>>2946729
This is for a 2-pole voltage-controlled filter (God help me). So, the OTAs control the cutoff frequency of each stage by sending current to the amp bias input, yes. I hadn't seen Figure 22, but that would seem to be the simplest option because the bias currents should always be the same. I was working from other designs where one PNP transistor per bias pin was used, but that was for a 4-pole filter. Maybe I can get away with it because, as you say, the amplifiers on the same IC are matched.

I think I have it all mainly working in Falstad, although the characteristic resonance peaks on my input square wave appear at the wrong place in the circuit. I also wonder whether I really need to invert the signal before feeding it back in the resonance loop.