Does anyone have a working example of the IIR filter block in GNU
Radio Companion that I can see, and the filter designer app, table, or
equation which generated the coefficients?
The GRC documentation states “fftaps and fbtaps must have equal
numbers of taps”, however biquad IIRs traditionally have three B
coefficients (feed forward) and 2 A coefficients (feed backward). This
conflicts with the ‘Direct Form 1’ shown here:
Is b0 already implied to be set to 1 automatically, with the
expectation that I would normalize all values to b0?
Or, do I need to specify b0, then for the A coeffs put a zero as a
placefiller?
Negating A coeffs doesn’t appear fix the issue in either case.
Also, I noticed that if I set an unequal number of coefficients (in
violation of the docs) there still is no error. (it still doesn’t work
though)
Does anyone have a working example of the IIR filter block in GNU
Radio Companion that I can see, and the filter designer app, table, or
equation which generated the coefficients?
Thanks for the example links, they were very helpful. There were two
things I was doing wrong: First I didn’t use square brackets for the
sets, and to use the biquad calculator featured on earlevel.com, the B
coefficients needed to start with a 1.0 filler and the rest negated
like this: