Presumably each voice terminates into its own vca, then they all are summed into a mixer.TheWaveWarden wrote: Wed Mar 17, 2021 9:32 pm The distortion I'm hearing must happen in the section where the voices are mixed already, since the spectrum clearly shows frequencies below the notes ones being played. This can not happen if the individual sine-waves are distorted, since that would just produce harmonics above the fundamental. I am still a bit confused by the lack of headroom in this section, but after playing for the evening it seems nothing bad happens unless I force it, like in the video.
The mixer if overloaded could clip, and if polyphonic more than one voice it would make intermodulation distortion, sum and difference frequencies above and below original vco pitches. That would happen with any wave shape. Even square wave, or high gain sine clipped to a square.
This isn't real surprising about headroom and extra gain added by resonance. For example set up a PA mixer or daw song mix adjusted to "nominal" output level, then boost all the mixer inputs by 20 dB or more and see if the mixer starts clipping.
You could possibly be getting distortion in all three stages, voice filter, then voice vca, then the output mixer. If one distorts it doesn't automatically rule out the others also distorting.
One possible way to get filter distortion harmonics lower than fundamental with only a single sine oscillator feeding the filter-- Overdrive the filter, turn up the resonance and tune the filter "close but not the same" frequency as the oscillator. This nonlinearity will cause intermodulation distortion, sum and difference frequencies, between the oscillator pitch and the different pitch of the filter resonant peak that is ringing, "trying to self oscillate" like a gooney bird hopping down the beach unsuccessfully trying to take off.
