Odd G54 behavior

I’m seeing some odd behavior with G54 across reboots.

  1. Open a web interface to Kinetic
  2. Configure a non-zero G54
  3. Use the Server page to initiate a machine reboot

Once the machine has rebooted, you can for instance command a G0 X0 Y0 and the DRO doesn’t read 0, but rather what the G54 offset used to be. The “Setup” page shows all 0’s for the G54 offset, but if you use the drop down to “clear current system”, then things go back to normal.

Closing and re-opening the web UI doesn’t seem to help at this point (I’m not sure if it would have if I had done so while the machine was powered off).

It looks like the UI is getting out of sync with what the machine thinks the current work offset is?

This sounds like similar behavior to a bug in MachineKit that we thought was fixed when switching over to LinuxCNC. Does switching to a different G5X offset and then back to G54 bring up the correct offsets?

Yep, switching to G55 then back to G54 clears up the discrepancy.

We implemented a work around for this. This shouldn’t be a problem in v5.1.3.

My machine shipped with v5.1.1, I have updated to v5.1.2 and the only bug I have seen is with the G54 item. I managed to figure out the work around by switching between a diff G5X offset and back to G54 after some frustration and lots of clicks :slight_smile: .

All that said, I see that v5.1.3 is supposed to release today. Kinetic Control Software Versions and all that is listed is “Bug fixes”. Are there plans to go into more detail what that bug fixes entail? Any chance of getting a glimpse into the commit details for this minor version?

These bugs were all fairly minor. As far as I know, we’ve only encountered them internally. One had to do with MDI command history from the legacy software could corrupt the Kinetic Control UI state and cause it to crash (so if you happen to have the legacy software open and run an MDI command, opening Kinetic Control would crash on the Manual tab). One had to do with an edge case around rapid movement of rotary axes causing a following error that triggered an E-Stop. We’ll try to be more descriptive in our version history in the future. BTW, code changes for most releases are housed here, though commits aren’t always clearly tied to a version: https://github.com/PocketNC/pocketnc/tree/emcapplication

1 Like

Perfect, thank you for sharing the github link, I was not aware that it was in a public repo. Another reason that I am so happy with the pocketNC hobby platform!