[sdiy] Replacing CA3080 with CA3046 ??

[Ian Fritz]

At 09:57 AM 4/9/05, Scott Gravenhorst wrote:
>karl dalen <dalenkarl at yahoo.se> wrote:
> >
> >--- Antti Huovilainen <ajhuovil at cc.hut.fi> wrote:
> >> On Sat, 9 Apr 2005, Tim Parkhurst wrote:
> >>
> >> > http://www.idg.se/personal/bergfors/bergfotron/VCA.htm
> >> > This comparison shows that actually the lowly 13600 makes a pretty darn
> >> good
> >> > VCA. This list includes the 3080 , the SSM2024, 3046 and 3086-based 
> OTAs,
> >> > amongst others. It doesn't include the 3280 or 5517 however.
> >>
> >> What is the reason for the higher offset of the LM13600 circuits compared
> >> to CA3080?
> >>
> >> Antti - Who will need some 40 VCAs for one project...
> >
> >The thing with the 13600 and 700 is that the offset is not constant,
> >it responds in a somewhat eliptical fashion depending on varying Iabc,
> >from some 500uV to 25mV and back, i tested this long time ago about
> >12-15 13700 they all shows this in varying amount. Used the classic PNP
> >Iabc source expo.
>
>That's interesting.  Is there an active method to compensate, like using one
>OTA with 0 volts on it's inputs, but varying it's Iabc the same as another OTA
>to generate a correction voltage to cancel the offset?
>
>Or would this be useful only if the other OTA is part of the same package as
>the compensator?

I recently looked at this offset problem quite closely.  I made a discrete 
OTA with high quality parts, but found I couldn't do any better that the 
CA3280.

The offsets for the 13600's that I looked at were all over the map -- even 
between the two units on one chip.  This will make your compensation idea 
difficult, I'm afraid.

I made a partial offset compensation in my chaos generator, but it works on 
the loop as a whole, so it isn't a general solution.

   Ian