JordanAceto
commented
3 years ago Done, centered envelope follower around +1v. I did not change the invert switch to a single pole version, but I thought about how to do it right after I ordered boards. Just apply the +1v reference directly to the non-inverting input of the opamp, and get rid of the bias applying resistor that gets switched in. Oh well, add this in some other time.
The envelope follower level pot attenuates the envelope signal that sits in the range [0v, +2v].
When you turn it down, it goes down to zero volts.
This feels weird with the way everything is scaled, it would feel more natural if it turned "down" towards +1 volt (the middle of its range), squishing peaks down and pushing valleys up. This is how the LFO is attenuated, it swings between 0 and +2 volts, centered around +1 volt, and attenuates to +1 volt.
As a quick fix, cut the ground leg of the envelope follower pot and inject a +1 volt reference. This could be created by a 2.7k / 1.2k resistive voltage divider across the +3.3v rail. Put a big cap across at least one of the resistors. If you do this, consider also changing the EF level pot to 100k, the bigger the pot to more attenuation you'll get when you turn it down. There still might be some EF bleed when the pot is at a minimum.
When a pcb respin happens, consider ways of scaling the EF to accomplish the attenuation behavior described above. Maybe this means rethinking the norm/invert switch and the method of generating the offset. Perhaps the EF could swing from [-1v, +1v], centered around zero, and then get pushed up 1 volt later in the CV summing amp.
Bonus points if you can change the norm/invert switch to a single pole switch while you do this, I bet you can.
If it seems like the best way to do this is to add another opamp, you could use the other half of a dual package for the LED driver and get rid of the existing transistor LED driver.