active crossover: fir vs iir

[netchris]

i'm going with an active setup using a computer and vst software to do my crossover, and am currently planning on using a l/r vst plugin filter. it's my understanding that the disadvantage of iir filters is phase shift (90* every 6dB/oct of slope), and that fir filters have no such phase shift.

but what if you just use a 24dB/oct (or a multiple thereof) slope? that's 360* of phase shift and thus no phase shift -- correct? even if i were to use a 12dB/oct or 36dB/oct slope (180* of phase shift), couldn't i just reverse the polarity of the mid and end up with the audio back in phase? is there anything wrong with my thinking on this? ..or is there some other benefit of a fir filter over an iir filter.

thanks,


-chris

[Portlandmike]

All natural systems must have a phase shift when there is a change rate of change of slope.

IIR's have this natural change. It isn't a bad thing in itself.

Use IIR's to do your crossover IMO. Bruno Putsey did an AES paper on this recently.

You might play with using a FIR filter to restore the phase shift at the crossover, but you risk causing some wierd stuff happening if the system is not perfectly linear.

my two cents

Mike

[phase_accurate]

There is no need for using FIRs at all IMO. Phase linear digital crossovers can be done using the subtractive delay method (IIR lowpass and delay).
EQing of driver response can be done in IIR as well.

You can even achieve linear phase-response down to the cutoff frequency using the subtractive-delay method. But this would come at the cost of increased total delay.

Regards

Charles

[Portlandmike]

Yeah, what he said!