Software Leveling a Kossel

rgordon

I'm trying to apply Matter Control Software Leveling option to a Kossel Pro that I've been having a hard time to get leveled.

During manual measurements, I only see a variation of 0.3mm in front of the 3 towers and middle, but that's enough to stop that all-important first layer from properly sticking.

I've already had the length of the carbon arms checked. I've already checked the width of the arms at the carriages and applied very very thin shims to them to balance that out.

On the theory that the hardware-leveling Z-probe itself may be the problem, I tried using MC's software leveling option.

After going through it - i was dismayed to find that the printer *still* displayed issues as if the bed were not leveled. Printing the default 'calibration circle' object - the X side is still too high, and the Y-Z side is too low.

Here are the measurements MatterControl displays after going through the software leveling ritual using a 200mm print area (x,y,z):

100, 0, 2.14
86.6, 50, 2.14
50, 86.6, 2.08
0, 100, 1.94
-50, 86.6, 1.9
-86.6, 50, 2.08
-100, 0, 2.28
-86.6, -50, 2.28

I'm interpreting the Z column as the measured height at that location. We can see that points #1 us midway between Y-Z, #4 is in front of Z, #7 is between X-Z, and #8 its almost in front of X. Already you can see that the leveling thinks that the X-side is higher than the Y-Z side.

(the odd thing is that these are all of the points that MC would display - but as I map them out I can see that i'm missing some points - another bug or user error - not sure but I could have missed the non-standard scroll bar.)

So that's showing me a variation of 0.4mm, slightly higher than I expected to see from previous measurements ((0.3).

Trying to print directly at this point - did not work - the head was too close to the bed. I had to add a Z offset via the other setting. That started to work, but was still showing skew. X side to high, Y-Z side squished into bed.

I ran an experiment where I dropped a couple of those (#7 and #8) .2 mm, thinking that if made the software think that the side as actually lower, it would compensate with and properly level things.

(It's hack I know, i'm just trying to determine a way forward.) Needless to say, that didn't really do it - it was still printing high than the bed. And obviously i need to find the other points and change them in the display.)

so - i'm at a loss - how would the software leveling NOT compensate for the physical characteristics of the bed/printer? Why would I still need to "lie" to the software to get the leveling matrix correct?