EXPERT SYSTEMS AND SOLUTIONS

Email: [email protected]
[email protected]
Cell: 9952749533
www.researchprojects.info
PAIYANOOR, OMR, CHENNAI
Call For Research Projects Final
year students of B.E in EEE, ECE,
EI, M.E (Power Systems), M.E
(Applied Electronics), M.E (Power
Electronics)
Ph.D Electrical and Electronics.
Students can assemble their hardware in our
Research labs. Experts will be guiding the
projects.
IIR Filter Design
 Introduction
 IIR filter have infinite-duration impulse responses,
hence they can be matched to analog filters, all of
which generally have infinitely long impulse
responses.
 The basic technique of IIR filter design transforms
well-known analog filters into digital filters using
complex-valued mappings.
 The advantage of this technique lies in the fact that
both analog filter design (AFD) tables and the
mappings are available extensively in the literature.
 Introduction
 The basic technique is called the A/D filter
transformation.
 However, the AFD tables are available only for
lowpass filters. We also want design other frequency-
selective filters (highpass, bandpass, bandstop, etc.)
 To do this, we need to apply frequency-band
transformations to lowpass filters. These
transformations are also complex-valued mappings,
and they are also available in the literature.
 Two approaches
Apply Freq. band Apply filter
Design analog Desired
transformation transformation
lowpass filter IIR filter
s-->s s-->z

Approach 1, used in Matlab

Apply filter Apply Freq. band

Design analog Desired
transformation transformation
lowpass filter IIR filter
s-->z z-->z

Approach 2, study
 IIR Filter Design Steps
 Design analog lowpass filter
 Study and apply filter transformations to
obtain digital lowpass filter
 Study and apply frequency-band
transformations to obtain other digital filters
from digital lowpass filter
 The main problem
 We have no control over the phase
characteristics of the IIR filter.
 Hence IIR filter designs will be treated as
magnitude-only designs.
 Main Content of This Chapter
 Analog filter specifications and the properties of the
magnitude-squared response used in specifying
analog filters.
 Characteristics of three widely used analog filters
 Butterworth, Chebyshev, and Elliptic filter
 Transformation to convert these prototype analog
filters into different frequency-selective digital filter
 Some Preliminaries
 Relative linear scale
 The lowpass filter specifications on the magnitude-squared
response are given by
1
| H a ( j) | 1, |  |  p
1  2

1
0 | H a ( j) | 2 ,  s |  |
A
Where epsilon is a passband ripple parameter, Omega_p is the passband
cutoff frequency in rad/sec, A is a stopband attenuation parameter, and
Omega_s is the stopband cutoff in rad/sec.
 1
| H a ( j ) | 
2
at    p
1  2

1
| H a ( j ) |  2
2
at    s
A
The relations among epsilon, A, Rp, As,
delta1 and delta2

1 R p / 10
R p  10 log10    10 1
1  2

1
As  10 log10 2  A  10 As / 20

A
1  1 1 2 1
  
1 2 1  2
1  1
2 1 1  1
  A
1  1 A 2
 Properties of |Ha(jOmega)|2
Analog filter specifications, which are given in terms of the
magnitude-squared response, contain no phase information. Now to
evaluate the s-domain system function Ha(s), consider
H a ( j  )  H a ( s ) | s  j
| H a ( j) |2  H a ( j) H a* ( j)  H a ( j) H a ( j)  H a ( s ) H a ( s ) |s  j
or , H a ( s ) H a ( s ) | H a ( j) |2 |  s / j
Therefore the poles and zeros of the magnitude-squared function are
distributed in a mirror-image symmetry with respect to the jOmega axis.
Also for real filters, poles and zeros occur in complex conjugate pairs (or
mirror-image symmetry with respect to the real axis).
From this pattern we can construct Ha(s), which is the system function of
our analog filter.
 Properties of |Ha(jOmega)|2
 We want Ha(s) to represents a causal and stable filter.
Then all poles of Ha(s) must lie within the left half-
plane. Thus we assign all left-half poles of Ha(s)Ha(-
s) to Ha(s).
 We will choose the zeros of Ha(s)Ha(-s) lying inside
or on the jOmega axis as the zeros of Ha(s).
 The resulting filter is then called a minimum-phase
filter.
 Characteristics of Prototype
Analog Filters
 IIR filter design techniques rely on existing analog
filter filters to obtain digital filters. We designate
these analog filters as prototype filters.
 Three prototypes are widely used in practice
 Butterworth lowpass
 Chebyshev lowpass (Type I and II)
 Elliptic lowpass
 Butterworth Lowpass Filter
 This filter is characterized by the property that its
magnitude response is flat in both passband and
stopband.
1
The magnitude-squared
| H a ( j ) | 
2
2N

response of an N-order 1   
lowpass filter  c 
Omega_c is the cutoff frequency in rad/sec.
 1
| H a ( j ) |  2
2N

1   
 c 

1.2

1
N=100

0.8 N=2
N=1 N=200

0.6

0.4

0.2

0
0 10 20 30 40 50 60 70 80 90 100
 Properties of Butterworth Filter
 Magnitude Response
 |Ha(0)|2 =1, |Ha(jΩc) |2 =0.5, for all N (3dB attenuation at Ωc)
 |Ha(jΩ)|2 monotonically decrease for Ω
 Approaches to ideal filter when N→∞
To determine the system function Ha(s)
1 ( j c ) 2 N ( j c ) 2 N
H a ( s ) H a ( s )   2N  2N
 s 
2N
s  ( j c ) 2N

1     ( s  pk )
 j c 
k 1

1 j 2N ( 2 k  N 1)
pk  (1) ( j c )   c e
2N
, k  0,1, ,2 N  1
 Poles of |Ha(jΩ)|2 Ω=s/j =Ha(s) Ha(-s)
 Equally distributed on a circle of radius Ωc with angular
spacing of pi/N radians
 For N odd, pk= Ωcej2pik/N
 For N even, pk= Ωcej(pi/2N+kpi/N)
 Symmetry respect to the imaginary axis
 A pole never falls on the imaginary axis, and falls on the real
axis only if N is odd
 A stable and causal filter Ha(s) can now be specified by
selecting poles in the left half-plane 
N
H a ( s)  N
c

 ( s  pk )
LHP poles
 Matlab Implementation
 Function [z,p,k] = buttap(N)
 To design a normalized (Ωc=1) Butterworth analog
prototype filter of order N
 Z: zeros; p: poles; k: gain value
 Function [b,a] = u_buttap(N,Omegac)
 Provide a direct form (or numerator-denominator) structure.
 Function [C,B,A] = sdir2cas(b,a)
 Convert the direct form into the cascade form
 Design Equations
The analog lowpass filter is specified by the parameters, Omega_p.
R_p,Omega_s, and A_s. Therefore the essence of the design in the
case of Butterworth filter is to obtain the order N and the cutoff
frequency Omega_c.
1 at    p ,  10 log10 | H a ( j) |2  R p
| H a ( j ) | 2  2N
 at    s ,  10 log10 | H a ( j) |2  As
1   
 c 
 log10 [(10 R p /10  1) /(10 As /10  1)] 
N  
 2 log10 ( p /  s ) 
p s
c  , or  c 
2N R / 10
(10 p  1) 2N
(10 As /10  1)
 Matlab Implementation
 Function [b,a] = afd_butt(Wp,Ws,Rp,As)
 To design an analog Butterworth lowpass filter,
given its specifications.
 Function [db,mag,pha,w] =
freqs_m(b,a,wmax)
 Magnitude response in absolute as well as in
relative dB scale and the phase response.
 Chebyshev Lowpass Filter
 Chebyshev-I filters
 Have equiripple response in the passband
 Chebyshev-II filters
 Have equiripple response in stopband
 Butterworth filters
 Have monotonic response in both bands
 We note that by choosing a filter that has an equripple rather
than a monotonic behavior, we can obtain a low-order filter.
 Therefore Chebyshev filters provide lower order than
Buttworth filters for the same specifications.
 The magnitude-squared response
of Chebyshev-I filter
1
| H a ( j ) | 
2
N is the order of the filter,

1   T 
2 2
N
 Epsilon is the passband ripple
 c  factor
cos( N cos 1 ( x)), 0  x  1 Nth-order Chebyshev polynomial
TN ( x)   1
cosh(cosh ( x)), 1  x  

(a) For 0<x<1, TN(x) oscillates between –1 and 1, and

(b) For 1<x<infinity, TN(x) increases monotonically to infinity
 Observations
At x=0 (or Ω=0); |Ha(j0)|2 = 1; for N odd;
= 1/(1+epson^2); for N even
At x=1 (or Ω= Ωc); |Ha(j1)|2= 1/(1+epson^2) for all N.
For 0<=x<=1 (or 0<= Ω<= Ωc)
|Ha(jx)|2 oscillates between 1 and 1/(1+epson^2)
For x>1 (or Ω > Ωc), |Ha(jx)|2 decreases monotonically to 0 .
At x= Ωr, |Ha(jx)|2 = 1/(A^2).
 Causal and stable Ha(s)
To determine a causal and stable Ha(s), we must find the poles of
Ha(s)Ha(-s) and select the left half-plane poles for Ha(s).
The poles of Ha(s)Ha(-s) are obtained by finding the roots of
 s 
1   T 
2 2
N

 j c 

It can be shown that if pk   k  j k , k  0,1, , N  1

are the (left half-plane) roots of the above polynomial,
then
pk   k  j k , k  0,1,  , N  1
  (2k  1) 
 k  (a c ) cos  
2 2 N 
  ( 2k  1) 
 k  (b c ) sin  
2 2 N 

a
2

1 N

  1/  , b 
N

2

1 N

  N 1/  ,

1 1
   1 2
 
The poles of Ha(s)Ha(-s)

The poles fall on an ellipse

with major axis b Ωc and
minor axis a Ωc .
Now the system function is

K
H a (s) 
 ( s  pk )
k

Left half-plane
K is a normalizing factor
 Matlab Implementation
 Function [z,p,k] = cheb1ap(N,Rp)
 To design a normalized Chebyshev-I analog
prototype filter of order N and passband ripple Rp.
 Z array: zeros; poles in p array, gain value k
 Function [b,a] = u_chblap(N,Rp,Omegac)
 Return Ha(s) in the direct form
 Design Equations
Given Ωp, Ωs, Rp and As, three parameters are required to
determine a Chebyshev-I filter

  100.1Rp  1, and A  10 As / 20
s
 c   p , and  r 
c
Function
g  ( A2  1) /  2
[b,a]=afd_chb1(Wp,Ws,Rp,As)

 log g  g 2  1  
N 
 

10

 log10  r   2r  1 
 Chebyshev-II filter
Related to the Chebyshev-I filter through a simple
transformation.
It has a monotone passband and an equiripple stopband,
which implies that this filter has both poles and zeros in
the s-plane.
Therefore the group delay characteristics are better (and
the phase response more linear) in the passband than the
Chebyshev-I prototype.
1
| H a ( j ) | 
2

1   T   c /   
2 2 1
N
 Matlab implementation
 Function [z,p,k] = cheb2ap(N,As);
 normalized Chebyshev-II
 Function [b,a] = u_chb2ap(N,As,Omegac)
 Unnormalized Chebyshev-II
 Function [b,a] = afd_chb2(Wp,Ws,Rp,As)
 Elliptic Lowpass Filters
These filters exhibit equiripple behavior in the passband as
well as the stopband. They are similar in magnitude
response characteristics to the FIR equiripple filters.
Therefore elliptic filters are optimum filters in that they
achieve the minimum order N for the given specifications
These filters, for obvious reasons, are very difficult to
analyze and therefore, to design.
It is not possible to design them using simple tools, and
often programs or tables are needed to design them.
 The magnitude-squared response
1 N: the order; epsilon: passbang
| H a ( j ) | 
2

 ripple; UN() is the Nth order

1   U 
2 2
N 
 c  Jacobian elliptic function
Typical responses for odd and even N

Computation of filter order N:

 
K (k ) K 1  k12    /2 d
K (k ) K  1  k 
, K ( x )  0
p
N ,k  , k1 
1
2  s A 1
2
1  x 2 sin 2 
 Matlab Implementation
 [z,p,k]=ellipap(N,Rp,As);
 normalized elliptic analog prototype
 [b,a] = u_ellipap(N,Rp,As,Omegac)
 Unnormalized elliptic analog prototype
 [b,a] = afd_elip(Wp,Ws,Rp,As)
 Analog Lowpass Filter Design: Elliptic
 Phase Responses of Prototype Filters
 Elliptic filter provide optimal performance in the magnitude-
squared response but have highly nonlinear phase response in the
passband (which is undesirable in many applications).
 Even though we decided not to worry about phase response in
our design, phase is still an important issue in the overall system.
 At other end of the performance scale are the Buttworth filters,
which have maximally flat magnitude response and require a
higher-order N (more poles) to achieve the same stopband
specification. However, they exhibit a fairly linear phase
response in their passband.
 Phase Responses of Prototype Filters
 The Chebyshev filters have phase characteristics that
lie somewhere in between.
 Therefore in practical applications we do consider
Butterworth as well as Chebyshev filters, in addition
to elliptic filters.
 The choice depends on both the filter order (which
influences processing speed and implementation
complexity) and the phase characteristics (which
control the distortion).
 Analog-to-digital Filter
Transformations
 After discussing different approaches to the design of analog
filters, we are now ready to transform them into digital filters.
 These transformations are derived by preserving different
aspects of analog and digital filters.
 Impulse invariance transformation
 Preserve the shape of the impulse response from A to D filter
 Finite difference approximation technique
 Convert a differential eq. representation into a corresponding difference eq.
 Step invariance
 Preserve the shape of the step response
 Bilinear transformation
 Preserve the system function representation from A to D domain
 Impulse Invariance Transformation
 The digital filter impulse response to look similar to that of a
frequency-selective analog filter.
 Sample ha(t) at some sampling interval T to obtain h(n):
h(n)=ha(nT) jT
w  T or e jw
e
 Since z=ejw on the unit circle and s=j Ω on the imaginary axis,
we have the following transformation from the s-plane to the z-
plane: z=esT
1   2  Frequency-domain
H ( z)   H a  s  j k
T k    T  aliasing formula
Complex-plane mapping in impulse invariance
transformation
 Properties:
 Sigma = Re(s):
 Sigma < 0, maps into |z|<1 (inside of the UC)
 Sigma = 0, maps into |z|=1 (on the UC)
 Sigma >0, maps into |Z|>1 (outside of the UC)
 Many s to one z mapping: many-to-one mapping
 Every semi-infinite left strip (so the whole left plane) maps
to inside of unit circle
 Causality and Stability are the same without changing;
 Aliasing occur if filter not exactly band-limited
 Given the digital lowpass filter specifications wp,ws,Rp and
As, we want to determine H(z) by first designing an
equivalent analog filter and then mapping it into the desired
digital filter. Design Procedure:

1. Choose T and determine the analog frequencies:

Ωp=wp/T, Ωs=ws/T
2. Design an analog filter Ha(s) using the specifications with one
of the three prototypes of the previous section.
3. Using partial fraction expansion, expand Ha(s) into
R
H a ( s )  k 1 k
N

s  pk
4. Now transform analog poles {pk} into digital poles {epkT} to
obtain the digital filter
Rk
H ( z )  k 1
N

1  e pk T z 1
 Advantages of Impulse Invariance Mapping
 It is a stable design and the frequencies Ω and w are
linearly related.
 Disadvantage
 We should expect some aliasing of the analog frequency
response, and in some cases this aliasing is intolerable.
 Consequently, this design method is useful only
when the analog filter is essentially band-limited to a
lowpass or bandpass filter in which there are no
oscillations in the stopband.
 Bilinear Transformation
 This mapping is the best transformation method.
2 1  z 1 1  sT / 2
s 1
z
T 1 z 1  sT / 2
T T
sz  s  z  1  0 Linear fractional transformation
2 2
Complex-plane mapping in
bilinear transformation
 Observations
 Sigma < 0 |z| < 1, Sigma = 0 |z| = 1, Sigma > 0 |z| > 1
 The entire left half-plane maps into the inside of the
UC. This is a stable transformation.
 The imaginary axis maps onto the UC in a one-to-one
fashion. Hence there is no aliasing in the frequency
domain.
• Relation of ω to Ω is nonlinear
ω= 2tan-1(ΩT/2)↔ Ω=2tan(ω/2)/T;
Function [b,a] = bilinear(c,d,Fs)
 Given the digital filter specifications wp,ws,Rp and As, we
want to determine H(z). The design steps in this procedure are
the following:

1. Choose a value for T. this is arbitrary, and we may set T = 1.

2. Prewarp the cutoff freq.s wp and ws; that is calculate Ωp and
Ωs using (8.28) :
Ωp=2/T*tan(wp/2), Ωs=2/T*tan(ws/2)
3. Design an analog filter Ha(s) to meet the specifications.
4. Finally set
 2 1  z 1 
H ( z )  H a  
1 
 T 1 z 
And simplify to obtain H(z) as a rational function in z-1
 Advantage of the bilinear
Transformation
 It is a stable design
 There is no aliasing
 There is no restriction on the type of filter that
can be transformed.
 Lowpass Design Using Matlab
 Matlab Function:
 [b, a] = butter(N, wn)
 [b, a] = cheby1(N, Rp, wn)
 [b, a] = cheby2(N, As, wn)
 [b, a] = ellip(N, Rp, As, wn)
 Buttord, cheb1ord, cheb2ord, ellipord can provide filter
order N and filter cutoff frequency wn, given the
specification.
 Lowpass filter design
 Digital filter specification
 Analog prototype specifications
 Analog prototype order calculation,Omega_c,wn
 Digital Filter design (four types)
 dir2cas
 Digital filter design examples
For the same specifications, do the following:
 Ex8.21: Butterworth LP filter design;
w p  0.2 , R p  1dB
 Ex8.22: Chebyshev-I LP filter design;
ws  0.3 , As  15dB
 Ex8.23: Chebyshev-II LP filter design;

 Ex8.24: Elliptic LP filter design;

Prototype Order N Stopband Att.

Butterworth 6 15
Chebyshev-I 4 25
Elliptic 3 27
 Frequency-band Transformation
 Design other kinds of filters:
 High-pass filters
 Band-pass filters
 Band-stop filters
 Using the results of Low-pass filter and
Frequency Band Transformation
 Frequency-band Transformation
Let HLP(Z) be the given prototype lowpass digital filter, and let H(z)
be the desired frequency-selective digital filter. Define a mapping of
the form Z 1  G ( z 1 ) such that H ( z )  H ( Z ) |
LP Z 1 G ( z 1 )

Given that HLP(Z) is a stable and causal filter, we also want H(z) to be stable and
causal. This imposes the following requirements:

1. G( ) must be a rational function in 1/z so that H(z) is implementable.

2. The unit circle of the Z-plane must map onto the unit circle of the z-plane.
3. For stable filters, the inside of the unit circle of the Z-plane must also map onto
the inside of the unit circle of the z-plane.
 Frequency Transformation for
Digital Filters
The general form of the function G() that satisfies the above
requirements is a rational function of the all-pass type given by

z 1   k
Z 1 1
 G ( z )   n
k 1 , |  k | 1
1k z 1

Now by choosing an appropriate order n and the coefficients

{alpah_k}, we can obtain a variety of mappings. The most widely
used transformations are given in Table
 Use zmapping function to perform the
LP-to-HP transformation
 Digital lowpass filter specifications
 Analog prototype specification
 Analogy Chebyshev prototype filter calculation
 Bilinear transformation
 Digital highpass filter cutoff frequency
 Lp-to-HP frequency-band transformation
 dir2cas
 Comparison of FIR vs. IIR Filters
 In the case of FIR filters these optimal filters are the equiripple
filters designed via the parks-McClellan algorithm, while in
the case of IIR filters these are the elliptic filters.
 For FIR filter the standard realization is the linear-phase direct
form, while for elliptic filters cascade forms are widely used.
 Comparisons: Multiplications per output sample
 For most applications IIR elliptic filter are desirable from the
computational point of view. If we take into account the phase
equalizers, then FIR filter designs look good because of their
exact linear-phase characteristics.