PID Problems: Jerking while in steady run state

Thanks tommylight. I disabled CoolStep, StealthChop, etc a while ago.

I've added some jump detection code to my driver. It reports the positional jumps are within 1 ms of being 1 second apart. So I guess now the challenge is to figure out what's happening once per second that could cause this.

At this point everything is telling me that the TMC5041 internal counter really is jumping ahead about 1290 microsteps every second just like clockwork. My only guess is that this can only be a clock issue. I've ordered some 16 mHz crystals to confirm or reject that theory.

I think I've finally licked it.

I hooked the TMC5041 clock up to the clock on the Raspberry Pi. It didn't detect any improvement. So then I created a test program that essentially runs the servo loop like LinuxCNC, calling the same code just outside LinuxCNC. The problem was still there only not as bad.

I then started commenting out an non-essential TMC5041 register reads. And like magic the position jumps stopped. So in the end my driver loop was too slow. Reminds me of the old saying, when you hear hoofbeats think horses not zebras.

It took me so long to find the issue because the halscope graph didn't look like there was a delay. I expected the position number to flatten out before jumping. Maybe this a case of the graphing library helping too much. I also thought the SPI reads would be much faster than they actually seem to be.

My plan going forward is to have separate functions for issuing the commands to the TMC5041 and reading data back from it.

To wrap this up, I was able to completely resolve the problem just by changing the SPI clock divider from 1024 to 256. 128 was too fast for the TMC5041.