This module is basically the Steiner VCF, made famous by Ken Stone. The original Nyle Steiner article is found here. There are a few pages on the net that describe this filter including the Nyle Steiner VCF page here and this one here. There are probably many more.
I chose to substitute the use of the venerable 2N3904 transistor in this design for Q1 and Q2. Beta match these as close as you can for more accurate tracking. Some people choose super matched pairs or transistor arrays for these. The choice is yours. Common net wisdom seems to suggest that for this particular service, a reasonably well matched pair of transistors will do here.
Another variation of the original, is that I chose to use an opamp shown as U9 in the circuit below. The original design used a pair of discrete transistors for this. I followed the lead of one of the links above and used their approach. I feel that this opamp circuit is easier to customize and tweak than a pair of transistors with several resistors.
Diodes D9 and D10 are an innovation of Yasunori Hirakawa. Their purpose is to limit the amplitude of the signal should it begin to oscillate with excessive resonance (excessive positive feedback).
A dual wafer switch then connects the “Audio” label to one of:
In the diagram, the connection is simply shown as connecting the buffered Audio signal to the LP input.
The diodes D1 through D8 act as voltage controlled resistors. C3 is the participating capacitor in the RC constant. For more information about diode strings used this way, see "Diodes as voltage controlled resistors".
To make the diodes conduct (and act as a resistor), transistors Q1 and Q2 generate a differential voltage. The anode of D8 is positive with respect to the cathode of D1.
Control inputs go through CV level controls U8 and U3, which are then summed and passed to the base of Q1. The frequency control (cutoff) is adjusted by pot U4. Trimmer U5 allows calibration of the VCF, so that the control voltage tracks with the filter cutoff. Q1 responds in an exponential manner, so the control voltage is exponentially converted.
C6 couples the RC filter output into opamp U9, which allows control of resonance through feedback pot U6. Output from U9 is (positively) fed back into the RC filter via R12.
The output of U9 is also boosted and buffered by opamp U7, which provides the VCF's output signal.
The voltages shown are the approximate steady state values of the circuit.
For this simulation, one control voltage was stepped through 0 to 5 volts, 1 volt at a time. The input AC signal was 3 volts in amplitude. Resonance was set to minimum.
The curve at the right shows the response with the resonance control set to maximum. In the actual circuit, this might oscillate. Some gain adjustment may be required to achieve this, if it doesn't after the build.
The plot at right shows the band pass response, with the same control and input signals. The resonant control is set to minimum.
The module construction started with the front panel (right), with the prototype PCB attached with two copper tubes. A ground wire was run over the top of the pcb as triangulated support. The pcb has the first TL072 opamp installed with incoming power connections.
At left is a picture of the module installed into the cabinet. The mess with the alligator clips is an improvised “Y” connection to the CV 1 input jack (patched also to the XR-VCO).
The schematic above is the final circuit used in this project. Pin numbers were assigned to the opamp connections and resistor R20=1k was added to the output circuit. The TL071 opamp in the earlier schematic was replaced with a TL072 opamp, leaving one half of it unused. I just happen to have quite a few of these on hand.