# Bandpass filter using firpmord

116 views (last 30 days)
Maria Stroe on 29 Mar 2018
Commented: Wei-Han Hsiao on 31 Jul 2020
Hello, So I have this bandpass filter with : 0 between 0 and 0.2*pi, 1 between 0.3*pi and 0.5*pi and 0 between 0.6*pi and pi. the passband ripple=0.01 and the stopband ripple=0.05 Sampling frequency=16kHz I have to use firpmord and I have no idea how to make the length of the f vector = 2*length(m)-2 Here is my code
m=[0 0 1 1 0 0]
f=[0 0.2*pi 0.3*pi 0.5*pi 0.6*pi pi]
dev=[0.05,0.05,0.01,0.01,0.05,0.05];
Fs=16000;
[n,fo,mo,w]=firpmord(f,m,dev,Fs)
Wei-Han Hsiao on 31 Jul 2020
1. First, you need to know that m (or so-called a) denotes the number of bands.
That is, m = 3 in your case since you have two stopbands and one passband.
2. Second, f is the edge frequency vecor that inherently removes the starting and
end points. Thus, 0 and pi in your f are redundant and should be removed.
3. Third, dev corresponds to m. Hence, your dev should have only 3 elements.
4. Last, f and Fs should have the same units. Use either radians or Hz.
So, under Fs = 16000 Hz, your codes should be modified as follows.
Fs=16000;
m=[0 1 0];
f=[0.2*Fs/2 0.3*Fs/2 0.5*Fs/2 0.6*Fs/2];
dev=[0.05 0.01 0.05];
Fs=16000;
[n,fo,mo,w]=firpmord(f,m,dev,Fs);
h = firpm(n,fo,mo,w);
figure,freqz(h,1,1024,Fs);
Hope it helps! ^^

Abhishek Ballaney on 30 Mar 2018
https://www.mathworks.com/help/signal/ref/firpmord.html