[sdiy] Reverse-exponential VCAs

[Jerry Gray-Eskue]

Hmm,

As I understand this sort of 4 stage setup, each stage phase shifts 90
degrees and the feedback loop is running a 360 phase shift, which is to say
in phase with the input. The per stage VCA you have could allow independent
control over each of these shifts, I am not sure what all the implications
are but I have a feeling this may give you control over harmonic content
that is not available in any similar filter. I suggest that you breadboard
this design with separate VC inputs for each stage and let us know how it
behaves with matched and independent control voltages. I think you may be on
to something really cool.

- Jerry

-----Original Message-----
From: David G. Dixon [mailto:dixon at interchange.ubc.ca]
Sent: Thursday, June 03, 2010 6:02 PM
To: 'Jerry Gray-Eskue'; Synth-diy at dropmix.xs4all.nl
Subject: RE: [sdiy] Reverse-exponential VCAs


> <<At this point, presuming a similar gain block on every filter
> stage, the filter will begin to self-oscillate. >>
>
> I guess I am missing something, I do not see a feed back loop in the
> diagram.  I believe that self oscillation requires a positive feedback
> loop.

This is just a tiny part of the whole filter circuit.  One of these hangs
off of every filter stage, which also includes a separate VCA (as a VC
resistor), an integrator, and an inverter (which is the source of the two
signals of opposite polarities).  Hence, the entire filter cascade involves
8 VCAs (two 2164s) and 12 opamps, not counting the various auxiliary circuit
bits.  The output of the fourth gain block feeds back to the first filter
stage, as usual.

The difference is that every filter stage has a resonance amplifier, not
just the fourth stage, and the resonance VC is "reverse-exponential" rather
than linear (or even exponential, as in the case of Neil Johnson's 3P SVF).
I've only shown one of the resonance amps from the first three stages.  The
fourth one has a slightly different arrangement of resistors.  All four
resonance amps generate a gain of sqrt(2) at the onset of self-oscillation,
which gives the requisite overall gain product of 4 (sqrt(2) to the fourth
power).

>
>                                V_c
>                                 |      |--[ 13k7 ]--|
>                                 |      |            |
> (+) Signal --> [ 33k0 ] --> 2164 VCA ----> Summer ----> Out
>                                        |
> (-) Signal --> [ 33k0 ] -------------->|
>       |                                |
>       |------> [ 13k7 ] -------------->|