Test theory to bounce off of you.
Hypotheses: x-axis is not perpendicular to y-axis, Im seeing a deviation of the calculated right triangle distance across the points of a 20mm cube do not equal the measurement i am getting after printing.
After some more calibration my 20 mm cube measures 19.98mm x 20.01mm with 0 slicer flow adjustments (all mechanical calibration).
I have a delta of .2mm from measuring across points 1-4 vrs 2-3
So it seams i have a parallelogram more then a true square.
each side measures parallel to the far side but that dose not mean they are at 90 deg from each other.
So my test thought is.
After lining up the print plate edge to edge with the heated bead on both the back and left edge.
(assuming the print bed is square )
print the following all boxes are on the same datum line from the edge. x,y= -125,-135 0,-135 125,-135 -125,0 -125,135
then if the red dimensions are equal to each other and all the blue dims are equal to each other then the tramming of the x vrs y axis should be perpendicular
I think the big possible error here to account for is making sure that the outer edge your measurements reference to is exactly parallel/perpendicular to the axis movements. A 0.5mm discrepancy at the corners could be pretty easy to come up with during install/assembly. I do like the theory though. I would probably start with using a carpenter or machinist square (preferably machinist) to check that the y-rails are trammed perpendicular to the z-rails. Make sure there is no torsional rotation on the x-axis (up and down slop in z-rail sliders) too.
Well, I’m not sure I’m going to get much better without trying a new rail. At first I tried to just adjust for backlash with the m425 settings, but at 0.2mm for the x axis, it was no good. It printed my little test cylinder OK (with nice round dimensions) but moving onto something bigger and more complicated (the clearance test wheels linked above) it started sounding horrible and the print had terrible layer shifting.
So then I gave in and took the rail apart to check the bearings, which were quite a bit closer together than they should be. Rechecked for backlash, which was still present, but down to around 0.13mm, so I changed that to just 0.1mm. Still getting layer shifting. Not quite as bad, but still definitely there - enough to make none of the discs free.
Disabled the backlash compensation again, and no layer shifting. So it definitely seems to be that the backlash compensation is introducing layer shifting for some reason.
At least I am a bit better off now. My first print of these test wheels, I could only just get the 0.35 disc free. Now I have the 0.2 disc free (although that did take quite a bit of prying to get it free).
Yeah identifying a possible issue, and testing for it with home tools can be two different things. I have some ideas but might have to wait till this weekend.
Model of the printer? A250.
This was with a print speed of 60, which I’ve printed plenty of things just fine with. This is the only time I’ve ever experienced layer shifting. And at times I’ve pushed the speed faster.
anyways i found that when i had layer skips it was actually running the infill to fast on some gyroid or elaborate fill with lots of little lines. 60 is fast but shouldn’t be an issue unless you cranked up the accel or anything.
The print was the clearance test that you posted a link to earlier. Just plain grid infill, all settings exactly the same as I normally print with.
Exact same gcode file prints normally when disabling the backlash compensation again.
Hi Spaced, I have been studying this thread with interest. After using the CNC for the first time, I have developed 0.2mm backlash in both Y axis modules, even though I was being careful with the CNC workspeed - not putting too much pressure on the axis. FYI, the nylon design has changed on my A350T (1yr old). The spring maintains a constant pressure on the opposing nylon bushes eliminating backlash. However, in my case the nylon has worn and reached the end of its spring loaded adjustment. So, I reckon the G & M code adjustments probably the next option for me. I had wondered about trying PTFE plumbers tape on the threads as a quick-fix, I’m sure it will end in disaster, but l’ll post how it goes. - Stu
Update to my previous nonsense… I have discovered that you can apply a small bit of force to the floating nylon (the one on the left in the picture above) as the carriage is installed back into the module . That re-instates the spring-loaded back pressure on the two opposing nylon bushes. Taking care to avoid the end of travel switch. I realize now that I introduced this error when I recently replaced the carriage bearings. I’m just putting the PTFE away and reaching for my dunce hat… LOL
Observation… If all the layers of a circle are printed in a consistent direction, either clockwise or anticlockwise, I get no flat spots or splits. However, since inner walls are printed in an opposite rotation to the outer wall by default, I get splits and flat spots. I guess this is classic backlash. At least I know now.
)
