.
> > Is it the fact that the wave no longer looks anything like a
> > squarewave, and that a VCA (obviously)shouldn't do that?(especially
> > not a 380 dollar one)
>
>
> Er...that's my ∗point∗!! :)
>
> Maybe JH knows a good answer.
Hmm, I will only comment on other's designs when I know a bit more about the
circuit.
One idea did strike me, however: In the common debate about "clean
and crisp" vs. "smooth" (attributes that are often adressed to Serge vs.
Moog,
just to name two examples) it's not necessarily the "smooth" circuit that is
adding something (like signal rounding, THD), but it might as well be the
"clean" one which shows the more serious artefacts.
> I do know that the CA3280s used are kind of strange.
But they are working well enough in Rev. 3 Prophet 5's.
> Bernie Hutchins in Electronotes called them ' the most slippery ICs ever
> designed", meaning
> they have many undocumented...errr...."quirks".
I remember that - and it was Serge himself who helped to shed some light
on the then new LM13700 and CA3280 OTAs. If memory serves Serge
offered an explanation for the seemingly wrong design equations in
preceding EN issues. So in the context of that time and that EN
thread, it's fair to say that Serge was ahead of the others in understanding
that chip.
Not that ∗all∗ his conclusions would have been right, though. He also
proposed a scheme for the 13600 that allegedly would approximate
the current sources in theory (and resistors in real life) from the data
sheet better than the data sheet's own application circuits. The goal
was to improove the linearizing effect of the diodes. I once built
a circuit based on Serge's verbal description, and in fact it does
not improove the linearizing effect, but really performs on over-
compensation, thus bending the nonlinearity in the opposite direction.
It's J�rgen Bergfors's merit to find this out. I had published a
Morphing Programmer that used this circuit, and J�rgen measured it
and prooved it wrong.
Disclaimer: This is not a comment about Serge's production circuits, just
about a casual remark he had made in that EN article.
And that's all just good for anecdotes, IMO. If there really is a stability
problem with that particular Serge VCA, I doubt that it's a CA3280
problem.
> they have many undocumented...errr...."quirks".
The biggest "quirk" here in Europe is that CA3280's are practically
unavailable, and if so, just for a high price. That's why I stick to the
LM13700 family for simple applications, and lately I started to use
the BA6110 which apparently combines the benefits of the 3280
(real current sources and low noise) with the buffer of the 13700.
JH.