Delta auto calibration deviation goal



  • Getting close to running my first delta print but I'm still a bit wary that I've navigated things correctly.

    I'm using a smart effector, haydn magballs and magball rods with a duet Maestro. My first point of confusion was setting the Z trigger height...should this not remain at -0.1 for a smart effector?

    I ran the auto calibration sequence (G32) a few times and received the following:
    1:28:47 AMG32
    Calibrated 6 factors using 16 points, deviation before 0.022 after 0.018
    1:27:07 AMG32
    Calibrated 6 factors using 16 points, deviation before 0.030 after 0.024
    1:25:46 AMG32
    Calibrated 6 factors using 16 points, deviation before 0.022 after 0.020
    1:21:51 AMG32
    Calibrated 6 factors using 16 points, deviation before 0.024 after 0.023
    1:19:31 AMG32
    Calibrated 6 factors using 16 points, deviation before 0.029 after 0.026
    1:17:57 AMG32
    Calibrated 6 factors using 16 points, deviation before 0.025 after 0.022
    1:15:43 AMG32
    Calibrated 6 factors using 16 points, deviation before 0.024 after 0.019
    1:13:48 AMG32
    Calibrated 6 factors using 16 points, deviation before 0.216 after 0.019

    After saving:
    M665
    Diagonal 450.550, delta radius 227.137, homed height 407.124, bed radius 185.0, X 0.253°, Y 0.297°, Z 0.000°

    M666
    Endstop adjustments X0.13 Y-0.22 Z0.09, tilt X0.00% Y0.00%

    Few additional questions

    1. I'm also wondering if for the M665 setting I should have used multiple L values since my Haydn rods were measured at 450.55, 450.54, 450.56. I ended up using 450.55 only.

    2. Why would I not want to use mesh bed compensation in conjunction with auto calibration?
      G32 followed by G29 in my startup sequence.

    I'm quite excited to get printing but a bit confused as to what good values are vs bad values coming from the cartesian world!



  • These seem like sufficiently narrow deviations to me (see below). I run a 600mm (24") delta and what I settled on is to do a calibration, an M500, and a mesh every so often (particularly if I work on the printer), and do a calibration only in the start G-Code of every print.

    I use a single value for L. Not right or wrong, just what I do. I also NEVER run the number factors in a calibration that can change L; my experience with that makes me shy away from it.

    I run a bed.g generated from:

    bed.g file for RepRapFirmware, generated by Escher3D calculator
    ; 16 points, 8 factors, probing radius: 280, probe offset (0, 0)
    

    And config.g (or override) entries of:

    M558 P5 R0.4 H5 F500 T18000     ; P5=Effector.  R0.4 = Recovery time after each probe. H5 = Height from which to probe. F500 = Feedrate mm/min T18000 = rapid between probes, mm/min
    G31 P100 X0 Y0 Z-0.1            ; Set Z probe trigger value, offset and trigger height
    M557 R300 S75                   ; Define mesh grid
    
    ; config-override.g file generated in response to M500 at 2019-01-18 16:32
    ; This is a system-generated file - do not edit
    ; Delta parameters
    M665 L720.280 R328.965 H626.773 B304.0 X-0.173 Y0.125 Z0.000
    M666 X1.176 Y-1.179 Z0.003 A-0.50 B0.11
    

    Resulting in a height map of:

    RepRapFirmware height map file v2 generated at 2019-05-25 13:51, mean error -0.065, deviation 0.040
    xmin,xmax,ymin,ymax,radius,xspacing,yspacing,xnum,ynum
    -225.00,225.10,-225.00,225.10,300.00,75.00,75.00,7,7
          0, -0.080, -0.135, -0.070, -0.046,  0.026,      0
     -0.066, -0.065, -0.074, -0.062, -0.075, -0.074,  0.034
     -0.103, -0.073, -0.106, -0.090, -0.094, -0.041, -0.052
     -0.044, -0.065, -0.040, -0.089, -0.085, -0.090, -0.059
     -0.105, -0.066, -0.075, -0.081, -0.086, -0.059, -0.120
     -0.145, -0.058, -0.013, -0.038, -0.067, -0.081, -0.131
          0, -0.102, -0.005,  0.011,  0.022, -0.024,      0
    

    You can see that with my large-ish bed, I'm satisfied with final deviations in the .04 range. You are getting things about half that. Looks good to me.

    P.S. Looks like five months since I bothered to do an M500 after a calibration (since I do one every time) and about a month since I did a mesh.


Log in to reply