[sdiy] Best place to get 3080s?

[Ian Fritz]

At 06:41 PM 1/2/06, harrybissell wrote:

>Making your own OTA is not trivial. You could make a VCA with an
>opamp and three transistors... (two of which should be well matched.
>
>A decent OTA would require many matched pairs... not easy to do
>in discrete.

But often you don't actually need a full OTA, since in many applications 
you don't need all the fancy current mirrors.  So, for example, you can 
make a really good VCA with (1) an accurate, symmetric Wilson mirror made 
from a MAT03 and a pair of 2N3906's, (2) a differential input pair from a 
MAT02 and (3) an opamp differential I/V converter fed from the MAT02 
collectors. Not cheap, but not difficult, either.

For filter (etc.) applications you can just feed one collector into an 
opamp integrator.  See the old Charlie Thompson designs in EN#62(12).

>Otoh, maybe I should but transistor pairs from you if you have that
>many, that well matched. one thing you can say for the IC process...
>the transistors may all ~suck~ but at least they suck in very very
>similar ways on the same silicon...

Well, similar, but often not similar enough.  For best performance you have 
to deselect OTA chips with large offset variations.

My recent research on this question showed that the cost comes out about 
the same using the MAT's or selecting well matched (<50 uV) monolithic 
pairs or selecting OTA's for low offset variation.

   Ian




>H^) harry