Digital Filter Design Using MATLAB

Notation; if we have a filter described as:

y(n)= b0 x(n)+ b1 x(n-1)+ b2 x(n-2)+ b3 x(n-3)+…..-a1 y(n-1)-a2 y(n-2)+a3 y(n-3)-…..
Rewrite the equation:
b0 x(n)+ b1 x(n-1)+ b2 x(n-2)+ b3 x(n-3)+…- a0(=1) y(n)- a1 y(n-1)-a2 y(n-2)+a3 y(n-3)-…=0
Rewrite in matrix form:

 x ( n)   y ( n) 
 x(n − 1)   y(n − 1) 
   
[b0 b1 b2 b …] *  x (n − 2)  – [1 a a a …]*  y (n − 2) =0
3 1 2 3 or B*X-A*Y=0

   
 x(n − 3)   y(n − 3) 
 .. . .   . . . . 
(eq.1)
Example 1: A moving average filter with time window length of 4 is defined as:
y(n)= 0.25*x(n)+ 0.25*x(n-1)+ 0.25*x(n-2)+ 0.25*x(n-3)
The A and B matrix for this filter are:
B=[b0 b1 b2 b3 …]=[0.25 0.25 0.25 0.25]
A=[1 a1 a2 a3 …]=[1]
Note 1; For executing the commends follow, Signal Processing Toolbox is needed.
A real digital filter in market is designed in a way that you can change the A and B
coefficient by means of software. It means that you can implement any kind of filter by
changing A and B. Commend named filter is in MATLAB for propose of simulating a
digital filter.
Y=filter(B,A,x)
Where A and B is the coefficients of the filter as in eq. 1. X is input signal to the filter
and y is output of the filter.
Example 2:simulate the filter in ex.1 with a input signal as follow:
time=[0:0.02:20];%Ts=0.02>>Fs=1/0.02=50 Hz
signal=cos(2*pi*0.1*time)+0.25*cos(2*pi*0.5*time)+ .1*cos(2*pi*2*time);
%0.1,0.5 and 2 Hz components
noise=randn(1,1001);
x=noise+signal;
 signal without noise x=signal with noise
1.5 2

1.5
1

0.5
0.5

0 0

-0.5
-0.5

-1

-1
-1.5

-1.5 -2
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200

The filter in ex1 can be defined in MATLAB as follow:

A= [1]; B=0.25*[1 1 1 1]; y=filter(B,A,x);
We also can test the signal for a moving average filter of length of
40, by just changing B to B=0.025*ones(40,1);
y=Output of 4 length moving avrage filter y=Output of 40 length moving avrage filter
1.5 1.5

1 1

0.5 0.5

0 0

-0.5 -0.5

-1 -1

-1.5 -1.5
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200

Digital filters usually consist of A/D and D/A modules that enable it
to work in analog circuits. After implementing coefficients A and B to
the filter we usually test the filter by using a signal generator and
oscilloscope. Command freqz in MATLAB do the oscilloscope and signal
generator job together.
[h,w]=freqz(B,A,n,Fs)
Where; A and B is the same in eq.1 , n is number of points we want to
see the frequency response graph, Fs is sampling frequency. H is
magnitude and w is angular frequency respond to the h. Note that if we
drop the n the default is 512 and if drop Fs the default is pi=3.14.
Ex.3; test the both filters in ex.2 by means of freqz. Suppose sampling
frequency is 50Hz. frequency responce of 4 length moving avrage filter

A= [1];
1

0.9
B=0.25*[1 1 1 1]; 0.8

[h1,w1]=freqz(B,A,10); 0.7

figure(1) 0.6

plot(w1,abs(h1),'r+:'); 0.5

0.4

A= [1]; 0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
B=0.025*ones(40,1);
[h2,w2]=freqz(B,A);

[h3,w3]=freqz(B,A,1000,50);
frequency responce of 40 length moving avrage filter
1

0.9
frequency responce of 40 length moving avrage filter
0.8 1

0.7 0.9

0.6 0.8

0.5 0.7

0.4 0.6

0.3
0.5
0.2
0.4
0.1
0.3
0
0 0.5 1 1.5 2 2.5 3 3.5 0.2

0.1

0
0 5 10 15 20 25

FIR Filter
Parks-MacClellan optimal Fir filter design:
Using remez command: B=remez(n,f,a)
Where; n is the order of filter(number of coefficients is n+1), f and a
are frequency-amplitude characteristics.
Ex.4,Design a low pass filter order 20 with a cut off frequency of 3 Hz
, transient band of 1 Hz, and sampling frequency of 50Hz. And test it
with the same signal in Ex.2.
Note that the frequency vector is between 0 and 1 which correspond with
0 and Fs/2 Hz respectively. 1

n=20; 0.9

f=(1/25)*[0 3 4 25]; 0.8

a=[1 1 0 0]; plot(f,a)>>>>>> 0.7

0.6
amplitude

B=remez(n,f,a) 0.5

0.4
A=[1]; % note that for FIR filter always A=1 0.3
[h1,w1]=freqz(B,A,100,50); 0.2

plot(w1,abs(h1),'r+:'); 0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
frequency responce of 20 order FIR filter frequency
1.4

1.2

B=
Columns 1 through 7
0.8

0.6 -0.1085 -0.0078 0.0019 FIR 0.0182 0.0399 0.0649 0.0908

filter b coefficients

0.4
Columns 80.2through 14
0.1148 0.1344 0.1471 0.1515 0.1471 0.1344 0.1148
0.2 Columns 0.15
15 through 21
0
0.0908 0.0649
0.1
0.0399 0.0182 0.0019 -0.0078 -0.1085
0 5 10 15 20 25

0.05

0
For testing just excute y=filter(B,A,x); FIR filter b coefficients
-0.05

-0.1

-0.15
0 5 10 15 20 25
 y=Output of 20 order FIR filter
2

1.5

0.5

-0.5

-1

-1.5

-2
0 200 400 600 800 1000 1200

Ex.5 Do the E.4 with 200 order FIR filter.

The only change is just n=200 the rest of Codes are the same.
Note that the delay in filtered signal.
y=Output of 200 order FIR filter togther with noise free signal frequency responce of 200 order FIR filter
1.5 1.4

1 1.2

1
0.5

0.8

0
0.6

-0.5
0.4

-1 0.2

0
-1.5 0 5 10 15 20 25
0 200 400 600 800 1000 1200

Ex.6; design a band pass filter to just pass 0.5 and 2 Hz component.
We choose n=200 and f=(1/25)*[0 0.4 0.45 2.1 2.2 25] then a=[0 0 1 1 0
0].
n=200; 1

f=(1/25)*[0 0.4 0.45 2.1 2.2 25]; 0.9

a=[0 0 1 1 0 0]; 0.8

B=remez(n,f,a); 0.7

A=[1]; 0.6
amplitude

[h1,w1]=freqz(B,A,100,50); 0.5

0.4

0.3

0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
frequency
 frequency responce of 200 order FIR filter
1.4

1.2

0.8

0.6

0.4

0.2
There is a problem here?!
N=201 or 199 will solve the Problem.
0
0 5 10 15 20 25

frequency responce of 201 order FIR filter

1.4

1.2
y=Output of 201 order band pass FIR filter
0.8
1
0.6

0.8
0.4

0.2 0.6

0
0.4

-0.2
0.2
-0.4

0
-0.6 0 5 10 15 20 25
0 200 400 600 800 1000 1200

Ex.7. Design a minimum-order low pass filter with a 3 Hz pass band

cutoff frequency and 4 Hz stop band cutoff frequency with a sampling
frequency of 50 Hz with maximum stop band amplitude of 0.1 and a
maximum pas band error of 0.05.
We need to use remezord command.
[n,f0,a0]=remezord(f,a,dev,Fs)
Where; f and a are frequency-amplitude characteristics (f and a are the
same as in remez command), dev is acceptable error in pass and stop
bands. n is the order of filter(number of coefficients is n+1), f0 and
a0 is the normalized frequency and amplitude.
f=[ 3 4 ]; a=[1 0]; 1.4
frequency responce of minimum order FIR filter

dev=[0.05 0.1];Fs=50; n=
[n,f0,a0]=remezord(f,a,dev,Fs) 45 1.2

B=remez(n,f0,a0); A=[1]; f0 = 1
[h1,w1]=freqz(B,A,100,Fs); 0
0.1200
0.8
FIR filter b coefficients
0.14

0.12
0.1600 0.6

0.1
1.0000
0.4
0.08

0.2
a0 =
0.06

0.04

0.02
1 0
0 5 10 15 20 25

0
1
-0.02 0
-0.04
0 5 10 15 20 25 30 35 40 45 50
0

FIR filter using direct method

The same example Professor Marti
explained in the Class in MATLAB:
Hw=[1 1 1 0 0 0 0 0 0 1 1]; >>>>
Inverse Fourier Transform
ha=ifft(Hw);

N = 11
ha2=[ha,ha]; >>>>>>>>>>> indStart = 7
N=max(size(ha)); indEnd = 17
indStart=(N+3)/2; B = Columns 1 through 11
0.0694 -0.0540 -0.1094 0.0474
indEnd=indStart+N-1; 0.3194 0.4545 0.3194 0.0474
B=ha2([indStart:indEnd]); -0.1094 -0.0540 0.0694

Hw=[1 1 1 0 0 0 0 0 0 1 1];
figure(5);stem(Hw);
ha=ifft(Hw);
figure(1);plot(ha);
ha2=[ha,ha]; function [B] = FIRDirect(Hw) >> B=FIRDirect([1 1 1 0 0 0 0 0 0 1 1]);
figure(2);plot(ha2); ha=ifft(Hw); >> plot(B)
N=max(size(ha)); ha2=[ha,ha]; >> B=FIRDirect([0 0 0 1 1 1 1 1 1 0 0]);
indStart=(N+3)/2; N=max(size(ha)); >> plot(B)
indEnd=indStart+N-1; indStart=(N+3)/2;
B=ha2([indStart:indEnd]); indEnd=indStart+N-1;
figure(3);plot(B); B=ha2([indStart:indEnd]);
 IIR Filter
There are butter,cheby1,cheby2,ellip,maxflat, and yulewalk commands for
IIR filter design. Formats for them almost the same as follow:
[B,A]=command(n,r,w)
where, B and A are filter coefficients and n is the order of filter as
FIR filter. R is ripples (if applicable) and w is cutoff frequency (or
frequencies). For example Butterworth filter format is as follow:
[B,A]=butter(n,w)
Where B,A,n,w are described in general format. This format just
generates the low pass filter with a cutoff frequency w.
Ex.8. Design a Butterworth digital filter with the same characteristics
of example 4. As in the ex.4 n=20 and w=3/25(note that Fs/2=25)
Fs=50; frequency responce of 20 order butterworth filter
w=3/25; 1.4

n=20;
[B,A]=butter(n,w); 1.2

[h3,w3]=freqz(B,A,512,Fs);
1

2 1 0.8

H ( w) = 2N
w 0.6
1+
w 

 c 0.4

0.2
If we want to design a high
Pass filter we need to use 0
The following format: 0 5 10 15 20 25

[B,A]=butter(n,w,’ftype’)
frequency responce of 20 order butterworth filter
1.4
Ex.9. design high pass filter
With the same values in ex.8. 1.2

The only difference is this Line:

1
[B,A]=butter(n,w,’high’);
0.8

0.6

0.4
frequency responce of 20 order butterworth band pass filter
2.5
0.2

2 0
0 5 10 15 20 25

1.5

1
Ex.10, Design a band pass filter
With cutoff frequencies at 3
0.5
and 10 as in ex.9.
From the example w=(1/25)*{3 10]
0
0 5 10 15 20 25
The rest is the same as ex.9.
 frequency responce of 2*20 order butterworth band stop filter
1.4 For a stop band , ‘stop’ will replace ftype
1.2 and order of the filter will become 2*n
Ex.11.change the filter in example 10 to
1
band stop filter.
0.8 The only change is:
[B,A]=butter(n,w,'stop');
0.6
The rest is the same as ex.10.
0.4

0.2 frequency responce of 20 order elliptic band stop filter

1.8

0 1.6
0 5 10 15 20 25

1.4

1.2

Ex.12.Repeat example 11 using elliptic filter. 0.8

The only change is: 0.6

[B,A]=ellip(n,0.5,10,w,'stop');
0.4
The rest is the same as ex.11.
The format for this command is: 0.2

[B,A]=ellip(n,Rp,Rs,w,'ftype'); 0
0 5 10 15 20 25
suggested values for Rp and Rs are
0.5 and 20. Then by try and error
will come up with a better result.

frequency responce of 20 order cheby1 band stop filter

4

3.5

3
Ex.13.Repeat example 11 using cheby1 filter.
The only change is:
2.5

[B,A]=cheby1(n,0.5,w,'stop'); 2

The rest is the same as ex.11. 1.5

The format for this command is:
[B,A]=cheby1(n,R,w,'ftype');
1

suggested values for R is 0.5. 0.5

Then by try and error will come up 0

0 5 10 15 20 25
with a better result.

frequency responce of 10 order cheby1 band stop filter with R=0.01

1.4

1.2

Ex.14.Repeat example 13, and try to get 1

a better response. 0.8

The only changes are: 0.6

n=10;
R=0.01; 0.4

I did these changes by try and error. 0.2

It is also possible to use cheb1ord
Command to calculate optimum order n. 0
0 5 10 15 20 25
 frequency responce of 7 order elliptic band stop filter
1.4

1.2

Ex.15.Repeat example 12, and try to get 1

a better response.
The only changes are: 0.8

n=7; 0.6

[B,A]=ellip(n,0.5,20,w,'stop');
0.4

I did these changes by try and error. 0.2

It is also possible to use ellipord

Command to calculate optimum order n. 0
0 5 10 15 20 25

1.4
frequency responce of 2*12 order butterworth band stop filter
Ex.16: Repeat example 11 , and try to get a
1.2
better response. The only change is:
n=12;
1 [B,A]=butter(n,w,'stop');
Plot A and B coefficients.
0.8
B = 1.0e+004 * Columns 1 through 25
0.0000 -0.0001 0.0005 -0.0028 0.0123 -0.0418 0.1149
0.6
.2625 0.5068 -0.8366 1.1906 -1.4684 1.5742 -1.4684
1.1906 -0.8366 0.5068 -0.2625 0.1149 -0.0418 0.0123
0.4
-0.0028 0.0005 -0.0001 0.0000
A = 1.0e+004 * Columns 1 through 25
0.2
0.0001 -0.0013 0.0084 -0.0355 0.1104 -0.2688 0.5330
-0.8824 1.2405 -1.4991 1.5699 -1.4324 1.1420 -0.7964
0
0 5 10 15 20 25 0.4852 -0.2574 0.1183 -0.0467 0.0156 -0.0044 0.0010
-0.0002 0.0000 -0.0000 0.0000

4 FIR filter A coefficients

x 10
4 FIR filter B coefficients 2
x 10
2

1.5
1.5

1
1

0.5
0.5

0
0

-0.5 -0.5

-1 -1

-1.5 -1.5
0 5 10 15 20 25 0 5 10 15 20 25
 Special Functions
FIR1 command uses window method to use n th order low pass FIR filter.
B=fir1(n,w,’ftype’,’windowtype’)
Where, n is the order of filter (number of coefficients is n+1), and w
is cutoff frequency (or frequencies). Default for ftype is low pass
filter. ‘high’ and ‘stop’ are other values that ftyp can have. Defult
for Windowtype is Hamming window other available window are Boxcar,
Hanning, Barlett, Blackman, Kaiser and chebwin.

Ex.17. Repeat the example 16 for 100 order filter with hamming window
using fir1 (defult is Hamming Window) command.
The only changes are:
n=100;
[B,A]=fir1(n,w,'stop'); 0.8
FIR filter B coefficients

0.7
frequency responce of 100 order fir1(defult is Hamming Window) band stop filter
1.4
0.6

1.2 0.5

1
0.4

0.3
0.8
0.2
0.6
0.1

0.4 0

0.2 -0.1

-0.2
0 0 20 40 60 80 100 120
0 5 10 15 20 25

Ex.18. Repeat the example 17 for 100 order filter with hamming window
using fir1 (defult is Hamming Window) command.
The only changes are:
[B,A]=fir1(n,w,'stop',Kaiser(n+1,4));
frequency responce of 100 order fir1 using Kaiser window with beta=4 band stop filter
1.4 FIR filter B coefficients
0.8

1.2 0.7

0.6
1
0.5

0.8
0.4

0.3
0.6
0.2
0.4
0.1

0.2 0

-0.1
0
0 5 10 15 20 25 -0.2
0 20 40 60 80 100 120
 [z,p,kc]=ellipap(n,Rpass,Rstop)
Rpass,Rstop are in decibels
Butterworth analog low pass filter prototype.
[z,p,k]=buttap(n)
» [z,p,kc]=ellipap(2,1,2)
n is order of the filter z=
» [z,p,k]=buttap(2) 0 - 1.1188i
z = Zeroes 0 + 1.1188i
[] p=
p = Poles
-0.0962 - 1.0517i
-0.7071 + 0.7071i -0.0962 + 1.0517i
-0.7071 - 0.7071i kc =
k = Gain 0.7942
1
Chebyshev type II filter order selection.
[n,w]=cheb2ord(Wpass,Wstop,Rpass,Rstop) » [n,w]=cheb1ord(.3, .4 ,10, 60)
Rpass,Rstop are in decibels n=
8
» [n,w]=cheb2ord(.3, .4 ,1, 6)
n = w=
3 0.3000
w = » [n,w]=buttord(.3, .4 ,10, 60)
0.3507 n=
» [n,w]=cheb2ord(.3, .4 ,10, 60)
n =
17
8 w=
w = 0.2869
0.3833

» [A,B,C,D]=zp2ss(z,p,k) ∗

A = x =Ax+Bu
-0.1923 -1.0561 Y=Cx+Du
1.0561 0
B =
1 1
0
0.9
C =
-0.1923 0.1290 0.8

0.7

D = 0.6

1 0.5

0.4
w=kaiser(n,beta)
0.3
n is order of the window
» w=kaiser(100,4); 0.2

» plot(w) 0.1

0
0 10 20 30 40 50 60 70 80 90 100
dct(X)
Discrete cosine transform
» x=[1 1 0 0];

» Dx=dct(x)
Dx = Note that; project must be written in a
1.0000 0.9239 0 -0.3827 reasonable format. With at least
» x2=idct(Dx) introduction and stating the problem,
x2 = methodology and theory, results with
1.0000 1.0000 0 -0.0000
figures and tables, conclusion,
appendix (all MATLAB codes+…)