Delta X-Axis Accuracy issues



  • Hey all, first off let me say how awesome this community is and how awesome the duet ecosystem is! I'm loving my duet wifi and smart effector!

    Ok, with that said I did come here because I am having an issue with my delta. It's a Tevo Little Monster I've converted to a Duet Wifi and Smart Effector with 360mm Haydn arms. It's printing BEAUTIFULLY, but things aren't quite dimensionally accurate. As it is printing in the +X direction, parallel lines will bow inward, and in the -X direction they bow out. This leads to walls and infill not touching. I can't find any build issues, as I have taken it apart and rebuilt about half a dozen times and can't get it to go away, and can calibrate to 0.019 deviation. I've rotated my arms around, and validated they are all VERY VERY close to the same length. Any ideas on what could be the culprit?!

    Pic #1Pic #2Pic #3Pic #4



  • Also I have pics, but can't seem to get them to show up in my post? lolHere's the google drive link



  • In 925 picture: did you check that the axis of the side where you get gap is prependicular and doesn't flex? Printing this line means the other axes are low and this one is high, so perpendiculary should be measured on the high position of the linear guide. You can check whether the guide bends somehow at this position.



  • I looked at your printer model. The aluminium extrusions are fixed on a plate and under a plate at the top, so if they have different lengths, the extrusions would bend. Just a possibility, you will find such a reason by measuring perpendularity.


  • administrators

    @joshc4566 said in Delta X-Axis Accuracy issues:

    Ok, with that said I did come here because I am having an issue with my delta. It's a Tevo Little Monster I've converted to a Duet Wifi and Smart Effector with 360mm Haydn arms. It's printing BEAUTIFULLY, but things aren't quite dimensionally accurate. As it is printing in the +X direction, parallel lines will bow inward, and in the -X direction they bow out. This leads to walls and infill not touching. I can't find any build issues, as I have taken it apart and rebuilt about half a dozen times and can't get it to go away, and can calibrate to 0.019 deviation. I've rotated my arms around, and validated they are all VERY VERY close to the same length. Any ideas on what could be the culprit?!

    Does it do the same thing if you print parallel lines in the Y direction? If so then it's likely that your rod lengths or tower steps/mm in config,g are incorrect. If you are using 7- or 9-factor calibration, reset the rod lengths in the M665 command back to the marked length and change to 6- or 8-factor calibration. Otherwise, check that you have entered the rod lengths and steps/mm correctly.



  • @dc42 Thanks so much for the help. It does not seem to be printing skewed at all in the Y direction. Just in the X. Would the steps/mm need adjusted for just the X Axis then? Also, I am 100% confident in the rod lengths being correct at 360mm.



  • @joergs5 Unfortunately I don't think that is the issue. I've taken all the towers off and laid them next to each other and the length is so even that there is no difference with all 3 being lined up with 2 machinist squares.



  • @joshc4566
    Did you consider warping of the filament? I know, the effect is only at one side and warping would occur at all sides, but maybe the print direction of the single printed lines differ and produce the effect being stronger at one side '). Warping could be strong at the walls because the walls will reinforce the effect by rewarming the lower layers. The weakest part is the connection of wall and low layer and internal forces will manifest at the weakest location.

    ') do you have an open window on one side cooling this side down more? 🙂



  • @joergs5 I really don't think it could be the filament warping. It has happened with multiple filaments, and multiple materials. The printer is locked away in a closet under the stairs like Harry Potter. Lol so no windows or anything like that that could cause a draft.



  • Can you find a systematic relationship between short/gaps with a specific position of the delta printer? If yes, there must be a difference between the axes or the print bed.

    If you want, you can try a detailed calibration:
    https://www.thingiverse.com/thing:2722557



  • In the think one commentator had a non flat bed at one corner. This is a possibility also.



  • @joergs5 it theoretically could be warped, but it's a circular mirror, and happens on all areas on the bed. And a G29 produces a pretty nice grid.



  • @joshc4566 I searched for "delta printer different x y" and found some information interesting:

    comments of hercek:
    https://reprap.org/forum/read.php?178,380688,381527

    comments of Doug LaRue:
    https://groups.google.com/forum/#!topic/deltabot/q65tcjykI4k

    Found additional possible reasons, e.g. that the J-Head must be mounted straight and you should calibrate with autoleveling off.

    Maybe you find something.





  • @joshc4566 a very interesing calibration guide, with Duet3D correction codes:

    http://www.sublimelayers.com/2018/03/musing-how-to-print-accurate-parts.html



  • @joergs5 that's a pretty awesome article. I might try that, but since my issue isn't consistent across the print bed, I think it will only help on smaller objects. I'm going to re-print some parts, tear it down and rebuilt again. Just to eliminate any discrepancies in the adapters for my magball ends I may have. Also, how important is 100% even belt tension?



  • @joshc4566 as far as I know, it is not important, because every belt has its own stepper and they are independent. It is important to have enough tension to avoid backlash.

    At CoreXY printers it is important because the XY steppers are correlated.



  • @joshc4566 my main bet is that the aluminium extruders are not perpendicular, i. e. they are not parallel to each other. The difference of the printed object is in the whole object, but is small enough to not being visible in the middle (because the rods are in middle height position), and at the edges one of the rods is at high position and error is maximized.

    To measure parallelity of the aluminium extrusions, you could use your linear guides: connect them pairwise with a horizontal shaft where the rods are connected, and move the guides up and down. If this stucks, you know that they are not parallel. Repeat it for all three combinations.



  • @joergs5 sorry for the huge delay! It's been a crazy busy month. I went and got married and had my honeymoon band have been out of the game for a month, but I'm ready to come back at it strong! I can't really seem to find much I'm the way of the extrusions being off from each other. I can't really connect the carriages to see if they bind, but using my metric tape and digital angle gauge, everything seems to be super square. I've even replaced the vslot wheels and bearings.



  • @joshc4566 Congratulations for you marriage, I wish you all the best!

    Regarding the printer, I would search for differences first:

    • is there play (backlash) in the connections
    • something not flat
    • lengths, even if sub mm
    • stepper speed settings (you can slow down printing in case one of your steppers looses steps). You can exchange the steppers (I mean exchange the existing ones) in case one is defect or inaccurate a bit
    • every pulley, belt etc. fixed
    • belt tensions

    Maybe this link and their sublinks are helpful:
    https://www.3dhubs.com/talk/thread/delta-printer-printing-dimensions-wrong

    There are Tevo Little Monster specific calibration on thingiverse:
    https://www.thingiverse.com/thing:2722557
    esp. the forum links with some additional ideas.

    And Duets own know how page:
    https://duet3d.dozuki.com/Wiki/Calibrating_a_delta_printer

    Hope something helps, as I cannot help you directly.

    PS I forgot: please make sure by using one of the calibration tools on thingiverse that your stl file is not the cause.... this would be a bit annoying.


 

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