MPG count going mad

Okay so today my MPG was giving me trouble. As I was jogging the X in one direction, the axis suddenly reversed at what seemed rapid speed. It was doing this in both directions. Keyboard jogging was no issues. I opened hal meter and the enc0 count would reverse from 127 to -127 at a single click and return to normal counting then repeat the reversal. Very strange. I thought the problem may lie in the cheap import MPG handwheel but maybe not so as everything seems to operate normally again after a linuxcnc restart. I did not re boot the pc. Any Ideas?? This is with 7i77 and axis on 2.8.0 pre1-3669. Last, what is a safe upper temp limit for PC and 5i25? My hmi enclosure was pretty hot the other day....

The 5I25 should be OK up to 70C, PCs usually quite a bit lower...


Was the 7I77s field power cycled when you exited/relaunched LinuxCNC?

Hi PCW,

No field power was not cycled. Although I did cycle the 5v to the MPG. Meaning there is a switch that just applies power to the a/b channel but the rest of the 5v and 24v was on and not cycled. I turned the mpg on before restart of linuxcnc when it worked.

I assume core temp is a different (higher) number than the overall pc max temp?

Thank you

Yes, but you can read the core temperature(s) by running "sensors"

OT-CNC wrote: Okay so today my MPG was giving me trouble. As I was jogging the X in one direction, the axis suddenly reversed at what seemed rapid speed. It was doing this in both directions. Keyboard jogging was no issues. I opened hal meter and the enc0 count would reverse from 127 to -127 at a single click and return to normal counting then repeat the reversal.

I have seen this a few times recently. Initially I thought it was just old firmware in the 7i73 in the mill, but it has also happened with the 7i73 i the lathe.

Is your problem with the 7i73?

Power-cycling (needs a power off of the board) fixes it until next time.

If it isn't just me and my very early hardware then I think this needs looking at.

This is with a 7I77 but it could be a similar issue

One diagnostic might be be to see what the mpg encoder rawcounts are doing when the problem arises

I will try to get a proper mapping next time.
Initially it looked rather like reversed-binary, but I need to check.

I loaded counts and rawcounts but don't remember if anything changed under rawcounts. The counts clearly switched from 127 to -127. When I changed the increments on the hadwheel, example .001, .002 it would happen every 2 revolution or every one etc. But always at the 127 count if I recall correctly. When I saw that I ruled out the actual MPG. Another thing, the max velocity slider or rapid override didn't slow anything down when it reversed at rapid rate.
What could cause this? I know I didn't change anything on the machine. The pc was running in hot enclosure but the core temps were in the 55 deg C range on the previous day running it.
My 7i77 and 5i25 are maybe 2 years old. I can get the specifics if needed. I'll see if it comes up again. It has been flawless. Just for clarification, I did not cycle power when it worked again. I just re-loaded linuxcnc and powered the 5v mpg after start up. Unless Andy you're saying that the board are cycled automatically when loading linuxcnc? In my case the 7i77 led is always lit when the machine is plugged in.

If you did not remove field power, the 7I77s digital I/O processor would not be reset
this suggests the the problem may be with LinuxCNC or the host rather than the 7I77 but I suspect the only way to tell is to read the actual low level register values when this happens. The +-127 count is suggestive of a rollover error somewhere since the actual MPG counter data is 8 bits wide.

PCW wrote: If you did not remove field power, the 7I77s digital I/O processor would not be reset
this suggests the the problem may be with LinuxCNC or the host rather than the 7I77

Both 7i77 and 7i73 share the same driver code in LinuxCNC. Do they also share the same low-level code in the FPGA?