I’m struggling with a strange alignment issue on my CO₂ laser with Ruida controller and LightBurn.
Problem:
On intricate layered mandalas / decorative cuts, each layer ends up slightly offset (typically around 1–2 mm). The strange part is that the offset often appears in roughly the same direction and distance, almost like it is systematic.
What I have tested so far:
Simple shapes (triangles/rectangles) cut perfectly, even at high speed
Step calibration is spot on in both X and Y
Belt tension checked
Pulley screws tightened and checked for slipping
Machine diagonals checked
Same issue happens with SVG and DXF
Same issue when running directly from Ruida memory (.rd file via USB)
Lowering acceleration from 10000 to 1500 did not solve it
Small sections of the mandala cut almost perfectly
Larger/more complex sections increase the offset gradually
Multiple passes of the same intricate design do not perfectly overlap
The issue has happened on several unrelated projects over time (lamp project, mandalas, Christmas stars etc.)
Optimization settings in LightBurn were also tested with most automatic optimizations disabled.
What confuses me:
The machine behaves perfectly on simple geometry, but complex fragmented designs seem to “drift” gradually. It almost feels like accumulated positional error or motion planner behavior rather than pure mechanical misalignment.
Has anyone experienced something similar with Ruida / LightBurn?
Any ideas are greatly appreciated
I wasnt shure to post it under Lightburn or hardware, so i used this cathegory..
I tested multiple passes of the same simple shape (triangle) at both low and high speed, and they overlap perfectly every time.
The problem mainly appears on intricate layered mandalas / decorative patterns with lots of small curves and fragmented/open paths. The more complex the design becomes, the larger the positional drift seems to get.
I also tested:
SVG and DXF versions
Running directly from Ruida memory (.rd file via USB)
Lower acceleration values
Tightening/checking pulleys and belts
Step calibration (spot on in X and Y)
Kerf offset is disabled during testing.
I am using Absolute Coords.
One interesting observation:
Small sections of the mandala cut almost perfectly, but larger sections gradually drift more and more. The drift also tends to happen in roughly the same direction each time.
As you see in the picure, this is run 3 different times. When i tried 4-5 pieses, no problem, all 3 times, perfectly on top of eachother.
Can you run those layers or passes separately? If so, try re-Homing the machine between them.
I am still suspicious backlash is present. Also, is it possible it is jumping a belt tooth in the Yaxis travel (where you have a lot of sudden direction changes)?
After a huge amount of testing, the issue appears to have been caused by different PWM Rising Edge settings between the axes in Ruida machine settings.
X-axis had:
PWM Rising Edge Valid = Enabled
Y-axis had:
PWM Rising Edge Valid = Disabled
After enabling PWM Rising Edge on BOTH axes, the positional drift disappeared completely — even on larger and highly fragmented/vector-heavy designs at higher speed.
Three-pass overlap tests are now perfectly aligned.
Huge thanks thanks to Ednisley for pointing me toward the PWM Rising Edge topic
The usual configuration has those switches set False, because the controller output is low-active and the signal should happen on the leading = downward transition. However, I haven’t dug into the electronics with an oscilloscope to figure out exactly how the STEP and DIR signals line up.
If it works, don’t mess with it. Keep in mind the switches may be backward should another weird misalignment appears (much) later on the calendar.
)