Z height issues



  • Just finished assembling my Kossel XL from Think3dPrint3d with the Duet WiFi. Every thing was going fine up until I sent the G32 command to calibrate the bed. The hot end crashed into the glass bed( at least it did not crack or break), at first I thought the IR probe had not triggered but it had. I spent several hours trying different configuration settings etc but nothing helped it keeps running into the glass. What I eventually observed was that as the effector moves out from X0 Y0 the Z Height drops (the readout shows that the z height has not moved). at a distance of 100mm from the centre it will drop somewhere between 5 and 10 mm. This drop appears to be uniform no matter which direction I move from the centre (at first thought the bed was badly off square.

    Has anybody else had this problem or any idea of what I need to do to correct it?

    Regards,
    Tony


  • administrators

    You delta radius and/or rod length set in config.g is way out. First measure the rod length between bearing centres and put that in the L parameter of the M665 command in config.g. Next, with the effector centred, estimate the horizontal distance spanned by each rod (between bearing centres again) and put that in the R parameter of the same command. Finally, temporarily set the H parameter (dive height) in the M558 command to a large value e.g. 40 (this will make it start probing from 40mm above where it thinks the bed should be). Restart the Duet, home and auto calibrate.



  • Thanks DC42. My delta radius was out quite a bit.



  • David,
    Delta with 250mm DIA mild steel with glass top bed.
    When I used 10 point S6 calibration it was ok.
    Now switched to 13 point S6 as some builds need the whole extent of the bed.
    Trouble is extreme ends still extrude too close or too far from the bed so
    I get poor adhesion to glass bed covered in Morrisons white glue stick.
    So I ran a bedGetH.g with S-1 to record the height errors.
    This gave: -0.750, -0.605, 0.074, 0.532, 0.352, 0.143, -0.205, -0.702, -0.758, -0.443, 0.193, -0.080, 0.747
    So I put these exact values as the H parameter values in bed.g.
    This seems to make matters worse, not better.
    Is my problem that I should negate the values provided by S-1?
    Or is there some other deep underlying issue with my setup?
    Thanks
    Maurice


  • administrators

    The H parameters are intended solely to compensate for variations of Z probe trigger height with XY position. So don't use them unless you have measured the trigger height at each probe point.

    The height errors you recorded are very large. This suggests to me that you have a geometrical problem with your printer build, or insufficient range of motion of the joints.



  • To improve the geometry I'm re-building the spider with new rods and adding new timing belts.
    However I noticed that for the last 12 months I have somehow run with L=325mm
    when in fact centre to centre the rod/bearing is 335mm!!
    Perhaps that would account for my large Height errors 😞
    Am I right in assuming that the values returned with a bed.g having S-1
    are the corrections needed at each Z test point, and i do not need to do each one manually?
    If S-1 gives an H value of -0.010 then is this value to add to that point in the S=6 bed.g
    or should it be H=0.010 ?
    Ye old Newbie


  • administrators

    @Merlin:

    To improve the geometry I'm re-building the spider with new rods and adding new timing belts.
    However I noticed that for the last 12 months I have somehow run with L=325mm
    when in fact centre to centre the rod/bearing is 335mm!!
    Perhaps that would account for my large Height errors 😞
    Am I right in assuming that the values returned with a bed.g having S-1
    are the corrections needed at each Z test point, and i do not need to do each one manually?
    If S-1 gives an H value of -0.010 then is this value to add to that point in the S=6 bed.g
    or should it be H=0.010 ?
    Ye old Newbie

    No, it doesn't work like that. You can't infer variation in trigger height with XY position from the results or auto calibration.



  • From your last reply I now understand that the H factors only affect auto calibration and not bed leveling.
    To improve my 'geometry'as you suggested:
    Replaced my awful home made Traxass arms with new ones made by T3P3
    Replaced all 3, year old GT2 timing belts.
    Re-installed Robotdigg metal carriages
    Had an engineer check that my metal effector widths precisely matched the metal carriage widths.
    Z height issues are now even worse, that those previously reported. 😞
    S6 settles down to a factor of 0.220
    The Delta skeleton is all metal and looks spot on and solid.
    IMHO The only 'geometry' left that could be at fault is the carriage alignment with the carriage trucks, affecting effector tilt.
    I don't have the carriage spacers installed as per your 'Big Delta' notes as I'm using T3P3 printed carriages
    that i believe do not need them?
    I'm still on firmware 1.15e - will the new features in version 1.17 correct for my carriage/effector tilt?
    Or is it just hit and miss to adjust them?
    Or have you any other suggestions?
    Thanks


  • administrators

    The reason I use carriage spacers is that I am using Igus joints, and without them the rod ends foul the carriages and the effector. The Traxxas joints don't need them because they already have a flange that spaces the joint from the other parts.

    I suggest you measure the Z probe trigger height at the G32 probe points, and also check whether any variation is due to effector tilt. There are a few of ways of checking for effector tilt. One is to mount a bulls-eye spirit level on the effector. Another is to lower the nozzle until it touches the bed or grips a sheet of paper, then use feeler gauges to measure the height of the bottom edge of the IR sensor above the bed. If there is no tilt and the bed is flat, the height should be constant regardless of XY position.


Locked
 

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