j  1 0    0.2
He   0.17783
j
0.3    
• Since sampling rate Td cancels out we can assume Td=1

• Map spec to continuous time
0.89125  Hj   1 0    0.2
Hj   0.17783 0.3    
• Butterworth filter is monotonic so spec will be satisfied if

Hc j0.2   0.89125 and Hc j0.3   0.17783
1
Hc j  
2
1  j / j c 
2N
• Determine N and c to satisfy these conditions

Example Cont’d
• Satisfy both constrains
2N 2 2N 2
 0.2   1   0.3   1 
1      and 1     
 c   0.89125   c   0.17783 
• Solve these equations to get
N  5.8858  6 and  c  0.70474
• N must be an integer so we round it up to meet the spec
• Poles of transfer function
sk   1 jc   cej / 12 2k 11 for k  0,1,...,11
1 / 12
• The transfer function

0.12093
Hs 
 
s2  0.364s  0.4945 s2  0.9945s  0.4945 s2  1.3585s  0.4945  
• Mapping to z-domain
0.2871  0.4466z 1  2.1428  1.1455z 1
Hz   1 2

1  1.2971z  0.6949z 1  1.0691z 1  0.3699z  2
1.8557  0.6303z 1

1  0.9972z 1  0.257z  2 74
May 31, 2024 Prof K.Venkat Reddy
Example Cont’d

Filter Design by Bilinear Transformation
• Get around the aliasing problem of impulse invariance
• Map the entire s-plane onto the unit-circle in the z-plane
– Nonlinear transformation
– Frequency response subject to warping
• Bilinear transformation
2  1  z 1 
s  
1 
Td 1  z 
• Transformed system function
2  1  z 1 
Hz   Hc   
1  
 Td  1  z 
• Again Td cancels out so we can ignore it
• We can solve the transformation for z as
1  Td / 2s 1  Td / 2  jTd / 2 s    j
z 
1  Td / 2s 1  Td / 2  jTd / 2
• Maps the left-half s-plane into the inside of the unit-circle in z

– Stable in one domain would stay in the other
• On the unit circle the transform becomes
1  jTd / 2
z  e j
1  jTd / 2
• To derive the relation between  and 

2  1  e  j  2  2e  j / 2 j sin / 2 2 j  
s      j    j / 2   tan 
Td  j 
1  e  Td  2e cos / 2 Td 2
• Which yields
2    Td 
 tan  or   2 arctan 
Td 2  2 


Example
• Bilinear transform applied to Butterworth
 
0.89125  H e j  1 0    0.2
He   0.17783
j
0.3    
• Apply bilinear transformation to specifications
2  0.2 
0.89125  Hj   1 0   tan 
Td  2 
2  0.3 
Hj   0.17783 tan   
Td  2 
• We can assume Td=1 and apply the specifications to

1
Hc j  
2
1   /  c 
2N
• To get
2N 2 2N 2
 2 tan 0.1   1   2 tan 0.15   1 
1      and 1     
 c   0.89125   c   0.17783 

Example Cont’d
• Solve N and c
  1 
2
   1 
2

log   1      1 
  0.17783     0.89125 
 


N  5.305  6  c  0.766
   
2 log tan 0.15 tan 0.1 
• The resulting transfer function has the following poles
sk   1 jc   cej / 12 2k 11 for k  0,1,...,11
1 / 12
• Resulting in
0.20238
Hc s 
  
s2  0.3996 s  0.5871 s2  1.0836 s  0.5871 s2  1.4802 s  0.5871 
• Applying the bilinear transform yields
Hz  
0.0007378 1  z 1 
6
1  1.2686z 1

 0.7051z 2 1  1.0106 z 1  0.3583z 2 
1

1  0.9044z 1
 0.2155z 2 
Example Cont’d

DSP Unit 4 Iir Filters

Uploaded by

Document Information

Original Title

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

DSP Unit 4 Iir Filters

Uploaded by

Copyright:

Available Formats

UNIT-V

IIR DIGITAL FILTERS

• The basic technique is called the A/D filter

May 31, 2024 Prof K.Venkat Reddy 3

MAIN PROBLEM IN IIR FILTERS is

Discrete-Time Systems Continuous-Time Systems

May 31, 2024 Prof K.Venkat Reddy 5

• If the conversion technique is to be effective it

2. IIR Filter design: Step-Invariance method

3. IIR Filter design: Bilinear Transformation

4. Frequency Transformations in digital domain

1. IIR Filter design: Impulse-Invariant Method

h(t )  h(nT )  h(n) for n  0,1,2,...

May 31, 2024 Prof K.Venkat Reddy 9

Impulse response of the discrete-time system be the

• The Z-Transform of an IIR filter is

Considering mapping of points from S-plane to Z-Plane

• Consider any pole

  0  r  eT  1  OUTSIDE UNIT CIRCLE

May 31, 2024 Prof K.Venkat Reddy 12

UNIT CIRCLE Im jΩT

Mapping of jΩ-axis, Stable and Unstable poles from

• Therefore the impulse invariance method maps

1. Poles from the jΩ-axis of S-plane to Unit circle

Impulse Invariance pole many-to-one mapping

The S-plane poles having imaginary parts between

• Let Ha(s) is the system function of an analog filter

if we sample ha(t) periodically at t=nT we have

May 31, 2024 Prof K.Venkat Reddy 18

4. Compute the Z-transform of the digital filter by

May 31, 2024 Prof K.Venkat Reddy 19

7. In addition, two factors containing real-valued poles

May 31, 2024 Prof K.Venkat Reddy 21

determine the digital transfer function using Impulse

May 31, 2024 Prof K.Venkat Reddy 23

May 31, 2024 Prof K.Venkat Reddy 24

May 31, 2024 Prof K.Venkat Reddy 25

Design a digital filter equivalent to this using impulse invariant

May 31, 2024 Prof K.Venkat Reddy 26

Step response of the discrete-time system be the

We also know that hs (n)  u[n] * h(n)

From I and II  z  H ( z )  Z .T  I .L.T H s( s )  

determine the digital transfer function using step

z ( z 2  1.7235z  0.7408  2 z 2  3.637 z  1.637)

Basic principle: application of the trapezoidal

• The Bilinear transformation is a conformal mapping

1. That transforms the jΩ-axis into the Unit circle

3. All points in the L.H.P of ‘s’ are mapped inside

4. All points in the R.H.P of ‘s’ are mapped into

May 31, 2024 Prof K.Venkat Reddy 32

Where y’(t) denotes the derivative of y(t).

May 31, 2024 Prof K.Venkat Reddy 33

The approximation of the previous integral by

May 31, 2024 Prof K.Venkat Reddy 36

By Comparing H(s) and H(z), it can be seen that

May 31, 2024 Prof K.Venkat Reddy 38

Between these limits the angle of z varies from 0 to  .

Conclusion 1: j -axis unit circle

Conclusion 2: The entire range in  is mapped only

May 31, 2024 Prof K.Venkat Reddy 40

May 31, 2024 Prof K.Venkat Reddy 41

May 31, 2024 Prof K.Venkat Reddy 42

May 31, 2024 Prof K.Venkat Reddy 44