[sdiy] Re: OTA comparisons, VCA shootout

[JH.]

> What I am wondering about in all of this OTA discussion, is about how
> transistor OTAs fare. I have seen schematics and heard critiques of
> some using hand-matched transistors, and the RCA/Harris arrays. Hardly
> any though of some of the transistor packages which sound like they
> were made for this purpose, like the THAT and MAT0X arrays. Seems to me
> like an MAT04 would make for a decent and inexpensive VCA. Anybody have
> any opinions of these to relate?

If you build OTAs from matched transistor pairs, you normally get much
better
results than with integrated OTAs. It's just much more expensive, too. And
taking much more space.

If you's want to build a CA3280 completely, with all the current mirrors and
linearising diodes, from transistor arrays ... just add up the costs and
find
it way too expensive for most applications.

That was the bad news.
The good news is, that for most applications, you don't need all the extra
stuff that's inside an OTA. Often, just a bare naked transistor pair is all
you
need, as this performs the multiplying function in any OTA; the rest is just
for interfacing. Like, giving the control current input a different
direction,
and making a single ended output with high common mode voltage range
for your output current.
If you get rid of these in your design, you will have a less comfortable
sulution
(i.e. you will have to put some days of work into a design of the best
interfacing
for your application), but you will also get rid of sources for errors, such
as
bad current mirror precision and CV feedthru that may come from this.

For service techs, a lack of integrated OTAs will be a problem. So stock
up on 3080's.

For synth-diy, the fun is in designing circuits instead of just using
combinations
of IC application notes, isn't it? So the fun will be increasing, at least
as long as
we can buy matched transistor pairs.

JH.