This was with the bench power supply, not powering anything else. There was an oscilloscope and a module tester on the same circuit… and an LED lamp… but turning the scope and module tester off didn’t affect anything.
I don’t think I ever powered the Noise Cornucopia with that supply — I think I got that supply later — though on the bench I probably was powering it with an ATX supply that surely was dodgier. But it’s true enough PSU issues work in mysterious ways. The synth it’s now in has linear supplies.
Anyway, I regard the problem as fixed, as far as the present circuit is concerned: The Rev 0 source works fine. If the Noise Cornucopia ever starts exhibiting similar behavior I’ll start by moving the transistor base to ground and that’ll probably fix it.
I prefer not to add digital chips where I don’t need to.