Over the past few weeks I was troubleshooting a few gremlins I was encountering.
SITUATION
MPCNC Primo
Jackpot controller running FLUIDNC
Fusion 360 education license with FlyFischer v3 post-processor
1/4” single flute carbide end mill
ISSUE
with each plunge into the material (cedar), the physical z height would cut increasingly deeper when it should end up at the same height across the work piece. I confirmed that the gcode was supposed to go to z-3 (for instance). As the program moved along deeper and deeper into the workpiece, it drifted further and further deeper than intended so that by the time it was complete with OP1 (3D adaptive clearing) there was multiple depths at the “floor” instead of the same, and my next OP was air cutting when it should be engaging material.
HYPOTHESIS
The tool is either losing steps while retracting, or the tool was moving out of the collet.
rapid retracts were causing missed steps, therefore the next cutting move was deeper than the previous one.
SOLUTION
originally I had z acceleration set to 800 mm/s^2 and after a few runs found that lowering the z acceleration to 300 mm/s^2 resulted in the issue going away.
I did check the collet tightness, and found no issue.
I checked the z leadscrew coupler grub screws to ensure it was secure and not slipping, and found no issue.
Just for curiosity’s sake, has anyone else run into this issue and what z acceleration setting seems to work best for you?
Another (not an issue) thing I found was that Fusion 360 would grossly under estimate the OP completion time, was that FluidNC acceleration settings will affect the actual time to complete the OP. Fusion 360 does not take acceleration limits into account. Therefore, the lower your acceleration settings for each axis, the longer the OP will take. In my example, a 3D Parallel OP (with LOTS of up/down movements while cutting) took more than 4 times as long than estimated by F360. I increased my z acceleration settings to 800mm/s^2 and the job was completed in only 1.5 times but then I ran into the issue above. By decreasing the z acceleration to 300, I’m expecting a happy middle ground of 2x the estimate.
This should only have more of an effect if the tool is accelerating into/out of a cutting move more often.
I first experienced this issue a couple of years ago when Fusion 360 removed support for rapids in the education license version. The result was that the Z feedrate would be the same as the X and Y feedrates. At the time, the recommended fix promoted on the forum was to limit the Z feedrate in the firmware. This fix worked for me. You might compare your solution (reducing the acceleration) to this solution to see which one has the best machining time.
Yes I had exactly this but on a lowrider using a jackpot. I could hear and see the skipping when it was lifting Z. It was at 1200 (I think), I reduced it to 800 and it appears to have resolved it but I’m keeping an eye on it.
With the E edition I have the option to remove rapid retracts, but have it enabled. I’m curious to see, so I’ll definitely try it out with it turned off.
My assumption is that due to the toolhead weight, the motor was skipping steps due to the limited torque available with higher accelerations. Once I install a fan on The controller box I may monitor the drivers temp and see if I can raise the run current a bit more to 1 or 1.2A to increase the torque. My only concern is that the motor may heat up too much for the PLA parts of my MPCNC.
It should never need that much current to hold. If it does, something else is wrong.
You’re not only going to heat up your motors to the point it may weaken the motor mounts, but you are likely to put the drivers into thermal protection, in which case your motors will be disabled while the driver is in thermal shutdown. You’ll for sure lose steps when that happens.
I have been playing with scrap wood and bs cuts. Just trying to figure out cam software in general. I have been working with f360 more and more but have yet to encounter that same issue of skipped steps. I’ll keep an sharper eye out for it though. I think maybe because I’m not really pushing anything towards any kind of limit, that I’m not likely going to see these issues until i actually ask for something more in line with real work.
I just got my ball nose tool in and really want to try a scallop cut with it.
At that low? What would cause thermal protection to trip if I have a fan blowing on them? I’ve ran TMC2209s at 1.7A actively cooled for hours on my printers while testing with no issues. The motor heat and PLA mounts are the main issue with raising current. I’m currently At 0.9A and ran the MPCNC for 5 hours yesterday with no issues thankfully.
Honestly I don’t think I’m going to push it anymore on this. It’s working well within my needs now. 300mm/s^2 z accel was pretty good for what it is.
I believe the skipped steps were a product of trying to lift the heavy core at 600-1000 accel. I had a few skipped steps at 400 but now it’s running smooth at 300.