AW: RE: PCB FOR MAT-0

[brad.sanders at circellar.com]

Subject: Re: PCB for MAT-04 VCA

I looked at national's site, and couldn't find AN-105. I had written you 
a reply asking about this, but just a few minutes ago found a later 
message from you in which you mentioned ANALOG DEVICES an-105! Ayyye 
Carrumba!

So, I finally got a look at the circuit, just now. It looks interesting, 
but much too complex for what I'm wanting to do. Anyway, it seemed a 
shame to toss out the rest of my reply to you, so...

F>     3. Signal inputs above around 8V p-p do not cleanly limit or
F>     clip, but  rather get a very nasty distortion as the top of the
F>     waveform gets  "cut and pasted" into the middle!

This sounds very much like a case of opamp nasties. With many opamps, if 
you exceed the common mode input voltage AND input current capabilities, 
you'll get this waveform "folding." Perhaps a better opamp? If you're 
using FET input opamps here, I'd be willing to bet that's the problem...

F>     4. When the CV travels too far in the negative direction some DC 
F>     component does affect the output (a simple diode may prevent
F>     this). 

Again, this may also be a function of the CM range of the opamp you're 
using (or mismatch in the 30K resistors). This sounds like a neat amp to 
try better parts in. One of those newfangled opamps with higher 
overvoltage tolerance might be very useful here...

F>     Chris Crosskey had an interesting thought when he told me that
F>     perhaps  it would be a good idea to split the VCA into two
F>     categories - CV-modulating VCA's, for control of DC control
F>     signals and some audio, and audio-only VCA's to allow for clean
F>     and noiseless amplitude moduation on the hi-fi side of things.

This sounds very useful for my digital osc/mixer/env generator chip.
It saves several clock cycles generating a linear slope ADSR 
rather than having to lookup a transfer function. If the log 
function is inherent to the VCA I use at the output of my synth module, 
then the only place I have to do lin/log conversion is in sending 
modulation sources to the digital oscillators. Since I have four 
oscillators and eight ADSRs on a chip - and only 25 microseconds/sample 
or so to do all this (and I'm NOT using a DSP!) -  those five or six 
clocks per device can add up REAL fast. Save five clocks per, spread 
over four ADSR generators per sample period, and I free up 2.5uS out of 
25uS!

Now, what's this about the electronotes gizmo? How would you rank the 
3080 based vca against more attainable devices, like the LM13700? 
I'm still just getting all this together, and I'd like to think I'm 
doing it right the first time... Of course, one usually never does - but 
I'd like to *think* so.