Hi,
This seems like a dumb problem, but after way too much thought I can't see a solution. In general for a stable filter the denominator term is of the form 1+a1(z^-1)+a2(z^-2). And if the filter is stable a1 is between -2 and 2 while a2 is between 0 and 1.
The problem is that the input need to be in 16-bit Q15. Of course one would like to scale the coefficients (and you generally must for the numerator terms), but because of that 1+ term, any scaling of the a terms will result in moving the poles.
So net effect: how the heck can one use iircas5 or the rest of those iir functions? I can't see a way to get the values of that one coefficient to fit. I can't imagine that no one else has had this problem, but the one place I saw this question posted there was no real answer. Any help welcome...
Thanks,
Mark
