[sdiy] Later MS-20 filter with 2164

[David G. Dixon]

Further to my previous postings...

I've just simulated my 2164-based MS-20 filter clone, and I am shocked at
how beautiful the simulations are!  The filtered waveforms look fantastic.
And, with no limiting diodes or anything, this filter oscillates just
beautifully!  (This is very different from the state variable filter, which
hits the rails hard at oscillation, creating trapezoids.)  At a feedback
gain of exactly 2, the oscillations are stable and perfect.  The sine waves
are +/-6.7V from each filter separately (when one is oscillating and the
other is at zero resonance).  They will be very easy to trim.  If the first
filter is oscillating, the second filter will give identical sine waves 90
degrees out of phase at a feedback gain of exactly 1.  THDs are in the
neighbourhood of 0.03% to 0.04%.  If the gains are cranked beyond these
limits, the sine waves distort, looking more like sharks' teeth, with larger
amplitudes, but nowhere near hitting the rails (!).

I wouldn't want to put any limiting diodes in the feedback loops of these
resonance gain cells, or if I did, I'd put them on a selector switch, or use
back-to-back 6.8V zeners (which would clip at about +/-7.5V) so as not to
lose those perfect sines.

I think this might be the nicest filter simulation I've seen.  I'm really in
shock.  Design is not supposed to be this easy!  I'm definitely building
this filter.