Well, it
is possible... However, we don't have the schematics for the MxB, so it would involve reverse engineering of the "open chest surgery" type. That is to say, we'd need to have the synth up and running while measuring the outputs of the CV demultiplexer. This is quite hard in practice since the construction used a modular approach with board interconnects via pins and sockets, which means that it's not easy to get to, or find without schematic and component placement diagrams, the points we need to measure. I.e. all the boards have to be plugged and screwed together for the synth to work (ref. the teardown pictures linked to in my previous posting). Some synths, especially older ones, use bundles of wires for board interconnects, which makes it easier to have each board out in the open and accessible for measurements. I should mention that MxB way of doing it eliminates that nest of wires which is good for reliability.
The array of ICs on the voice board (the left hand side of
VoiceBoardFront.jpg) make up the 96 channel CV demultiplexer and its sample & holds (hi-res version of the demultiplexer + s&h here:
VB_1.jpg). With the voice board exposed and the synth running, we could measure each CV (the outputs of the 064 and 074 opamps) while twisting the knobs on the front panel to determine which ICs/output-pins represents which CV. When that is mapped out, the output pins of the opamps could be desoldered and lifted from the board, and normalized jacks connected between the pins and the pads to provide overriding CV inputs. CV mixers could conceivable be made, but then we'd have to start adding quite a few components, most likely requiring a custom daughter board. CV outputs can be taken straight from the opamp output pins, with no need to lift the pins if only the outputs are wanted.
It may be possible to measure the CVs from the solder side (
VoiceBoardBack.jpg), which is exposed when the synth is screwed together, since there are capacitors and resistors there which could give access to the required circuit nodes - more reverse engineering would be needed to map that out. If not, we could in theory make extender cables to allow the MxB to run while we have the voice board out in the open for access to the component side, but then the project starts to get extensive.
Good reasons to not do it would be that it's fiddly and risky; it could all become a big mess with a non working MxB which no tech might want to touch. It's not impossible though, but definitely not for the faint of heart or the inexperienced. BTW, I've had mine open and taken out the voice board, so I do have a bit of feel for working on it, and can tell you that personally I'm perfectly happy with the modulation CV I/O that's already in place
(all the linked to images are by Tony Karavidas)
