Bad cutting results with good servo PID loop. Out of ideas..

[SM_1]

Hi, I can't get good cutting results with my converted Esab plasma.

I have tuned the servo PID good (609 pulses/mm feed) and the maximum counted lagging pulses are 86 @12000mm feed and around 12 pulses at 2000mm/min feed. I think that 0.02 to 0.14mm as maximum fault is OK, and probably not worse than the machine was with oem controller.

Please watch this video and see the bad cutting results and give me some tips of what to do next. https://youtu.be/FHapPuibU-0?si=oeKJZrUlc8W8bHob

[SM_1]

https://youtube.com/shorts/ZAzqdpQkqg4? ... QPCFrA1nRu

Tested and calibrated backlash compensation. 0.35mm.

Cutting speeds below 1000mm/min gives adequate cutting results, but that's way too slow for thinner materials.

400mm/s acc. on both axes. I have tried lower and higher but still bad results.

[cncdrive]

It could be mechanical (resonance) problem.
Or it could be CV setting problem, too high tolerance problem, try to reduce the tolerances in the constant velocity settings.

[SM_1]

Ok, I will test with some different CV settings. Do you have any ideas of a good starting point?


Step resolution is about 1/609, approx. 0.0016mm. Since my tests shows the servos will lag a bit, the practical resolution would be about 0.02-0.14mm (less than 0.14 if I don't count the rapid speed travel).

The servos seem to handle quite large accelerations, I will make some tests of how high I can set it without "loosing steps". I have read that 20-40mG would be enough, but higher is better right?

Since the machine is a serial production machine from ESAB, that only has been used in a school for oxyfuel cutting, I would think that the mechanical part should work nicely.