When I cut a complex shape that covers most of the bed I have position errors that build up. I’ve been through all of the usual steps – checking for loose pulleys, belt tightness, backlash, current to the motors, acceleration. I lubed all of the ways, everything is moving ok.
I can sometimes mask the problem with order of cuts (in the optimizations screen), but never really fixed.
The error is quite repeatable.
Is there an issue here?
Any advice is appreciated.
My first thought is lost steps…Complex shapes mean a lot of acceleration and deceleration.
What if you do the same exercise at half the speed?
Is it different in Y and X, Is it always in Y or X and not the other axis?
I had the issue in Y. In my case the Y-stepper has to move mass of the whole X-configuration on the gantry. So I lowered the acceleration on Y and increased the current.
How big is the error? Can you share the file?
There will be SOME accumulated error even if everything is mechanically perfect. The physical step size is going to be a large number of digits after the decimal, if not irrational. The floating point calculations in the CPU and the DSP are also imprecise.
That’s at least three sources of error. All likely quite small, but cumulative.
Can you break the shape down into sections and print each section individually?
Thanks Walter, I had the same thoughts.I did the same run at about 1/4 speed and saw the identical errors. I then increased the motor current to x and y, no change. I then halved all of the accelerations, again no change.
This really has me scratching my head. I’m not ready to point any fingers, but perhaps it’s some kind of rounding errors or some-such between Lightburn and the Ruida? I would think others would be seeing this (and reporting it!), so this seems unlikely.
Neil
Understand where your thoughts are going after those tests. Seems to be a plausible explanation. If it’s the Ruida then there is no solution to be expected… I have a Ruida 6442, but probably my designs aren’t complex enough 
Here’s the item I made today, although this is NOT the first time I’ve see this:
If you do manage to cut one of these, let me know how it goes please.
Regards
Neil
Hi Rick, just saw your post thanks. I posted a link to the current design (see my reply to Walt), I used the DXF version. The error is on the order of 1 mm or so by the end of the job. It’s in both axes.
As you suggest, where possible I break the design into regions and do all of the cuts in each region as a group. This doesn’t eliminate the absolute error, but at least I don’t have so many non-connecting line segments and unfinished cut throughs.
Neil
Aaargh! My wife saw the box; Now I’m in for a test…but not tonight she says!
Not true - there are no decimals. Everything sent to a Ruida controller is done in microns, expressed as integers, so the numbers are perfectly repeatable with no error. All computations done by LightBurn are done in absolute locations, converted to microns, then offset from the origin point, to ensure precision. Any relative values that are sent are also computed from the micron coordinates. I can assure you there are no accumulated errors either in the software or the controller.
What most people fail to account for is resonance. Acceleration and speed play the largest roles, but your motion system has mass and flex, which means vibration, and if you ask it to change direction very rapidly in just the right places, that vibration can work against you and cause lost steps.
Watch what happens to the mass on the right when the platform hits 4hz at 1:15 sec into the video:
The belts in your laser have tension, and the metal has flex, so the system will have a natural resonance (probably a different frequency for each axis). If you make direction changes that happen to coincide with that resonant frequency at just the right times, you’ll amplify the resonance. Lowering acceleration will combat this because you aren’t imparting as much energy into it, and with the acceleration low enough, the mechanical damping will kill the resonance before it builds.
You haven’t said whether the issue is on the X, the Y, or both. What speed and accelerations are you running, and how large is your system?
Not true - there are no decimals. Everything sent to a Ruida controller is done in microns, expressed as integers, so the numbers are perfectly repeatable with no error. All computations done by LightBurn are done in absolute locations, converted to microns, then offset from the origin point, to ensure precision.
Thanks for the clarification, and a GREAT visualization of resonance. 
You’re welcome - I’ve put a lot of effort into making LightBurn accurate (within the specified tolerances). When a system uses GCode with decimal numbers, I often flip to relative moves for scans (because it makes the data smaller) but internally track the rounding errors and compensate for them so they don’t build up.
And yes, that video is great - I especially like how little the platform is moving compared to how much the sprung mass is. It’s a good argument winner.
Still working on the problem, will report back as soon as I ferret out the cause…
Thanks all for the replies.
Neil
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.