[sdiy] Feedback in filters [was: Re: From a commercialstandpoint...]

[Richie Burnett]

> Oh, like positive or negative feedback on a flanger.

Exactly.

> So, if it oscillates at roughly half the cutoff frequency would that give 
> me a big spike in the passband, then a normal rolloff?

Sort of, but I'm not sure whether you will visibly or audibly notice just 
this feature, because the 8th order filter will have a steeper rolloff too 
so is going to sound different anyway.

What will alter is the perceived pitch of the self-oscillation relative to 
the perceived brightness of the open-loop filter.  A dramatic example of 
this effect is the TB-303 filter...  It's three dominant poles that are 
spread over several octaves result in a self-oscillation frequency with 
feedback that is almost a decade higher in pitch than the -3dB point of the 
open loop filter!  That's why it always feels like the range of the cutoff 
pot doesn't go low enough when you've got the resonance cranked up, but then 
it doesn't seem to go high enough to get the filter fully open when the 
resonance is turned right down to zero.  (The lowest frequency pole 
determines where the rolloff starts and where the -3dB down point is, but 
the self oscillation can't happen until you at least get to a frequency 
above the 2nd pole because it needs 180 degrees of phase shift to happen.)

> I can feel some experiments coming on, this weekend...

Have fun!  You might need to introduce a DC blocking capacitor somewhere in 
the feedback loop, otherwise putting positive feedback around the loop will 
probably just cause the bias point to integrate towards one of the rails and 
stay there.  Putting a zero at DC will break the loop for DC content and 
stop this from happening.

-Richie,