Moog VCF (TomG's VCF2B)

[Martin Czech]

> Anyone used a LM13700N instead of the CA3080? Or any other chip for that
> matter?
> 
As I have finished some clones, I will.  For the clones I think the 3080
distortion should be kept (if the clone is in heavy distortion this
makes a difference). A clone is a clone.

Of course I'll use the anti-distortion network (Gilbert). Very often
this is not used, I think people had problems understanding the data
sheet (the ca3280 data sheet being worse). I remember an E.N. discussion
about exactly that fact by S. Thernepin. 

This Gilbert anti-distortion makes a low input impedance for the ota.
I'll not use the darlington buffer anyway, but instead an opamp integrator
or follower. This has severall advantages, one of them beeing that the
lower input impedance is not important.

I think the main reason for not using the more modern otas is simply
history: many circuit concepts are very old (which doesn't mean that
they are bad), at that time only the ca3080 existed. From then on there
where just copies of copies...

As J. Haible pointed out, the 3080 is ok if you have low dynamic input
signals, e.g. for waveshapers, filters with vco input, vco mixers.

OTOH if you want a filter to deal with any high dynamic input (CD,
DAT, whatever) you will surely want something better, in terms of S/N
ratio. I'm not sure if the 13700 is good enough in these situations.

Perhaps I'll take again the SSM2018 route, I allready made an attempt but
this showed nasty dc-offset problems due to a stupid error of mine.
But the S/N was ok. The idea was to implement a fully balanced
state var ciruit, using the balanced design of the ssm chips.
The 2218 now has current outputs better suited for filter applications.

It all depends on your recording stuff. If you're doing live performances,
you don't need to care much about S/N. If you're going to digital
recording with REAL 16 bits or even more (how many ADCs claim to have 16
bits and have an effective resolution of 13-14 bits in an usual studio
environment, this means including rather long unbalanced cables etc
...) things are much different. Especially if you want to run a final signal
through your analog vcf. Any noise/artefacts will NOT be masked in this case.
You might cure this in the digital domain afterwards by using multi band
noise gates, but I think this is the wrong way. Better record a clean signal.

m.c.