This is a very strange issue. It’s strange that it is only happening on the high side, but it could be that both sides are affected the same, but the LPF is filtering out the noise on the low side. The fact that the noise seems to change characteristics with the rate knob makes me think that it has something to do with the microcontroller, although I can’t imagine how it would be behaving this way. If you have a USBTiny programmer, you may consider re-flashing the microcontroller to see if that changes anything. If you happen to own an oscilloscope you can scope the PWMs coming from the microcontroller on each side (high and low) and compare the wave forms.
One more thing I wanted to ask is if it behaves any differently when the mode switch is flipped.
-Brach