Why cant Auto Calibrate actually calibrate End Stop positions?



  • Im using a Duet IR probe, Duet2 Wifi (2.05.1), E3D Chimera on a modified Anycubic Predator.

    Despite many changes, anything near the X tower its printing way too high off the bed. On the Y side its squashed down to scraping point. At this point I'm strongly suspecting the terrible endstops and endstop "flags" (to break the optical signal) that Anycubic supplied are the problem. I am measuring height differences on all 3 endstop flags.

    Today I went old school and instead of Auto Calibration (S6, S8, diff diameters tested hundreds of time) I manually measured and set offset via M666 positions directly infront of each tower. I finally get these measurements when probing (G30 S-1):

    M666
    Endstop adjustments X0.86 Y-1.20 Z-0.11, tilt X0.00% Y0.00%

    The auto calibration (which of course overwrites my custom M666) puts the endstop adjustments at figures much smaller and completely incorrect. Like Y is is auto calibrating to around +0.09 instead of-1.20mm !?

    Thinking I could correct this by adjusting the "H" value in Bed.g I manually calibrated all 16 points (following Sublime Layers guide. Annoyingly, this makes for no improvement!?

    Which my crude, manual tweaks to M666 I does indeed start to print nearly uniformly across the board. But why is auto calibration failing so much with the End Stop calculations?



  • @Kodachrome you would need to post your Config.g and Bed.g files as a bare minimum to get any sort of useful feedback?


  • administrators

    @Kodachrome said in Why cant Auto Calibrate actually calibrate End Stop positions?:

    Im using a Duet IR probe, Duet2 Wifi (2.05.1), E3D Chimera on a modified Anycubic Predator.

    Does the IR probe have a large X offset relative to the nozzle? If so then the errors you are getting are almost certainly caused by the effector tilt varying with XY position. See https://duet3d.dozuki.com/Wiki/Calibrating_a_delta_printer. It's useful to mount a circular spirit level on the effector to show this up.



  • Well I have twin E3D Volcano's so there is only so close I can put the probe to the bulky and hot hotends.. but I had a similar thought tonight and I've edited my effector platform and am reprinting it now.

    Previous IR probe offsets:
    G31 P500 X28 Y18 Z1.5

    Diagonally positioned (for reasons, to get fan is nice spot), the new effector will have it 9mm close to ![alt text]Anycublic 713 Style effector Smaller - bottom.png

    Anycublic 713 Style effector Smaller .png

    Do you have any insights on the effect differing EndStop trigger points (between the 3 stops) would cause?



  • Here is a website that has nice graphics showing delta miscalibration effects (endstop shift results under the Tower Zero heading) http://boim.com/DeltaUtil/CalDoc/Calibration.html



  • @oliof said in Why cant Auto Calibrate actually calibrate End Stop positions?:

    Here is a website that has nice graphics showing delta miscalibration effects (endstop shift results under the Tower Zero heading) http://boim.com/DeltaUtil/CalDoc/Calibration.html

    Thanks.

    And now you know why I gave up Delta's in favor of "XYZ" machines.

    Frederick



  • @fcwilt I once started with a Delta and I must confess this was a mistake some years ago, because as beginner it is better to start simpler. It has it's strengths, I am sure, e.g. for speed, but it's better to have some expertise already. My opinion.

    I then had a Prusa, and although it was immediately defective (heating of bed), I still had some initial successes.



  • I started with a delta too (-; but now I build even weirder machines that are harder to tame.



  • .. so deltas are a gateway drug?!

    I think bad hardware and firmware have given deltas a bad rep, fix the hardware and give a god rep (rap firmware) and you'll be amazed.

    My friend got a kossel mini eons ago, struggled with inductive probes and calibration, never worked right and collected dust for years. Last week(?) we spent 2 hours wiring in a Duet 2 Maestro and squaring up the frame, and less than an hour to configure and (auto)calibrate using manual probe / paper method to eliminate effector tilt. My friend was in awe by the simplicity of the Duet and RRF compared to RAMPS and Marlin, not to mention the near silent printing.

    My experience is similar, but I got it working until the non v-slot profiles wrecked the v-wheels and its sitting in a box pending new carriages to be made for new v-slots..



  • @oliof said in Why cant Auto Calibrate actually calibrate End Stop positions?:

    I started with a delta too (-; but now I build even weirder machines that are harder to tame.

    Yes you do and they are fascinating. The kind of project I love to watch someone else dive into.

    Frederick



  • @JoergS5 said in Why cant Auto Calibrate actually calibrate End Stop positions?:

    @fcwilt I once started with a Delta and I must confess this was a mistake some years ago, because as beginner it is better to start simpler. It has it's strengths, I am sure, e.g. for speed, but it's better to have some expertise already. My opinion.

    I got mine to work but they were mini-d's with a limited build diameter and they took a lot of work to get everything just so.

    Then I built my first non-d, the FT-5. Once I saw how much easier it was to get first rate results and a large build area I was a convert.

    Then I equipped it with a Nimble and I didn't see that d's had anything more to offer for the kinds of things I print.

    My MarkForged style printer has even less moving mass than the FT-5. For some reason the MF style just appeals to me.

    Frederick



  • Aside from Resin SLA printers, my entire history with 3D printing is delta's (5years?). I guess Ill go easy mode on the next big build and try out Railcore type design. Ive had no real problems with smaller delta's but the Predators huge build volume (and dubious end stops) have proven problematic from day 1. So much potential though.

    On this problem I've made yet another IR Probe mount, super slim and just 1mm gap between the mount itself and the hotends (which are in silicone socks). The IR probe is another 3mm further out.. this is realistically as close as its going to get. I'll test it later when more filament gets delivered!



  • After moving the IR Sensor closer (again) to the middle I did indeed get much much better results.
    Makes be wonder.. do the Nozzle(s) have to actually be in the middle?
    I could really stretch the design and put the IR Probe in the middle (which is as close to eliminating effector tilt as it gets) and offset the nozzles, by maybe 3mm.

    Thoughts?



  • @Kodachrome said in Why cant Auto Calibrate actually calibrate End Stop positions?:

    I could really stretch the design and put the IR Probe in the middle (which is as close to eliminating effector tilt as it gets) and offset the nozzles, by maybe 3mm.

    the kinematics probably have to think it is, applying tool offsets would probably let you move it but at the end of the day its the distance between the nozzle and the sensor that causes tilt to affect the calibration.

    use the nozzle as a sensor with a smart effector or piezeo sensor (or other solutions to probe with the nozzle) to rule out the tilt (or just true up the mechanics if at all possible)



  • Nozzles, not nozzle, so cant use the Smart Effector (as per my other thread, thanks for the response there too, appreciate it).
    On the piezo sensor, not sure how to physically integrate on a E3D Chimera .

    I did have JohnSL FSR's setup too, to compare. Definately gave different results, but cant say for certain it was more accurate. I think FSRs are not a great solution for this size of bed.
    Its maybe not as pressing as before, Im printing my 5th sensor holder now, seeing how close I can get before the PETG takes heat damage. Printing is gone way up in quality. I also changed the crap Optical sensors to the old SeeMeCNC mechanical switch endstops.



  • So to update my own experiments, I made an easily attachable mount (magnets) for the IR probe and mounted it directly below the nozzles. Its dead on centered but the results from Auto Calibration are kinda the same at 6 Factor and useless at 8. Rear right of the bed is about 0.25 too high and front left is -0.15.

    Deviation is ~0.110 .

    So far:

    • changed effector platform from metal to custom PETG one
    • changed all Endstops from inaccurate optical ones to tried and trusted mechanical endtop switches (from SeeMeCNC)
    • moved IR probe close to center, about 10mm off nozzles
    • then moved IR probe directly below nozzle (G31 X0 Y0 Z35)

    Literally none of these changes make any actual difference. The same error positions occur all the time? What else am I missing?



  • To add:

    • stripped the Whambam bed off, back to bare clean aluminium plate.
    • added 3 more bolt down washers to snag the aluminium plate into the frame even tighter
      Got the deviation in Auto Cal down to "0.043" but...

    Literally none of these changes make any actual difference. The same problem positions (rear left vs front right of bed) occur all the time? What else am I missing?