Delta printer calibration always off just a touch at the same angle

  • Ultibots D300VS (delta style) running a Duet Wifi at latest firmware 1.18.1. Interlink Electronics FSR w/ JohnSL FSR Endstop Controller…

    Prints come out ok but i notice that the G32 calibration is always off just a bit at the same angle. I say this because on one side of the first layer the lines have just a bit of space between them and on the other side the layer is obviously squished a bit. I run the G32 before each print and each print i see this same thing in the same direction.
    Adding mesh compensation removes that problem but then tends to have the same "issue" but in different areas each time G29 is ran.
    Since im calibrating each print and it seems to be calibration that's off a bit i wasn't sure if there was anything in my config to share.
    My bed file is generated using the wizard at (13 points, 6 factors, probing radius: 120)

    As an additional note, it just dawned on me that the mesh compensation map shows this "error"... Not 100% sure what that means yet though... Other than useful for visualizing the problem and confirming when its resolved.

    Thoughts? hehe

  • What kind of Z probe are you using?

    Do a "paper test" at each of your G32 probe points. (move to the probe point, set the nozzle to just touch the paper, set Z0 with G92 Z0, move up 5mm, run a G30 S-1). See if the probe height changes per probe point. If so you either have mechanical inaccuracy in your build, such as tilt (this shows up more with offset probes like an IR probe) or something else affecting your probe trigger height (can happen with probes that use the nozzle, too). You ideally want to fix it mechanically so your probe triggers at the same height across the surface. Failing that, put the probe offsets into bed.g.

  • administrators

    It sounds to me that you may be getting an inconsistent Z probe trigger height. Using a lower probing speed (F parameter in the M558 command) may help. If you are using FSRs, then one FSR may be more or less sensitive than the other two.

  • Probe only in between the towers where the FSR's are instead of at the towers (between the FSR's) like the automatically generated file does. Takes more movement downward to trigger and is less repeatable at the towers. I use 3 sets of 3 points moving inward and it is so much better than the bed.g file that comes from Ultibots or the automatically generated one. I now get .01 to .005 deviation. All this is of course assuming your glass is flat.

  • cr8tor,

    I ran into the same situation with my T3P3 Kossel XL for a couple weeks (2 issues resolved a week a part)

    1. Pulley not tight enough (slipping)
    2. Pulley not on correct position of shaft (too far in or out)

    Once I corrected those build / mechanical issues, flawless with near consistent 0.017 std dev between calibration runs; no bed mesh compensation needed.

Log in to reply