Really bad calibration results



  • I've been resurrecting whatever remained of my old trusty Mini Kossel kit to a semi functional state and ran into some issues I'm having trouble to explain/deal with.

    Basically, this - 0_1557524898235_2019-05-10 23_44_54-Window.jpg

    The printer has a set of 3 under bed piezo sensors, so there shouldn't be any influence from the effector tilt. It's clear there's some underlying mechanical issue here, but I'm at a loss what it could be. I've tightened what could be tightened (pulleys, belts, screws, delta arms...), the bed itself is a cast aluminum plate flat to 0.1 mm. If anyone has any ideas what could be the cause of that, I'd be really grateful.





  • @phaedrux Yep, although 1 thing I haven't done was put springs/rubber between each pair of rods. Not sure if that could cause that much of an effect, but will do that nonetheless.



  • what factor calibration did you use?



  • it seems like a combination of
    1 position of one tower to high and
    radius to short

    http://boim.com/DeltaUtil/CalDoc/Calibration.html



  • @veti Wow, that's a really nice reference page. I'm running a 6-factor auto calibration, so both endstop positions and delta radius should have been accounted for.



  • just making sure. you did save the settings from the auto calibration in your config.g?



  • @veti Yes, with config-override disabled for simplicity's sake.

    The plot thickens... Once I allow the auto-calibration routine to adjust delta rod length with 9-point calibration, I get decent calibration, with 0.030 deviation and an almost perfectly flat bed on a G29 probe. A 8-point one nets a 0.4 deviation and some monstrous distortion.
    I'd be pretty happy with the former if it didn't adjust my diagonal rod length all the way up to 267 mm. Which is really different from the real life measurement of 215 mm. That's getting confusing.



  • 267 to 215 is nearly 25%

    that could be the difference between using a 16 tooth pulley with 100 steps/mm and a 20 tooth pulley with 80 steps.

    so which pulley are you using and what steps have you configured?



  • @veti You're absolutely correct! No wonder that my homed height seemed higher than it should have been.
    Thanks a lot! Case closed - I've configured steps per mm for 20-tooth pulleys while running 16-tooth ones.


 

Looks like your connection to Duet3D was lost, please wait while we try to reconnect.