smells of bikes on Nostr: I'm characterizing an Arduino's ability to make precise pulses. If I bitbang it ...

I'm characterizing an Arduino's ability to make precise pulses. If I bitbang it (yellow trace) using timers and direct port operation, I get, for example, 63.3uS when I want 64uS. If I set up a hardware pwm (red trace) I get 64uS. Pretty similar. But look at the sigma. The bitbang variation is about 7uS. The pwm variation is 10 _nanoseconds_. Like 500x better. This project needs <4uS jitter so I have to slog through the math for precise single shot pwm.