VCA chips (2)

[jhaible]

> I think the 3280 claims matched gm.
> Does the 13700 also ? I don't know.

The 13600 / 13700, too. 3dB tracking, if memory serves.
That's the major (if not only) advantage over two 3080's:
You can just short pins 1 and 16 together in stereo vca
or SVF VCF applications.

> By the way, what's the difference 13600 13700?
> Better output darlington? I think for new designs 13700 is recommended.

Not better. Different.

The 13600 gets the Iabc of the OTA mirrored into the emitter of the first
transistor of the darlington buffer. This decreases input impedance of the
buffer,
but makes the impedance track to the gm of the OTA. Good for VCF stuff.
Extremely bad for vocoders, because you get crosstalk from the analysis to
the
synthesis section.

The 13700 has a plain darlington. Constant high impedance. Still not high
enough
for VCOs that shall also be used as LFOs, or for VCADSRS. You will often
find
the recommendation to use an opamp buffer instead. Personally, I go for a
discrete
FET in front of the darlington buffer when it is possible.

BTW, CA3280 and LM13700 are more different than one might think. The
13700's
linearization diodes are crap, because without real current sources you are
far
from linearizing the input stage. You can slightly improove this by using
resistors
to the negative supply instead of the resistors to GND (Keep in mind that
these are
not used to divide down the input voltage anymore ! They rather act as
return path for
diode bias current, and an impedance of 500R to 1k is quite ridiculous for
a current
source ...)
The 3280 is much closer to the ideal "Gilbert" case.

JH