I’ve had network disconnect issues since the beginning, but haven’t ran anything more than tests. I am now on v0.781 and the last disconnect was a message about not hearing a heartbeat. Sometime I see a cable stepper motor studder then a stop. I moved a mesh node to the bench with my laptop, and ran a heatmap showing good wifi signal.
The job I was running was a grid of holes set at 6" intervals to test accuracy. It stopped after drilling the 8th hole of 105. Is it possible the heat sinks I upgraded are blocking the signal?
Getting a console is difficult, I should have taken a screen capture because when I reconnect it is restarted.
What would be my next steps to diagnose what is happening.
Yes, it is in alarm state and does not know where it is. I have to retract all and re-extend, add tension, and restart the job from the beginning. I will re-run and collect console data.
I reproduced the error and have logs! Looks like the disconnect was because of an alarm state. The first log is the failure while running a job. The second is after reconnect showing the machine was in a confused state (A TKO?) and required a restart, belt retracted, belts extending, tension applied to get it back to knowing where it was. I had hands off my laptop during the run. Normally I am exploring the UI or doing other laptop things.
We ran into this just the other day I think! Computer duster and cleaning the ethernet port fixed it for us. If you jiggle the wire can you get the green light to go off on that encoder?
Thanks! I blew out all RJ45 ports (I used keyboard cleaner - I did not have any more CRC contact cleaner on hand) and got in a new run that went farther but failed with a different ack. type error. Could this be a laptop to M4 timeout (maybe a processor sleep? the laptop itself did not sleep or hiberbate). I ran it again with the laptop on the power brick (next time I will try a performance battery profile) and it failed again but this was on me - I think the power cord came out (it has been over 100 here and the receptacle connectors are loose). Baby steps!
[MSG:ERR: Emergency stop. Update function not being called enough.1002ms since last call]
This is the important error message there. I think that this could still be the same issue of dust on the encoder connectors. I think that what is happening here is that it’s connecting to the encoders and then losing that connection and trying again and that process is taking so much time that it isn’t able to read from the encoders fast enough. I’m not 100% sure, but I was seeing this same issue at the same time as encoder read failures.
Success! Two runs back to back succeeded. Thank you for your encouragement and assistance.
The first time I attempted to correct from the connection error I blew out the ports and seated and re-seated the cables a few times. This time I blew out the ports but swapped the cables with some 18" I had on hand. I will get some new 12". I’m guessing cable issue. I was also going to move the cables to different spots to see if the problem tracked, but had something that sorta fit, though a little long…
When I made the second run, I changed the home by 0.25" and restarted the job. I noticed the pac-man blue/purple tracking dot no longer showed as a progress indicator (it wasn’'t visible).
Is there a way to command the sled to an absolute x,y? A go to from the command console? It would make it easy to get home in the same spot each time. I suppose I could define the job with home in the starting position (center) as well, but that would require more math.
What do you feel about covering the base of the sled with a thin UHMW plastic sheet? It is getting scratched up. I did wax it once, but that is something I’d stay away from in a production job so as not to leave a residue that would need to be washed out before finish.
Attached is the calibration test grid I was drilling. I’ll measure these out and report the accuracy in a future post.
Is there a way to command the sled to an absolute x,y? A go to from the command console? It would make it easy to get home in the same spot each time. I suppose I could define the job with home in the starting position (center) as well, but that would require more math.
I had experimented with G Code that just had a single travel command in it for this, as well. It worked fine, but the extra files taking up space in the list is somewhat annoying. Might be worth it to do for specific multi-part or batch jobs, though, so you can 1 click that if you have to re-hang. I wouldn’t necessarily trust it outright, even though I’ve had good results using the 0,0 home without manually homing and cutting jobs across power cycles/rehangs, and would probably want to still check the alignment before I tell it to update the saved home location.
What do you feel about covering the base of the sled with a thin UHMW plastic sheet? It is getting scratched up. I did wax it once, but that is something I’d stay away from in a production job so as not to leave a residue that would need to be washed out before finish.
That is something that has been suggested by several people. I haven’t noticed
any posts from people actually trying it. It should work.
in the fluidnc tab there is a place to enter raw gcode for the machine to execute.
You can just enter g0 x12 y12 to move to (12,12)
this assumes that you are in absolute coordinates (the common situation) and that the units are set as you expect (are you saying 12 mm or 12 inches?)
but I think that the vast majority of the time, this will ‘just work’ without you having to lookup/learn the gcodes to set absolute mode and set the units that you prefer
I have used UHMW tape on the bottom on the maslow Classic. Since they both use the sled I would imagine that it would work great on the M4 as well. I got the 3" 10 mil roll, but if I had to do it again I’d go with the 5 mil
