Hi all,
I was asked by a musician if I could do something about the temperature instability of his old Moog 342 synth. At the core of its oscillators is the CA3046 transistor array:
This design turns out to be quite sensitive to temperature changes -- as in: at least a half tone for every 6 degrees centigrade. This isn't much of a problem in a studio setting, with a constant temperature of some 20 degrees centigrade, although even then, it's necessary to let the instrument heat up internally for half an hour before playing. However outdoors, the synth is totally useless.
I tried replacing the dual transistors with an SSM2210 matched transistor pair (with a far better matching that the 3046), but to my annoyance, this didn't result in the improved stability I hoped for. Perhaps this is because the whole design isn't really symmetrical and balanced -- I haven't yet taken the time to investigate & calculate exactly what is happening.
Anyway, it seems that it's not just the transistor matching that's important, but also the absolute Vbe value. This latter can't be stabilized in any way except by providing a constant temperature -- so I'm thinking about another classic solution for cases like this: heat up the '3046's to a precise temperature, e.g. 50 degrees centigrade.
Now I already drew up a simple design based on a TC1047 temperature sensor, an opamp and a power transistor, to be stuck onto the transistor arrays, but I wondered if there's a simpler or more integrated solution available, such as a precision PTC or a special "heater chip". Or even a transistor with a built-in temperature sensor would be nice already.
As always, thanks in advance for any suggestions.
Richard Rasker