[SciPy-User] scipy.signal.butter function different from Matlab

[braingateway]

gwenael.guillaume at aliceadsl.fr :
> Hi,
>
> I'm trying to rewrite a Matlab code in Python and I encounter some difficulties
> using the butter function.
>
> In Matlab, my code is:
>
> % Parameters:
> N=5;
> F_inf=88.38834764831843;
> F_sup=111.3623397675424;
> Fs=32768.00013421773;
>
> % Filter coefficients computing:
> [z,p,k]=butter(N,[F_inf F_sup]/(Fs/2));
>
> % Result:
> z=[1;1;1;1;1;-1;-1;-1;-1;-1;]
> p=[0.999020109086358+0.021203989980732i;...
>    0.999020109086358-0.021203989980732i;...
>    0.99789313059316+0.020239448648803i;...
>    0.99789313059316-0.020239448648803i;...
>    0.99762168426512+0.018853164086747i;...
>    0.99762168426512-0.018853164086747i;...
>    0.998184731408311+0.017659802911797i;...
>    0.998184731408311-0.017659802911797i;...
>    0.999249126282689+0.017020854458606i;...
>    0.999249126282689-0.017020854458606i;]
> k=5.147424357763888e-014
>
> In Python, the code is quite similar:
> import numpy as np
> import scipy.signal as sig
>
> # Parameters:
> N=5
> F_inf=88.38834764831843
> F_sup=111.3623397675424
> Fs=32768.00013421773
>
> # Filter coefficients computing:
> z,p,k=sig.butter(N,np.array([freqmin,freqmax])/(Fs/2),output='zpk')
>
> # Result:
>  Error message:
> C:\Python26\lib\site-packages\scipy\signal\filter_design.py:221:
> BadCoefficients: Badly conditionned filter coefficients (numerator): the results
> may be meaningless
>   "results may be meaningless", BadCoefficients)
>
> z=array([], dtype=float64)
> p=array([
> 0.99464737+0.01603399j,0.99464737-0.01603399j,0.98633302+0.00982676j,0.98633302-0.00982676j,0.98319371+0.j])
> k=4.2522321460489923e-11
>
> Does anyone could help me to understand why the results are not the same?
>
> I have also another question. In Matlab, after having computed the filter
> coefficients, I apply the filtering to my signal X using the following code:
>
> % To avoid round-off errors, do not use the transfer function.  Instead
> % get the zpk representation and convert it to second-order sections.
> [sos_var,g]=zp2sos(z, p, k);
> Hd=dfilt.df2sos(sos_var, g);
>
> % Signal filtering
> y = filter(Hd,X);
>
> Is there any Python functions that could do the same?
>   
No, there is only naive implementation of filter(). As I know, there is 
no SOS convertion exists. The Scipy.signal is only for very very simple 
signal processing. In this state, you either implement your own SOS 
system convertion code and then use concatenated Scipy.lfilter() call to 
achieve your go, or you stick to matlab.

LittleBigBrain
> Thanks
>
> Gwenaël
>
>
> _______________________________________________
> SciPy-User mailing list
> SciPy-User at scipy.org
> http://mail.scipy.org/mailman/listinfo/scipy-user
>