Moog Music Forum

In the process of calibrating my Moog 55, I abandoned for the moment the early 911 EG problem (see my previous post).
Now facing the 902 calibration, I have seen that the procedure is quite tricky. The major difficulty is with the DC output bias. After the whole calibration process completion, as described in the schematics (in the first step I got about 0 V DC from the outputs), I find about a hundred mV DC in the VCA output, that is a problem in some situations, like when using it as a modulation VCA (eg. delayed vibrato), in that it alters the basic tune of the VCO. I see that, also when I finally readjust the output bias trimpot to 0 V, the voltage tends to fluctuate... Is there any modification or critical components to check? I replaced the three electrolytic caps, but nothing changed.

902 VCAs can be a b*tch to calibrate sometimes.
It helps to take their + and - outputs and feed them to a dual trace scope.
You don't actually need a meter for them if you have a good scope.

But there can be a lot of drift and offsets.
They are more temperature sensitive than the oscs. (!)
Just plug one into a scope or meter and blow a little cold or hot air their way and you'll see a change.

Each 902 is different (and there are about 5 revs too), so each must be evaluated with consideration to their intended purpose.
To a certain extent, one must either choose to calibrate them for low offset on one output (normally +) or calibrate them for balanced + / - outputs and accept whatever offset you get.
Sometimes one achieves everything correct in terms of one output, only to have the other suffer in one way or another.
This is especially true when going from expo to linear operation or vice-versa.
If one truly wants a 902 that's good in every way (modes, offset, level, balance) resistor substitution may be necessary.

Thanks Kevin, your explanations are always very exhaustive. As to the replacement resistors, are you referring to the couples near the transistors like Q8/9 and Q11/12?

There are 3 sets of resistors that should theoretically be matched.
This based on the assumption that the transistors they're feeding are matched.
They're the 100 ohm, 150 ohm and 15 ohm pairs.

But if any of these components (transistors or resistors) have aged or drifted, correction must be provided with their respective trimmers.
However these trimmers generally adjust balance and not offset.
So one could get two transistors of a current mirror balanced, but still have an offset.
This is where changing the resistors can improve things.

It's not easy work tho.
One should have an accurate resistor substitution box, a meter, scope and ideally be able to gauge the transistors also.

Sorry, but this is as much help as I can provide via text.
Trying to help tho.

Current mirrors with decent specs are a lot harder to implement on the discrete level than on the substrate of ICs. That was the best technology had to offer back in the 1960s.

And then there's some expo conversion going on in the 902 circuit - without temp compensation!

thats good info from Kevin. I have 3 cloned 902's I made myself and yes they are tempramental....some days fine other days iffy.
A particular problem I had was "thumping" (with a fast attack EG) and ouput noise but trimmer tweaks help fix this...but back it comes sooner or later.
But I will try matched transistors and the mentioned resistors as well in one of the modules and see if theres any improvement.