There was a discussion thread on synth-diy about the lowest parts count multimode VCF that doesn't use a CEM type chip (does this remind us that some of us need to get a life?) Anyway, here's my low parts count multimode VCF that sounds good and fat.
It's a pretty standard cookbook design with Vactrol 5C2's substituted for the resistors that control cutoff frequency.
Eagle files of REV3 are here (schematic) and here. for the board layout
PDF file for schematic of REV3 is here and pdf for board layout is here here.
BUILDING THE VCF: You can get the parts layout in pdf format here and a parts list in Microsoft Windows "doc" format here.
About the layout: Not much to say, this is very straightforward stuff, but, for me, normal 5V P/P signal inputs may cause the output to distort, especially at high-Q settings. You may need to R11 can be anywhere between 10K and 4.7K, experiment with that to set the initial cutoff frequency. Also for modular freaks I suggest buffering inputs and outputs and put a 1K resistor in series with buffered output--otherwise if an output goes to a dead short you will heat up the op amps and maybe fry them. You may want to gain up the output regardless.
About using this VCF: Well it doesn't sound that good at every setting.
Some of the Q=1 sort of settings sound a bit thin. It won't oscillate, but you can mess around with the resistors in the feedback network (R10; R8) and change the amount of Q for the circuit and make it squeal if you really want. As far as mods, I parallelled another 10K on top of R10 and put a switch on it to take the new value in and out of the circuit, which added a bit of sharpness to the Q--a sort of "turbo-Q" switch. To further muck with the Q, experiment with the value pot that goes between "Q" and "Q Ret". I used 100K linear for the Q pot, see the wiring diagram. To tune this to your setup, you can adjust the input levels with R12/R13, and control the overall cutoff freq. modulation amounts with R18, R5, and R1....the 5C2's give it a sloppy, slow, phat sort of feel, which is good for certain audio applications and not for others. I found that 5C2's are not all created equal, so you may need to experiment with parts values C1 and C2 as well to get the sound you want or try different 5C2's in the circuit until you get a sound you like. This circuit in general is simple enough that you should be able to go completely crazy with it.