Delta auto bed level madness .. :)

myschetzky

Hello,

I've upgraded my Kossel build from Ramps to Duet Wifi, but I can't get my bed to autolevel. I'm using the IR sensor.
I feel like I've tried everything. It probes fine. The Z trigger height is consistent at 8 different places when testing.

But what made me write this post is that I can't understand my latest issue.

When I do a G32 and then a G29 on my regular build plate. A fiber slate with PEI, the height map shows a consistent slant going up towards the back side. Quite a bit actually. But its very consistent with the test print failures I've been doing. However it didn't seem to compensate for it even tho the height map made sense..

So I removed the fiber slate, and tried with the glass plate I had underneath. But then my height map went crazy. It looks like a hedgehog.
And of course the print is even worse since the adhesion is far from as good as PEI when the level is off.

So why is it probing so bad on my glass bed?
And why isn't it using the probe data I was getting on my fiber slate?

And last but not least - If my Trigger height is consistent all over the plate when testing manually, does that mean that I shouldn't worry about effector tilt?

Thank you for any advice!

/Niklas

myschetzky

From my console on the last probes on my glass bed

myschetzky

Next G29 try, also on glass bed

And console

myschetzky

And then because I wanted to see if it could be reflections or something about the glass, I did a G29 on a piece of paper.
Which resulted in this bonkers result

dc42

I'm sorry to hear that you are having problems. Some questions and observations:

1 it's strange that after calibration you had a consistent slant towards the back. Your height maps show evidence of such a slant. Did you print immediately after calibrating? If you reset the Duet first and you didn't save setting after recalibrating, it's possible that settings in config-override.g were overriding your calibration.

2. How did you measure trigger height at each point? In particular, did you lower the nozzle until it was just touching the bed or just gripping a sheet of paper, then send G92 Z0 to tell the firmware where Z=0 is, before raising the nozzle and running G30 S-1 ?

3. To use the IR sensor on a glass or other transparent bed surface, you need to have a black surface under the transparent material such as black paper.

HTH David

myschetzky

Hi David,

Thank You for your reply. Much appreciated!

Yes, but the slant measured by the probe matched the result on the print. So that suggests that the probe data isn't being used. I have G32 in my start script. And I deleted the the override. I don't think there is anything overwriting it.
I have a G29 S-1 in the end of bed.g

I lowered the nozzle until it gripped a piece of paper, then did a G92 Z0 followed by G1 Z5, and then a G31 S-1. Then I moved to another YX pos and did the same thing. Hmm or did I actually reset Z between new XY pos?! I'll have to retry that tomorrow to confirm. (At the pillow now)

Ah okay. My glass bed is clear with the heat thing glued underneath. So that explains the unreliable probes on the glass. Back to my PEI slate then.

Which means that I have to check if anything overwrites after G32. And check if I checked the Z trigger height correctly. If I didn't, could the slant be effector tilt?
I checked my rods and paired them in as equal pairs as I could. There isn't much more than 0.2mm in difference from shortest to longest rod.

dc42

Please post your bed.g file.

myschetzky

Here is my bed.g

[[language]]
M561						; clear any bed transform, otherwise homing may be at the wrong height
G31 X0 Y0					; don't want any probe offset for this
G28						; home the printer

; Probe the bed and do 6- or 7-factor auto calibration
G30 P0 X0.00 Y88.97 Z-99999 H0
G30 P1 X83.80 Y48.38 Z-99999 H0
G30 P2 X95.26 Y-55.00 Z-99999 H0
G30 P3 X0.00 Y-110.00 Z-99999 H0
G30 P4 X-95.26 Y-55.00 Z-99999 H0
G30 P5 X-83.80 Y48.38 Z-99999 H0
G30 P6 X0.00 Y37.35 Z-99999 H0
G30 P7 X47.63 Y-27.50 Z-99999 H0
G30 P8 X-47.63 Y-27.50 Z-99999 H0
G30 P9 X0 Y0 Z-99999 S7

G1 X0 Y0 Z150 F9000			; get the head out of the way of the bed

G29 S1

myschetzky

I re-mounted my fiber plate, upside down to protect my PEI from further damage. (Scraped a bit in earlier auto level tests)
Then I went through the https://duet3d.com/wiki/Calibrating_a_delta_printer page again, measuring trigger height. Checking repeatability and then running G32 a few times until it settled. And then finally a M500 to save.

I didn't change my bed.g since I just uploaded it here for review.

Don't you think these results looks fine and consistent enough?

myschetzky

So I did the above, and then started a print, but first I commented the S29 out in bed.g to try only with the G32.

This resulted in a print starting 2 mm above the plate. So apparently the M500 created an override file with a wrong height.

But also I could see a quite active up/down compensation going on. Several mm up and down over a few cm travel. Which I guess is odd considering I didn't do a G29. And I disabled the load of the old height map.

I'm so confused.