taesungh
commented
5 months ago From last Friday with Sam: tested running the same program on two different Picos, connecting both to a dual-USB power adapter so they would turn on simultaneously. Observed on the oscilloscope that the apparent waveforms (after filtering) stayed pretty much uniform with no deviation in phase whatsoever (less than 0.01°). The deciding factor now remains on whether or not the timing of the serial communication will be consistent enough to maintain equidistant phases.
Going back on #25/#27 and UCI-HyperXite/hx9-control-system#31, we might want to reconsider whether or not we want to separate or unify the three-phase generation. Powers initially expressed concerns with lower driving frequency tor reaching the desired cutoff frequency of the filter as well as complexities in board layout and potential propagation delay issues. However, these have become more irrelevant since we are omitting a filter, the boards can still be separated per phase even with a single Pi Pico, and propagation delay differences are practically negligible. Additionally, different Picos may have slight timing inconsistencies which introduces a synchronization problem of maintaining the three phases equidistant as even small frequency differences will cause drifting over time.
As such, we'll need to experiment with how much of an issue timing inconsistencies will be and investigate possible solutions (perhaps with closed-loop control) or revert to generating all three phases together from a single Pico.