This behavior was observed on the bench testing platform (using an Atmega2650) until hysteresis and a few other enhancements were added in #24.
Currently those enhancements have not yielded the same results on the Rev.2.0.1 production boards. Screencap of the current PWM output behavior:
It's possible the slower clock speed of the At88 vs the At2560 is a factor, as the ADC is getting fewer checks and less able to correct for overshoot compared, or there could be other unforseen factors like PCB layout and capacitor type.
First steps will be to reduce the delay for polling the ADC pins to see if that improves the behavior, other tests will probably reveal other solutions
This behavior was observed on the bench testing platform (using an Atmega2650) until hysteresis and a few other enhancements were added in #24.
Currently those enhancements have not yielded the same results on the Rev.2.0.1 production boards. Screencap of the current PWM output behavior:
It's possible the slower clock speed of the At88 vs the At2560 is a factor, as the ADC is getting fewer checks and less able to correct for overshoot compared, or there could be other unforseen factors like PCB layout and capacitor type.
First steps will be to reduce the delay for polling the ADC pins to see if that improves the behavior, other tests will probably reveal other solutions