Delta twisted Prints
-
It's usually best not to let autocalibration adjust the diagonal rod lengths, because unless you can probe well outside the towers there is not enough information for autocalibration to determine them accurately.
-
+1 for that, measure the rods accurately from centre of rotation to centre of rotation at each joint/magball whatever, then use that value and choose S6 or S8 in bed.g for autocalibration.
Now just try printing 2 small cubes one in the centre of the bed and one right on the edge.
If they are both square, the same size and the right size you're done.
If they are both square but too large in XY then increase your rod length, if they are both square but too small in XY decrease your rod length, after changing steps/mm or rod length perform calibration again using S6 or S8.
If the central one is square but the outside one is not, then something is still not correct in terms of calibration, geometry (effector tilt?), or possibly motors skipping/umbilical pulling on effector etc.. I'd say few delta owners print precise parts near the edges of the bed, mainly as its much slower, the carriages have to move a lot faster to generate the movements. Sadly, as I love deltas, if you want to fill your build volume, you need a cartesian machine, even corexy's suffer from more error near the edges of the travel unless they are mechanically very precise. Its the price you pay for linking the motion of multiple axes together.
This thread covers it all in more detail https://www.duet3d.com/forum/thread.php?id=1912
-
ok, I have taken all your suggestions lowered max speed and motor amps several times ,at the moment i am at 8400 mm/min max speed and .6 A, took off the diagonal rod compensation in bed.g. I even lowered my max speed changes a bit.
Sizing is close to perfect 20x20x5 block comes out within .06mm in all axis. However i still can not get rid of the shifting and still can not get better than .033 deviation no matter how many times I run the auto calibration (ran it up to 12 times in a row.
Like i said before I have been printing with this delta for quite a while with a different board and firmware without any of this skewing, even when I print across the whole print bed.
Could I be missing some setting in the firmware to cause this? -
The cylinder sticking up at the corner looks OK, so I doubt it is a geometry calibration issue, or the cylinder would be skewed as well. Adjust your slicer settings to print very slowly and see if it makes a difference. I suspect dc42 Dave is no the right track; you are moving the tower carriages too fast when you get to the edge of your build plate and dropping/skipping motor steps.
-
the picture didn't show that cylinder very well but it is skewed as well.
-
i have also printed with my speed turned down in the slicer as well and it didn't help either.
-
Do all of the corners of your print look like this? It almost looks like your print is slowly revolving around the Z axis for each successive layer. Is it possible your print bed is not secure and slowly rotating? Have you tried printing something small but positioned at the edge of the print bed to see if it does the same thing?
-
that is what it looks like to me as well. that was one of the first things I checked and yes my bed is secure. and as DjDemonD suggested i printed 2-20x20x5 blocks at the same time (one in the center and one at the edge) the one in the center looked great but the one at the edge was twisted.
-
Sorry, I should have read the earlier posts more carefully. Is there more than one image on postimg? I can only see one. Maybe post a pic of your delta and what you modified to install the Duet board.
-
I didn't modify anything to install the Duet board. That is part of why I am so perplexed.
-
That means it's probably due to different settings for motor current, speed, acceleration or jerk. But just in case there is a firmware issue, run M122 and check the "Step errors" field.
I suggest that for now you try 6000mm/min max speed (M203), 1000mm/sec^2 max acceleration (M201), and 600mm/min max jerk (M566) until this is sorted.
I suggest also you do the calculation at https://duet3d.com/wiki/Choosing_steppe … e_you_need to check that you are not trying to move the motors too fast.
0.033mm deviation isn't too bad. Mine was a lot worse than that before I changed to the smart effector and magnetic rods.
-
0 Step errors, changed to the settings recommended now the lowest deviation I can get is .039 instead of .033 but the twisting is almost gone. It is still there but barely and nowhere near as much as it was.
