Another interesting finding. I downgraded the firmware to 1.0.4 and guess what, the DISCONNECTED issue was gone. Although this old firmware has other more serious bugs on it. LOL
The disconnected bug start at 1.0.7 on my testing:
And to make sure this is not a fluke, this video shows how consistently I was able reproduce it:
what does the local HDMI connected screen show. Everything you show seems to be a remote view…
Might also be evidence that even if it says Disconnected, I can still launch a new job. I just can’t move the axis manually
I can’t remember on witch version of firmware was added the asynchronus cleanup. This can monopolyse the rpi cpu.
what does the local HDMI connected screen show. Everything you show seems to be a remote view…
Yes, everything is remote. I don’t have an HDMI display anymore. I sold it. I don’t think it matters here.
1.0.7 did, this is most likely what is happening in this case and why it does not happen on his previous trys with 1.0.6 and below.
the cleanup is necessary as many users will run out of space causing major issues later.
cleanup is necessary but maybe there is way to limit cpu pourcentage (afinity) to this procedure so other process still continue to run.
@blaghislain Can you try the g-code that I posted here?
Not at home for the moment. Pretty sure it will do the same because I also get it on some cut.
I did try your gcode and could not get the issue. I tried both from my local monitor attach to controler and from my computer. I recently upgrade my SD card with this LAXTEK - 64GB Micro SD Card C10+U3+V30+A1 Memory Card so maybe it’S something you can try. Was not able to upload video
@blaghislain Thanks for testing that I will try to flash a new sd card and if that helps. I think we are pretty close in terms of configuration (spindle, custom switches, etc).
@blaghislain Do you mind if you can export your settings (json file). Maybe there is a settings on mine that might trigger this issue. I can perform diff on our settings.
Here is my config backup. onefinity-20210612_GB.json (7.1 KB)
Thanks. I will check on it later.
I had this happen once before and thought that I would have to reboot the controller. I can’t remember if it was version 107 or 108.
I found that If I hit the play button to re-run the tool path it look like it would attempt to run. I then hit cancel and re-zeroed my work before resuming.
Mike
Finally found possible solution to my case. You can try this one. Hopefully it works too.
Basically what I did is downgrade to 1.0.7 and then set max-devition to 0.001. The disconnected issue doesn’t happen anymore when I hit stop. Thanks to @blaghislain for helping.
Doing more test, it only solves one of my code. This g-code still has an issue.
If anyone can test it maybe we can have some more consistent test gcode.
BoltMountCNCMode.nc (27.0 KB)
While running a job this morning I had the issue happen. I am attaching the system log and gcode I ran in a zip file. I hope it will help Onefinity find and fix the issue.
Sequence of events leading up to the issue.
Powered on System.
Homed system.
Loaded 3 code files to the system.
Used the XYZ probe to find work 0
Change bit to a Keyhole bit and zeroed bit.
Probed Z only.
Ran profile tool path to cut key holes.
*** at end of the job the system returned to work 0 and the job looked like it ended normally.
Attempted to jog with joy stick (Failed)
Attempted to with screen jog button and received Disconnected message.
I also was able to recover by pressing the Play button and at the Continue prompt pressing Cancel. After doing that I am able to jog.
Disconnected system log and g-code.zip (12.4 KB)
Definitely this is a software bug.
I just pulled there source code of the controller on GitHub since its open source, modified, compiled and pushed it the controller. So far 2 of my gcode files that are consistently reproducible are no longer having issue.
Wished I can share the version of firmware updates I’m using but don’t think Onefinity will like it. Not sure I want to create a Pull-Request too.
Awesome. PLEASE do a push/pull request to their github so we can all have the fix (i’d personally love it!), and I have not doubt @OnefinityCNC would as well!