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

[rburnett at richieburnett.co.uk]

> So, if I made an eight-pole filter would I need to feed the resonance
> back in phase with the input then?

Yes.  45 degrees phase shift from each of 8 stages for a total of 360 
degrees around the loop.  So no inversion required.

(It *would* actually oscillate if the feedback *was* inverted too, but 
not at the cutoff frequency of the individual poles.  If you introduce 
inversion in the feedback path, then it will oscillate at whatever 
frequency the total phase shift through the filter adds up to 180 
degrees.  So for an 8-pole cascade each stage would contribute 22.5 
degrees at about 41% of it's cut-off "knee" frequency.  That's where the 
filter will self oscillate with enough feedback around the loop.)

> Let's leave aside how high the positive rail would need to be, for all
> those diodes :-D

LOL. You could use a cascade of OTA stages instead, like two cascaded 
IR3109s!

-Richie,