[sdiy] 2 queries -- SVF self oscillation

[Magnus Danielson]

From: Scott Gravenhorst <music.maker at gte.net>
Subject: Re: [sdiy] 2 queries -- SVF self oscillation
Date: Sat, 03 May 2003 08:03:24 -0800

> patchell <patchell at silcom.com> wrote:
> >
> >
> >Gavin wrote:
> >
> >> hey all,
> >> i've got 2 totally unrelated ?'s
> >> 1. anyone know have understand the tuning procedure for the yamaha CS-40m.
> >> I know there are some CSersout there!
> >> 2. AmI correct in thinking that in the basic state variable filter (2 3080's
> >> etc.)  MORE resistance in the resonance feedback path equals MORE resonance,
> >> LESS resistance equals LESS resonance?
> >> thanks
> >
> >    Yes, in a standard SVF, more resistance (which corresponds to less gain)
> >decreases the damping (which is what that feedback path really controls), or
> >increases resonance.
> >
> >    If you use an OTA for the resonance control...this translates more current,
> >more damping (lower resonance) less current, less damping (higher resonance),
> >zero current = oscilator (well, in theory anyway).
> 
> Ah yes, "in theory"...
> 
> I've always wondered about an SVF and self oscillation.  I have 2 FatMan synths
> which use an LM13600 SVF which looks pretty much like the schematic in the data
> sheet.  I've never been able to make either one self oscillate, even when I
> break the feedback path open (unsolder one leg of the rez pot).

The SVF feedback term works "in reverse" to what you expect from "classical"
filters like the Moog ladder where greater feedback results in greater
Q/resonance. In the SVF the closer to 0 feedback from the middle section, the
closer to resonance you get. It may be hard to reach for exact zero, so if you
reach for slight sign reversal you should get there.

This feedback term sets the real-axis position of the pole-pair, and when the
real-axis position is zero you have resonance, if it is positive you get
accute acceleration of amplitude (which basically drives you into self
resonance just as our SVF-building friend got when he used by schematic with
a polarity error in the feedback term).

So, for those that want their SVFs to be able to go into self-resonance or darn
near it (at their choice of control naturally) you should really pan between
the feedback signs. Since the input summer has both polarities of input, it
should be a simple thing to make use of this fact and set things up. The exact
form can be elaborated, but it would all be a bunch of resistors in some
suitable form. One trick could be to fix the feedback of one polarity and
change the feedback of the other, this might solve the inverse log pot problem,
I don't know.

In the schematic I have up, you would then toss a fixed resistor from the BP
output to the positive input of the input summer and keep the variable feedback
on its incorrect position (do take a failure and make the best of it!).
This fixed resistor should maybe be a trimmer, to set how deeply into
self-resonance you can get.

I haven't tried it, but it should work.

Cheers,
Magnus