[sdiy] SSM2164 state variable filter

[Neil Johnson]

Hi Tom,

Tom Wiltshire wrote:
> I'm playing with a SSM2164 SVF circuit again. I'm still not happy
> with it, or perhaps still not happy with my understanding of it.

What's your starting point?  There are already some 2164-based SVFs on the net:

mine: http://www.milton.arachsys.com/nj71/index.php?menu=2&submenu=0&subsubmenu=1

Roman: http://www.sowa.synth.net/modular/vcf.html

Sam: http://houshu.at.webry.info/200602/article_3.html

> There are several things that I've been told/learned which I have no
> particular reason to doubt:

Ooohhh.....ok......(danger will robinson danger)

> 1) The circuit is essentially an oscillator, given the feedback from
> the LP output.

No, its a filter who's poles you can push towards the imaginary axis.

> 2) The "resonance" path from the BP output actually controls the
> damping by cancelling the oscillation. This is why the control works
> "back to front".

Actually, it pushes the poles *away* from the I-axis.

> 3) Some type of limiting or clipping is required for the oscillation to
> be stable and low distortion.

Generally yes, as the response tends to infinite amplitude, and we
don't have op-amps that extend that far!

> 4) The best place for such a network would be in the
> resonance/damping BP feedback path.

Nope. Utter "bulls reproductive organs".
Because the limitation is needed when there's little to no damping.

> Now, it seems to me that (4) doesn't fit. If the circuit is oscillating,
> there's no signal (or very little) going back along this path, so I can't
> see why a clipping network would do anything at all. On this basis, I'd
> be better off putting the clipper in the LP feedback path.

Yes.  And see examples linked above for real-world illustrations.

> I also don't understand why having no clipping network doesn't
> achieve the same thing but at a higher voltage level. If left without
> any clipping, the circuit will oscillate and the output starts to crunch
> into the rails, producing a clipped sinewave at the level set by the
> limits of the op-amps. Why does having a clipper with a lower
> threshold make any difference at all? Why aren't the two clippers
> (overdriven opamp / clipping diodes) equivalent?

Because you're operating outside of 'normal', the maths no longer
applies, all bets are off.  When you limit to a lower voltage the
op-amps and integrators are operating normally so the system behaves
itself.

> Is one of my "facts" wrong? If so, which one? Am I right that putting
> clipping in the LP feedback path is a better idea?

Try putting the limiting in the LP path.

> I've tried many variations on the breadboard, but at the moment I'm
> not getting reality and theory to match up at all. That and there's too
> many options.

Try some of the proven designs, linked above.  Then go from there.

Cheers,
Neil
-- 
http://www.njohnson.co.uk