Ineffective autolevel results

I have a delta printer running repetier .91 with a capacitive Z-probe.
The problem I am having is that after running autolevel with G32, small prints in the center of my bed stick well but larger prints drive the nozzle into the bed or have it raised 1-2mm above the bed, it seems like the autolevel script does not accurately calculate the tilt of my bed.

There are a few things that might be causing this issue:
I don't know if the horizontal radius or diagonal rod length is perfectly asccurate, I am fairly sure they are close as when I level the bed manually using x,z,y offsets I get fairly consistent results.
I am a little confused as to how the x,y offset from the probe to the nozzle works, if it is +20mm from the nozzle in the Y direction, is the offset +20mm or -20mm.
my probe seems quite reliable and is fixed in place so its unlikely that there is any mechanical error there.
I am using a 6mm thick float glass mirror as my print surface so it is extremely planar, the discrepancy is not accounted for by an uneven print surface.


  • If horizonatl radius or diagonal length is wrong you get a dome or bump with extreme in the center. Autoleveling does NOT fix that. You need to iterate one of them until the error is at minimum. Also wrong alignment of towers e.g. 90,05° instead of 90° make a difference in z height.

    0.92 has the possibility to create a bump map to reduce such errors a bit, also it is meant more for 0.2mm error, not 1mm. So first calibrate geometry that the error is rather small.

    All offsets are with teh same xy coordinate system your print does, only with carriage center as origin (same goes for extruder nozzle if it is not in center anyway).
  • Thanks for the response repetier!
    Where can I get a download of .92 there is only .91 listed on the firmware page?
  • Follow that link and then you see a link to 0.92 or just replace 91 by 92 in the url.

    See repetier.ino for a list of commands.
  • I spent most of this afternoon carefully calibrating the horizontal radius and I can now print much larger prints, though I am still not making use of the Z-probe, my bed is removable so it can be easily cleaned and the reason I want z-probe function is that after removing and replacing it a few times, the printer tends to need calibrated again.

    I got repetier .92 running but the .ino file doesn't contain any G codes beyond the usualy G29, G30, and G32 for autoleveling the bed, I can't see a code for running a "bed correction" and google doesn't seem to be turning up any results tutorial-wise

    From the description in the configuration wizard, it sounds like the auto-correct runs a grid rather than a 3 point probe, g29 and 32 both run 3 point probes currently.

    here are the G-codes from my .ino
    - G0  -> G1
    - G1  - Coordinated Movement X Y Z E
    - G4  - Dwell S<seconds> or P<milliseconds>
    - G20 - Units for G0/G1 are inches.
    - G21 - Units for G0/G1 are mm.
    - G28 - Home all axis or named axis.
    - G29 S<0..2> - Z-Probe at the 3 defined probe points. S = 1 measure avg. zHeight, S = 2 store avg zHeight
    - G30 P<0..3> - Single z-probe at current position P = 1 first measurement, P = 2 Last measurement P = 0 or 3 first and last measurement
    - G31 - Write signal of probe sensor
    - G32 S<0..2> P<0..1> - Autolevel print bed. S = 1 measure zLength, S = 2 Measue and store new zLength
    - G90 - Use absolute coordinates
    - G91 - Use relative coordinates
    - G92 - Set current position to cordinates given
    - G131 - set extruder offset position to 0 - needed for calibration with G132
    - G132 - calibrate endstop positions. Call this, after calling G131 and after centering the extruder holder.
  • I see.

    G29 creates the distortion map

    M323 S0/S1 enable disable distortion correction P0 = not permanent, P1 = permanent = default

    Without parameter shows current operation mode.
  • Ok, last question hopefully, you've been really really great thanks!
    I got that working with the M323 command but how do I incorporate the distortion maxtrix? The host kicks out this after running the g29 command, but the performance of the printer seems unchanged:

    22:15:18.730 : Z-probe:12.52 X:-100.00 Y:100.00 zCorr:0.00
    22:15:29.389 : Z-probe:12.79 X:-50.00 Y:100.00 zCorr:0.00
    22:15:40.171 : Z-probe:12.80 X:0.00 Y:100.00 zCorr:0.00
    22:15:51.031 : Z-probe:12.93 X:50.00 Y:100.00 zCorr:0.00
    22:16:02.062 : Z-probe:13.15 X:100.00 Y:100.00 zCorr:0.00
    22:16:14.313 : Z-probe:12.68 X:-100.00 Y:50.00 zCorr:0.00
    22:16:24.939 : Z-probe:12.61 X:-50.00 Y:50.00 zCorr:0.00
    22:16:35.550 : Z-probe:12.64 X:0.00 Y:50.00 zCorr:0.00
    22:16:46.254 : Z-probe:12.76 X:50.00 Y:50.00 zCorr:0.00
    22:16:57.068 : Z-probe:12.91 X:100.00 Y:50.00 zCorr:0.00
    22:17:09.069 : Z-probe:12.45 X:-100.00 Y:0.00 zCorr:0.00
    22:17:19.522 : Z-probe:12.47 X:-50.00 Y:0.00 zCorr:0.00
    22:17:29.991 : Z-probe:12.48 X:0.00 Y:0.00 zCorr:0.00
    22:17:40.508 : Z-probe:12.58 X:50.00 Y:0.00 zCorr:0.00
    22:17:51.165 : Z-probe:12.76 X:100.00 Y:0.00 zCorr:0.00
    22:18:02.933 : Z-probe:12.15 X:-100.00 Y:-50.00 zCorr:0.00
    22:18:13.137 : Z-probe:12.28 X:-50.00 Y:-50.00 zCorr:0.00
    22:18:23.451 : Z-probe:12.36 X:0.00 Y:-50.00 zCorr:0.00
    22:18:33.874 : Z-probe:12.48 X:50.00 Y:-50.00 zCorr:0.00
    22:18:44.375 : Z-probe:12.56 X:100.00 Y:-50.00 zCorr:0.00
    22:18:55.796 : Z-probe:11.67 X:-100.00 Y:-100.00 zCorr:0.00
    22:19:05.626 : Z-probe:11.99 X:-50.00 Y:-100.00 zCorr:0.00
    22:19:15.673 : Z-probe:12.13 X:0.00 Y:-100.00 zCorr:0.00
    22:19:25.878 : Z-probe:12.31 X:50.00 Y:-100.00 zCorr:0.00
    22:19:36.207 : Z-probe:12.37 X:100.00 Y:-100.00 zCorr:0.00
    22:19:42.458 : Info:Distortion correction matrix:
    22:19:42.458 : 0, -27, -28, -41, -63
    22:19:42.458 : -16, -9, -12, -24, -39
    22:19:42.458 : 6, 4, 3, -6, -24
    22:19:42.458 : 36, 23, 15, 3, -4
    22:19:42.458 : 84, 52, 38, 20, 14

  • M323 S1 P1

    should enable correction permanently. How and when it works depends on your configuration. I have set it for me to 100% correction until 0.5mm and 0% correction at 1mm.

    The distortion map also shows that your printer axis are not 120° apart. Between A and C axis you go up while the rest goes down. Changing angles e.g. A to 329.8 would change that. That is what I meant with printercalibration first. I use the hosts bed height map feature to calibrate this.
  • I don't know how to access the bed height map, sounds like there is an awful lot about this firmware/host I don't know. Are there any resources I could use to find this stuff out or a walkthrough for delta calibration?
  • No, there is a tutorial about z calibration, but even that is expired with all the new possibilities. Hard to keep up to date with all the work.

    It is in tools menu of host. It simply measues a rectangular area with z probe. You say whcih area and how many points. Go a few mm above bed for faster measurement and start. Then you get a height image and can copy the numerical results to clipboard.
  • ok, but from there how does one calculate axis offsets? is that just trial and error?
  • I can not calculate. It is try and error and feeling. If you increase angle between 2 pillars that area goes up. Already small values have an effect which explains why it is nearly impossible to get it exactly 120° if 119.8° already make a difference.
  • How do you control the bed leveling process? It takes my printer over and hour to complete it and it crashes the sliders in the max endstops multiple times during the movement so I don't trust the output?
  • Sounds very wrong. I always start them near bed so they keep awayy from max endstops. At the top the cylinder gets smaller and smaller so never do print operations over full width at that height or you get what you said. At the bottom you never have this problem.
Sign In or Register to comment.