Gmoccapy & 7I96 jerky movement

I managed to catch it! There's a section of g code that's milling a letter S in XY that seems to cause particular problems.

The outline shape of the plots is recognisably the same but for the good S as you'd expect the velocity commands are smooth WRT time and with the bad S they're very much not....

These commands come from the linuxcnc motion controller so it can only be something wrong with that?!

It really looks like the feed-override has been changed between the two plots
(or the G64 settings)

Both plots are on the same timebase on the scope and at 100% feed rate in linuxcnc.... counting the divisions between the peaks on the pink trace it looks as though the bad S is running a bit slower, but maybe that's because whatever is making the waveforms is getting bogged down and that's why the bad one is also so ragged.

I still don't really get how this can be intermittent!

OK so it really looks like the G64 /G61 settings are changed

Do you have G64 settings in the prefix of your gcode file?

No G64 in the gcode (attached). It's gcode output from the Fusion 360 EMC post processor.

The intermittent nature is the most baffling thing to me. Usually these things work or they don't!

You need to have g64 settings (say g64 p 0.05 q 0.05)
in your gcode file or your trajectory settings will be whatever
was set by the last file that changed them...

I'll add that to the gcode and report back. Thanks a lot!

Hi Peter,
You were right about the G64...

I'd made a macro for putting the machine in centre position using one of the example macros, and on inspection it had a G61 in it cancelling any G64. This is why restarting linuxcnc cured the 'fault'. Now I feel silly!

Thanks a lot for your help again.

Yes, having an optional macro that changes a state variable can cause
tricky issues

Too bad LinuxCNC doesn't have "restore previous motion mode" gcode