by crochambeau » Wed Oct 06, 2021 10:52 am
I'm not familiar with the TAPLFO or published circuits for it, is your implementation based on a published circuit or is it something you have cooked up?
Again, I'm not familiar with the part in question, so take everything I say with a box of salt...
It looks like you're trying to slap an AC circuit (passive low pass filtering/wave shape) and a DC circuit (passive voltage divider) together without buffering. Generally speaking, insofar as AC (signal) is concerned, either end of your power rails are essentially "common" - so the 10K voltage divider is "seen" as a quick path to ground regardless of setting. I mean, you've got over 1M of load resistance in your wave shaping that is then planted a maximum of 30K away from ground (ignoring the power rail). That's a big voltage divider in itself.
What does your MIDI to CV need to see to do its job?
I would suggest slapping an opamp in where your 10K summing resistors are. If you want to be stingy with parts you can probably stick your AC waveform into the non-inverting input, add a smallish (say 10K) resistor in the negative feedback path to set some gain (I'm assuming your waveform might need it, I may be wrong), connect your voltage divider to the inverting input as a sort of sliding ground reference that will push and pull the output DC up and down. Note that you're inverting the signal and connect the pot accordingly.
Adding gain can push stuff out of whack really fast (especially if you're trying to fly the analog section inside the 0 to +5 voltage window, summing AC on top of a DC rail to rail swing will distort, and it would be advisable to connect the op amp Vcc to a +9 if available to buy yourself some headroom), so before you connect everything it can be helpful to test the state of this node (variable voltage plus your LFO waveshape) before hooking it up to anything. If you need to scale up or down you can do so before you hit your MIDI to CV stage.
This has been a speculative over simplification concocted over morning tea, and may be rife with errors. But I'm guessing your woes are tied to too much uncompensated resistance in your wave shaping filter basically dead heading into the convenient ground path your DC reference creates.