HOW TO DESIGN AN IIR Low PASS FILTER WITH MATLAB

I have asked this type of questions many times but can you give me the matlab code for designing a lowpass IIR filter.

thankyou in advance

Answer 1

[Wayne King]

Hi Raj, you can either use basic functions like butter(), cheby1()

For example:

Lowpass filter for data sampled at 10 kHz, passband below 1 kHz

 Wc = (2*1e3)/1e4; 
 [B,A] = butter(10,Wc);
 % view magnitude response
 fvtool(B,A,'Fs',1e4)

Or something like:

     [B,A] = cheby1(10,0.5,Wc);
     % view magnitude response
     fvtool(B,A,'Fs',1e4)

Or you can use the fdesign workflow. Using fdesign.lowpass

     Fs = 1e4;
     d = fdesign.lowpass('N,F3dB',10,1000,Fs);
     Hd = design(d,'butter');
     fvtool(Hd)

There are a number of specification strings for fdesign.lowpass that support IIR designs. After you specify a filter, you can use

    designmethods(d)

to see which design methods are supported.

[raj]

Thanks for your help an other question which is a continuation of the above question I want to design an IIR filter with only poles that is an ar process ?? How can I do it with MATLAB

[Fathima Bareeda]

how to pass this filter through a matlab in built sound signal splat

Answer 2

[Wayne King]

You can easily design an FIR filter, or use the impulse response from the filters above, and then use linear prediction on those coefficients (or impulse response). The problem you are going to have is choosing the order. It won't take a very large AR order at all before you start getting a peaky response, which isn't going to match your filter's frequency response very well at all.

   Wc = (2*1e3)/1e4; 
   [B,A] = butter(10,Wc);
   h = impz(B,A);
   A1 = lpc(h,2);
   fvtool(1,A1,'Fs',1e4);

Or starting with an FIR filter

 d = fdesign.lowpass('Fp,Fst,Ap,Ast',1e3,1.1e3,0.5,40,Fs);
 Hd = design(d);
 Afilt = lpc(Hd.Numerator,1);
 fvtool(1,Afilt,'Fs',1e4)

[raj]

If suppose I have an ar(p) process white noise and then add some signal components and then when I estimate the coefficients of the signal(signal + ar(p)process white noise) by getting the order using akaike criteria and rissanen criteria what can i expect I usually cannot estimate the white noise even after reversing the transfer generated by the coefficients by ar process

[raj]

If I want to perform least square estimation in frequency domain how should I PROCEED