IIR design and stability

>"Sam (r&#39982; sans -no-sp-am)" <totalsam-no-sp-am@hotmail.com> wrote in
message
>news:3ff754cd$0$726$5402220f@news.sunrise.ch...
>> Hi all !
>>
>> Thanks for reading ! I would like to design a 1st order digital high
pass
>> filter with a 50 us time constant. The analog transfer function is
>> Ha(s)=50E-6*s/(1+50E-6*s).
>>
>> Then I use the bilinear transformation. Sampling freq is 48 kHz. The
>> pre-warped pulsation is 0.4228 rad/sec and the z transfer function is
>> H(z)=236.4E-6*(1-z^-1)/(1+0.999527*z^-1).
>>
>> When I compute this filter, it seems to be unstable at 24 kHz, and the
FFT
>> of a filtered signal gives a spike at this frequency.
>>
>> Is something wrong in my design ?
>
>Yes, but not in your calculations for the digital fitler.
>
>> How can I avoid having this oscillation ?
>
>You need to plot the results for a 20khz signal for 1 millisecond to see
how
>the sampling is affecting your results.
>
>> Putting a low-pass filter after this one seems not to be a good idea I
>think
>> ...
>>
>> Who could help me ??
>>
>> Thanks a LOT in advance !
>>
>> Sam
>>
>>
>You are right.  Try putting the analog low pass filter before the
digital
>filter.
>Check out anti-aliasing.
>
>Peter Nachtwey
>I suggest to introduce a ZERO at this point, this hopefully will cancel
the effect of oscillation.
regards
al  
>
>