[sdiy] SSM2164 state variable filter

[Harry Bissell]

As the filter sweeps through the harmonics of the input waveform,
you get a characteristic 'sound'.  Lower harmonics have higher
amplitudes than higher ones, and the higher harmonics (of course) are
closer together.

If your input signal amplitude is near the zener voltage, this will
effectively 'compress' the amplitude of lower harmonics selectively.

This might sound 'good' ot 'weird' depending on your preference.

Of course if you are most interested in an oscillator, the zener should be
chosen for the amplitude you'd like.

Its a BIG trade-off. Higher zener voltages sound more like a classic SVF,
lower ones like an SVF followed by a compressor.

The idea of pre and post filter amplitude is really worth consideration.
The sounds are drastically different...

even more so if its ~not~ an SVF, but an MS-20 type filter :^)

H^) harry



----- Original Message -----
From: David G. Dixon <dixon at interchange.ubc.ca>
To: 'Harry Bissell' <harrybissell at wowway.com>
Cc: 'Tom Wiltshire' <tom at electricdruid.net>, 'sdiy DIY' <synth-diy at dropmix.xs4all.nl>
Sent: Wed, 14 Jul 2010 16:10:30 -0400 (EDT)
Subject: RE: [sdiy] SSM2164 state variable filter

> The zener clipper (as described) also serves a s a Q limiter when you are
> not
> oscillating. This will cause an apparent reduction of resonance if the
> input
> signal amplitude is large, but will not be noticed when the input
> amplitude is
> small.

Yes.  This is why I suggested a zener voltage somewhat larger than the
signal level -- 6.8V vs 5V.  One could perhaps stand to go even a little
higher, but one wouldn't want to go too high lest the LP and HP outputs
start clipping against the opamp rails at high Q.  It's a compromise.

> I used the same technique in the SVF filters in my guitar processor
> (Muffy). If you
> accidently clip the filter with hard playing or the filter locks on the
> fundamental,
> it sounds like hell. (not good hell)

If the filter locks the fundamental, then admittedly the BP output would be
pretty distorted at high Q.  However, the LP and HP outputs should be smooth
as silk, no matter what.

I suppose one could attenuate the signal input and amplify the filter output
to drop the signal level through the filter, but this would generate sine
amplitudes much larger than the input signal during self-oscillation.  It
seems to me that it is still best for the filter to have an overall gain
factor of 1, and for the zener voltages to be in the same ballpark as the
input signal level.


-- 
Harry Bissell & Nora Abdullah 4eva