Compensating multi-stage feedback (was: RE: all tranny vca+ )

[jhaible]

Hi Larry,

just a short reply - I'm in no way anti-opamp. There are reasons to use
them,
even in the signal path, in a modular synth, when you want a module to work
for both audio and CVs. Audio normally doesn't need precision, but low noise
and a good overdrive characteristic. This is possible with discreete
circuits
or with modern opamps, but not with cheap opamps or discrete opamps
that are worse than that. (Both of the latter *can* be the object of choice
if we want to emulate the oddities of some vintage circuits, though, so even
these are an option sometimes.)
For CVs you want precision: Processing a keyboard output voltage that
controls VCOs calls for low offset and high linearity at low frequencies.
Modern opamps can do that, cheap old opamps can often do that as well,
but need trimpots for offset voltage / currents.

So for modules that should excel for both audio and CV processing, a good
modern low offset / high speed opamp is a good choice.

If audio processing *only* is required (as in synths with fixed topology),
IMO
discrete circuits without multistage feedback have a *tiny* advantage.
Maybe it's audible, maybe it's not. But one thing is for sure: if you want a
simple unity gain buffer, there are choices more reasonable than sending
the signal thru dozens of transistors with high nonlinearity and at least
one pole in the frequency response each, and then restore the overall
linearity with a strong feedback loop, and artificially decreasing the
frequency of one of these many poles to such a low frequency that it becomes
dominant and the FB loop won't cause oscillations. It just doesn't feel
right.
At best, it's overkill. At worst, you get unpleasant side effects, like
increasing the unpleasant artifacts when the feedback loop is out of work
for an instant, namely during a momentary overdrive.

JH.

> OK, I am new to this game and have read this discussion with interest.  I
> will have to claim that I was totally ignorant to the use of op amps until
a
> few months ago.  Now, I know just enough to make some op amp circuits
work.
> I have a couple  of references that I am using on the subject.  1 - the
> simple Radio Shack booklet and 2 - the very in depth "Art of Electronics"
I
> have built a few of these circuits for summing and inverting and have the
> basic hang of it.
>
> What is escaping me here is why the use of an op amp and negative feedback
> in the audio path is perceived by some to be a bad thing.  For example, if
I
> wanted to boost the level of a line level audio signal to use in my higher
> voltage modular audio path, seems like a simple op amp (one with the right
> characteristics for audio) would be easy choice.
>
> Without starting a discrete vs. op amp vs. tube all out onslaught, I would
> appreciate hearing individual reasons why one would choose to build a
> discrete circuit in the audio path as opposed to a good op amp.
Subjective
> reasoning is really what I am after.
>
> As always, thanks.  I don't post much here as my knowledge level is so far
> below the qualified ones on this list.  But, I read almost all of it and
> appreciate the great spirit of sharing.
>
> Larry Hendry
>
>
> ----- Original Message -----
> >100% agreed. I'd even go one step further: The complete audio path can be
> >done without negative feedback. (And hence no compensation.)
>
> From: jhaible <jhaible at debitel.net>
> That's what I intended to say. Been there, done that - the audio path of
my
> JH-4 synth is completely built that way. The only multi stage feedback
loop
> is in the VCFs, to get resonance.
>
>
>
>
>