Moog VCF Transistor Matching Question

[Haible Juergen]

	> I couldn't hear any difference in the 
	>sound of the filter, and my ears are pretty good.  However, some
people 
	>insist that the filter works best when all of the devices in the
filter are 
	>matched to each other.  You'll probably have to go through around
100 
	>transistors to find ten that match, though.

Which is more or less exactly what I did when I built my Minimoog clone
back then, because I didn't know better. (And you have to keep
your room temperature constant while you're testing these 100 transistors,
and you have to wait 2 Minutes for the transistors to settle to room
temperature
after you've touched them with your fingers. That was a whole evening with
a tiny hourglass next to the lab equipment ...)
But today I wouldn't do that again, because I think Michael is right.

Just look at the possible sources of errors: 

Ladder balancing is just the lowest and uppermost pair. For low
frequencies all the other stages are just cascodes.

Slight mismatch between the 2 transistors of one "middle" pair *might*
change the type of distortion around the cutoff frequency *slightly*,
but that's with full emphasis on "might" and "slightly", and even when
there would be a difference it would be filtered out by the following
stages.

Mismatch from one ladder rung to the next - that would be the same as
an little pole spread, just the same effect that is caused by the capacitor
tolerance. Which is what - 1% ? 5% ? 10% ?
Self oscillation will not start at the (theoretical) 12dB feedback loop
gain,
but slightly later. Now that's what the trimpot is for, I guess.

JH.