Auto leveling fail with firmware 2.03 final



  • Hello,

    After updating to firmware 2.03 auto leveling will not work anymore.

    Error Description:
    All points are measured with the G32 command.
    The print head automatically moves up to 14 mm after the measurement.
    The display shows a height of 14 mm.
    Then the printhead should be moved up another 5 mm.
    The display then shows a height of 19 mm.

    Unfortunately, the printhead does not go up. Instead, he drives completely into the print bed.
    The display still shows the 19 mm.

    I have uploaded a video who shows the error.
    https://vimeo.com/342503831

    config.g
    0_1560644884624_config(7).g

    bed.g
    0_1560646014850_bed(2).g

    It still worked with the old firmware 2.02 xx, but I must update to 2.03 (fix for I2C Error).

    I would be glad about suggestions!


  • administrators

    That's a strange problem! [And an interesting triple-extrusion delta!]

    You didn't show the full bed probing at the start of the video. Can you confirm that that the video started with the last 2 points being probed?

    Please can you:

    1. Home the printer
    2. Run a copy of your bed.g file as a print file or a macro, but with the S6 on the final G30 command changed to S-1 so that it prints the height errors but doesn't adjust anything.
    3. Send M665 and M666, both without parameters, and report the responses here.
    4. Run auto calibration. Report the before/after deviation here.
    5. Send M665 and M666, both without parameters, and report the responses here.

    What type of Z probe does your printer have?



  • @dc42 said in Auto leveling fail with firmware 2.03 final:

    That's a strange problem! [And an interesting triple-extrusion delta!]

    Thanks. The triple extrusion delta works great (3 x 0.8 nozzles)! (except the I2C error 😉 )

    0_1560694473917_TBS.jpg

    @dc42 said in Auto leveling fail with firmware 2.03 final:

    You didn't show the full bed probing at the start of the video. Can you confirm that that the video started with the last 2 points being probed?

    Yes.

    @dc42 said in Auto leveling fail with firmware 2.03 final:

    Please can you:

    1. Home the printer
    2. Run a copy of your bed.g file as a print file or a macro, but with the S6 on the final G30 command changed to S-1 so that it prints the height errors but doesn't adjust anything.
    3. Send M665 and M666, both without parameters, and report the responses here.
    4. Run auto calibration. Report the before/after deviation here.
    5. Send M665 and M666, both without parameters, and report the responses here.

    0_1560694787961_Auto-leveling.PNG

    @dc42 said in Auto leveling fail with firmware 2.03 final:

    What type of Z probe does your printer have?

    Your Duet3D IR-Probe.

    Note: I have changed the M665 command in the config.g for this test.
    From: M665 R562.5 L1021.6 B450 H1021.4
    To: M665 R562.5 L1021.6 B450 H1028.57
    It is the real high. The result was the same with the old config, too.


  • administrators

    The image you posted shows why it is going wrong: auto calibration is changing the homed height from 1028.6mm to 1050.6mm. This is obviously wrong. My guess is that it is connected with the change in delta radius from 562.5mm to 527.9mm, which is a large change.

    I will run some tests and examine the auto calibration code to try to work out what is happening. Meanwhile, here are some suggestions:

    1. After running the first auto calibration cycle, home the printer, send M558 H35 to increase the dive height, and run auto calibration again. Then run M665 and M666 again, and report the values. With luck it will correct the homed height value and only re-adjust the delta radius a little.

    2. Revert to firmware 2.02 at least temporarily, run auto calibration a few times (until the deviation stabilises), then send M665 and M666 to report the delta parameters. That should show us what values auto calibration ought to converge on. I am particularly interested in whether the delta radius ends up close to 562.5mm or 527.9mm



  • @dc42 said in Auto leveling fail with firmware 2.03 final:

    1. After running the first auto calibration cycle, home the printer, send M558 H35 to increase the dive height, and run auto calibration again. Then run M665 and M666 again, and report the values. With luck it will correct the homed height value and only re-adjust the delta radius a little.

    0_1560727863684_2malLeveling.PNG

    It seems the double auto calibration (G32) hast worked. The printhead doesn't move in the print bed anymore.
    And the height is correct (+0.1mm). The real height is 1028.68.

    @dc42 said in Auto leveling fail with firmware 2.03 final:

    1. Revert to firmware 2.02 at least temporarily, run auto calibration a few times (until the deviation stabilises), then send M665 and M666 to report the delta parameters. That should show us what values auto calibration ought to converge on. I am particularly interested in whether the delta radius ends up close to 562.5mm or 527.9mm

    I can test it tomorrow.


  • administrators

    I have identified the problem. When the auto calibration adjusts the delta radius, in firmware 2.03 it makes an incorrect adjustment to the homed height. Your initial delta radius was a long way from the true value, so it make a large incorrect correction to the homed height.

    I will fix this in a 2.03.1 release. Meanwhile, if you set the R parameter in your M665 command close to the value found by auto calibration (527.726mm) then you will not have this problem.


  • administrators

    I have put a temporary 2.03+1 firmware binary for Duet WiFi/Ethernet at https://www.dropbox.com/s/zfv4p7dr2ycmjzd/Duet2CombinedFirmware.bin?dl=0. Please check that it fixes the problem on your printer.



  • I have updated to 2.03+1.
    1_1560897269582_2.03+1_.PNG

    The auto leveling worked successfully with the first G32 command.
    0_1560897269582_2.03+1.PNG

    Thanks for your fast help.

    (Ticket can be closed).


  • administrators

    @tbs, thanks for confirming this. I will do a 2.03.1 release soon to include this fix, but I'll wait a few more days first in case any more bugs emerge in 2.03.


Log in to reply