Want to address this problem. It comes from the greater springiness in the Y axis, because of its greater weight. The belts are tight enough.
I could slow down the accel in Y, but that also impacts job time negatively, in a significant way. I still think the setting is best overall.
This can be greatly improved in LB though.
I see this a lot- all these circles and things seem to prefer to start where the line is parallel to X. Which means you stop on X without Y in motion.
That’s the worst place for it. If the stop point is where the Y axis is moving, this feature is much less prominent because the Y’s bounce is inline with the cut.
So, basically, when looking at curved closed shapes, pick a start point where it’s cutting parallel to the Y axis instead of X.
That error comes from mechanical backlash, not an acceleration error: the Y axis does not return to the starting point after an excursion. The flat section across the opposite side of those holes shows the same problem.
In those spots, where the path is parallel to the X axis and tangent to the maximum and minimum Y axis travel, the Y axis is not moving, so its acceleration (or lack thereof) doesn’t affect the trajectory.
Given the overall size of your laser, it may not be possible to reduce the backlash below that fraction of a millimeter. Snugging the setscrews anchoring the Y axis pulley to the motor may help, but they’re probably already tight enough.
LightBurn can hide some of the backlash, but it’s not a cure-all and will certainly increase the job run time:
It is not “backlash” exactly. It’s not a dead space when you change directions, backlash would be from gear meshing and, yes, a loose pulley. This is stretch in the belts (and possibly torsion in the loooooong connecting rod linking the two Y sides). Backlash compensation will not help this.
Of course it can never be zero.
But this is all about acceleration, not velocity. The stretch happens upon accel. At constant velocity there is no stretch. Again, we can turn down accel, but that limits performance everywhere.
Accel in the Y axis is highest on the top and bottom edges, even though the Y velocity is zero there.
Putting the stop point on the left or right side means the Y axis accel is zero there. There’s less bounce.
It’d be interesting to characterize that with a square pattern along the X axis:
The center sections should all line up, regardless of whether the beam approaches from Y+ or Y-. Running at different speeds & accelerations could identify offset / overshoot as the Y axis velocity ramps up and down at the corners.
You’d need a much larger version to reach full speed, because a few millimeters will cramp your machine’s style something fierce!
Similarly, turn it 90° to see what happens along the X axis.
