[quote author="PRR"] You can use a second "tracking" opto-R, and this works well enough as a meter (see LA-2). But it does not track well enough for gain control unless you get very carefully made cells, age them in light and dark, and then match them about 6 ways. Yield would be very poor: two that match the same dark may mis-match in likght, or half-light, or end up the same but attack and decay very differently.
We probably want a single-substrate opto-R cut into 4, 6, or 8 elements and laser-trimmed. [/quote]
Dual Vactrols are not so uncommon. These have a single LDR which is center-tapped. I've played with two types of these, VTL5C4/2 and VTL5C3/2 (from memory; must look up the exact numbers to be sure).
The two halves can have slightly different absolute values (I measured this), but I expect them to track nicely (memory effect and aging included).
I have _not_ measured this, mind you, but I expect this because it's the same, center tapped, element.
So you might be able to build a servo system with these, one half for audio, other half for sensing a DC for the control loop. There might be a problem with the (small) resistance from th ecenter tap that is shared by both elements. Haven't tried this. But I know that Moog have used these things for a compander system in their classic 12 stage phaser: One half for the compressor, other half for the expander. The control signal is quite unusual in this circuit: instead of the usual fast attack / slow decay they have fast attack / some ten milliseconds of "hold" / fast decay, which creates some kind of staircase signal instead of ramps. I meantion this for one reason only: This "staircase-compressed" signal sounds rather aweful after the compressor, but very smooth again after the expander. So the two staircase operations cancel each other pretty well, which tells something about the matching of the two halves of th edual vactrol. I've built this on bread board some months ago, using VTL5C3/2s.
Another circuit worth mentioning is Don Buchla's ring modulator. He built a four quadrant multiplier with two vactrols for the audio signal, and another two vactrols to form a servo loop. AFAIK, he used VTL2C3's (same thing as 5C3, but other enclosure). And we're talking about full scale audio rate modulation here! (one audio range signal modulation another audio range signal). So I guess (I have not built this, but I want to try this some time), I guess vactrols can be made to act much faster than expected with just the right amount of cuircuitry - and not to forget: A lot of component matching and trimming. (Buchla's circuits are (in)famous for being expensive and hand-adjusted.)
JH.