[sdiy] Irwin dual VCA--any interest?

[René Schmitz]

Hi Charlie and List,

synth at charlielamm.com wrote:
> What I found empirically:
> 
> You build an ADSR using the same cap n 4066 arrangement a lot of
> designs use (there are 2 on my site that do, for instance).  Since the
> voltages at attack and release rise in a linear fashion over time, I
> consider these to be "linear" devices.  Someone should correct me if I am
> using this terminology wrong.
> 
> You set the decay to be 5 seconds, from the end of the gate, to when you
> want, say -100db (ie., silence).
> 
> For argument's sake, lets say that 5V CV equals 0db (no boost or cut in
> amplitude) when plugged into the VCA's modulation in.
> 
> In 2.5 seconds, you have 2.5V at the ADSR output, and you want the VCA to
> be -6db down.  In my mind, this means you have a "linear" ADSR, and so
> you want a "linear" VCA.

I disagree, with the linear decay you need (IMO) an exponential VCA. 
(I.e. one that responds in dB per Volts) And with an already exponential 
decay, you need a linear one. (Responds with a gain factor that is 
proportional to the voltage. AKA Two quadrant multiplication.)
THe major difference in the two combinations is what happens in the 
attack phase. Exponential EGs do have a convex shape, while the 
exponentiated linear shape will result in a concave curve.

> As I understand it, and I may be wrong and get cut to shreds by the group,
> an "exponential"  VCA won't do this, since it's set to some fixed V per
> DB, it will probably be way more than 6db down after 2.5 seconds; for
> practical purposes, you won't hear anything after a much shorter time than
> this.

Natural instruments (i.e. plucked strings, drums) decay exponentially.
So linear slopes with linear VCAs don't sound natural.

> Hence a lot of folks say expo VCA's are "good for percussion" but I think
> this assumes the sort of cap n' 4016 ADSR I just described.
> 
> The Irwin VCA, by my tests, works that way.  With a REF IN of -5V, if you
> send in 2.5V positive CV, the output is about -6db relative to where it
> was when the CV was 5V.
> 
> The important thing: When you hook it up to a standard, cap and 4066 type
> ADSR, it sounds "right".  The release fades out in a satifying, slow
> manner.

This one here http://www.charlielamm.com/synth/schematic.pdf
is haveing the exponential slopes. (When the switches are on, all that 
is in effect is an RC Stage with its usual exponential behaviour.)
So it sure sounds "right".

  Cheers,
  René