I’m running into an issue with my CNC during longer jobs. Over time, I lose accuracy, and by the end of the job, when I return to X0Y0, the bit no longer aligns with the original origin. This indicates I’m losing steps.
I recently upgraded to the EX Controller and servos, expecting this setup to resolve such issues. I’d like to understand what might be causing this.
If steps are being lost despite using servos, could the problem be isolated to mechanical issues, such as the belts or pinions? My understanding is that if I were pushing too hard into the material or running the machine too fast, the servos would detect it and adjust accordingly—am I correct? By the way, I don’t think I’m pushing it too hard for this job, but I have some 3D carving in it.
Since my migration to EX+Servos, I experienced accuracy issues on the right side of the table. I replaced the belt and later discovered a loose pinion, which I also replaced. Since then, I’ve been able to run complete jobs without any problems and I inspected all the belts and pinions.
However, during a recent long job (~3 hours), I encountered random inaccuracies. By the end of the job, X0Y0 was off. Interestingly, when I re-home the machine, X0Y0 aligns perfectly again.
Has anyone experienced similar issues, or could anyone point me toward potential causes or troubleshooting steps?
I don’t expect this to be your cause but it happened to me so I’ll share. My system is stepgen-driven servos but it’s homebrew. Sitting idle, the head would just… move. A little. Every once in a while. Turned out to be noise on the STEP signal; every EMI blip caused it to step.
My solution, and if EX supports it I’d recommend it, was to switch from STEP/DIR to quadrature drive for the servo signals. This way, noise on one of the lines will cancel itself out.
More likely, though, I’d check two things: first, belt tension and spring tension in the pinion drives. Second, manually try to move the servos and see how forcefully they hold position. If you can move them at all, they’re not tuned right.
It’s likely. Servos have very high resolution encoders in them. If you have any step loss they will immediately go into a fault state and halt the controller.
This really sounds like mechanical issues. If you haven’t already please open a ticket with support so we can dig in on this one for you.
Hi @djdelorie,
Thanks for your advices. I checked the bel and pinion drives again and everything looks tight. I tried to move the servos manually and they are hard to move and come back in place as soon as I release. So I think it’s good. But that’s a good thing to add in my check list. Thanks.
I’ve already opened a support ticket, but there’s at least a 24-hour delay between responses. At this pace, I might be stuck for another week. I’m trying everything I can to speed up the process, so any advice from the community is greatly appreciated.
Thank you for explaining how the servos work. I agree—it’s likely a mechanical issue. I recently replaced a belt and then the pinion, but I’ve noticed that the new belt I installed yesterday is already showing signs of wear, probably due to the loose pinion. This suggests there’s still an underlying problem in that part of the machine.
Today, I decided to replace everything at once: a new pinion and a new belt. I also verified that the rack is properly aligned, tightened everything, and ran a test job. It worked perfectly. I’ll continue running jobs to see if everything holds up and if the machine becomes more reliable.
So glad you got it sorted. Please do let us know if you need anything else. And if you haven’t already: Respond to that ticket and say you’re good for now
The support team hasn’t responded yet, and I didn’t want to close the ticket until I was sure the issue was fully resolved.
However, after two days of running various jobs—including a 6-hour 3D carving—I’m feeling confident that the problem is now fixed.
To be more specific, I’m genuinely impressed with the accuracy of this new setup. I’ve completed a few complex double-sided 3D carvings, and everything aligned perfectly. The results are far better than what I could achieve with the steppers and Mach 4.
What’s more, I can now pause a job, shut down the computer, and pick it up the next day, with the machine re-homing very precisely. This new level of reliability has made the migration worthwhile, and I’m finally starting to enjoy the benefits.
Now THAT is interesting. As someone who doesn’t make time for longer operations because there are too many other things competing for my attention, the ability to walk away and come back another time sounds pretty valuable.
I have notice that too. I can just power it down, come back the next day, home it and load up the same file and it all runs exactly the same. The feed hold and tool check are very robust as well.
