VC envelopes

[Joachim Verghese]

Juergen wrote:

> (And wouldn't the CV feedthru be filtered by the ENV's time constant?)

Yes, but when you vary the release time you can see the envelope
output fluctuate around the steady-state release asymptote.
Not a very dramatic effect, but a bit disturbing nevertheless.

> Other / better OTAs:
> Yamaha have built their VC envelopes in the CS synths around 
> a special chip that only containd the env *logic*, plus an external VCA.

Yes, I noticed that. Those VCAs are probably a lot better than
the 3080. And the env "logic" chip probably includes expo converters
for the time parameter inputs too.

> As far as I see it, the problem with the 3080 is that the offset voltage
> varies with the amplifier bias current, so you can only compensate
> for one point.

Exactly.

> Discrete transistors can be much better here. My
> BC550-based VCAs have much less CV feedthru than average 3080's,
> so maybe a VCADSR would also profit from a discrete OTA.

Good idea. But then you would have to select components,
which is a bit cumbersome if you intend to build a large
number of synth voices, such as in a polysynth.

> With "a la Oberheim" you mean the CEM3310 ? 

Yes, or the preceding discrete circuit, as described in one of
the Synthesource issues.

> Excellent solution, but a lot of transistors involved.
> Stupid question: where would you still need the 3080 then?

Lots of matched transistors -- I just thought that the 3080 would
be useful since it can both sink and source current. You know,
trying to avoid PNP current mirrors, because PNP arrays are not
widely available.

> Special Award for "most different" VCADSR solution goes to Roland's
> JP-4. Variable resistors are emulated by analogue switches and smoothing
> capacitors, and each switch is controlled by a little expo VCO.

Great! Never heard of such a design before.

Thanks to Martin and Juergen for the replies. I'll give the problem
some more thought.

-joachim