Print dimensions off



  • Now I understand that the whole kit I purchased is only about $250 max retail price. Injection molded parts, in high volume cost cents to produce. From what I've learned in my precision machining course is that injection molding, depending on the tooling, can be very consistent. I know it's not the level of precision you'd get from CNC Milled parts, is +/- 0.1 mm too much to ask of it?? . I'm a little confused about the setup of that…I digress.

    I've checked my setup and was careful assembling my Kossel Mini.

    I still cannot get a 20x20x20 mm cube! And I'm struggling to understand why. I broke a rod a while back and had to remake one of them, at the time I was careful to reproduce it and when the epoxy was drying it was in between two other rods to ensure it was the same length.

    My cube comes out
    X19.66
    Y19.54
    Z20.65

    Then I printed a 40.0 mm cube

    38.60
    39.00
    40.40

    Could my IR sensor not be exactly perpendicular to the bed? Or my bed not be level? Could the calibration keep getting messed up by an issue like that? Should I be extruding more out of my nozzle? It's only about 0.45 out of a 0.4 nozzle. I find it difficult to find in depth guides on delta 3D printers. I don't know how to use one of these and it's been 6 months of tinkering!

    This one is only to learn about these printers, I'm designing parts to cnc mill every single part then I'm going to use very precise Faulhaber Servo motors, if I can, to control my next printer, if I have to I'll use their stepper motors, I'd prefer servo.

    I have good PLA, an e3d hot end, e3d titan, robotdigg belt tensioners, and robotdigg effector. I've tried spending money where it matters and I feel I should be able to get better accuracy despite using 1.8° nema 17 motors and using a frame from a kit that was less than a $250.

    I do not understand why my dimensions are not closer to spec. Could someone help? This is frustrating.

    Thank you in advance.



  • Okay so given your x and y are reasonably consistent (check your calipers I had a cheap set which measured wrong, that really doesn't help) you need to change your diagonal rod length to slightly shorter, your under by 2% so shorten rod length by 2%. Then recalibrate and print again. That should get x and y to 20+/-0.1mm.

    Your z could be over extrusion and given that it's not proportionally the same for a 20mm cube as a 40mm one that's worth checking. Or more likely your steps per mm are slightly too much. Adjust this first before rod length and recalibrate.



  • My steps per mm are 100 because I use 1.8° stepper motors and I use 16 tooth pulleys and what I believe to be GT2 belts. Default 16x microsteppung should be 100….right?

    My M665 is:

    M665 L236.051 R116.386 H271.223

    So you're selling me to take off 9 factor and use, what 6? To prevent the rods from changing?

    And make my L231.330?


  • administrators

    You are correct on the steps/mm for those axis. Maybe DjD meantnthe extruder steps/mm which if too high gives over extrusion which can cause the top layer to bulge give a higher Z number



  • I would do a 6 or 8 factor Calibration and do a test print with L set to as close as you can measure the centre to centre of your Rod Pivot points.

    Then do the Measurements.

    Z is affected by Steps/mm and x and y by the rod length.

    So as Dj says if after you do this you side are short then reduce the L value and try again.

    Bare in mind tho that the Plastic does Shrink when it cools and different plastics do so at different rates so you have to decide what is important to you as you can allow for the Shrinkage rates in the Slicing software. (ie if you know the different shrink rates you can allow for it in the Slicer so you need to get it as close as possible taking shrinkage into account)

    I Would not allow the Firmware to adjust the Rod Length for you as that is better done by Careful measurement if possible or by calibration as above.

    Hope this makes some sense

    Doug



  • According to e3d's documentation for their titan extruder, I calculated that my E steps should be 418.5 for a normal 1.8° stepper

    See here: http://wiki.e3d-online.com/wiki/Titan_Assembly

    If it is recommended that I change my E steps, should I do the same percentage calculation?



  • @Dougal1957:

    I would do a 6 or 8 factor Calibration and do a test print with L set to as close as you can measure the centre to centre of your Rod Pivot points.

    Then do the Measurements.

    Z is affected by Steps/mm and x and y by the rod length.

    So as Dj says if after you do this you side are short then reduce the L value and try again.

    Bare in mind tho that the Plastic does Shrink when it cools and different plastics do so at different rates so you have to decide what is important to you as you can allow for the Shrinkage rates in the Slicing software. (ie if you know the different shrink rates you can allow for it in the Slicer so you need to get it as close as possible taking shrinkage into account)

    I Would not allow the Firmware to adjust the Rod Length for you as that is better done by Careful measurement if possible or by calibration as above.

    Hope this makes some sense

    Doug

    What Slicer do you use that takes shrinkage into account? I use Slic3r and have never seen that setting.



  • The best way to calibrate an extruder is to remove the Hotend
    extrude enough filament so you can get a clean edge on the output of the Extruder.
    Command it to extrude 100 mm
    Measure how much comes out and adjust your Steps/mm accordingly so for example if you have 100steps/mm set and you extrude 100 mm and you get 105 out then you have to reduce your steps/mm by 5%

    I think I have my Titan set to about 405 but not in a position to check at the moment you just have to try it and adjust.

    Then once you have the theoretical set you can adjust the Extrusion factor in the DWC to fine tune further.

    HTH

    Doug



  • Thank you I will….what's DWC stand for??



  • Just to clarify yes do 6 or 8 factor (might as well do 😎 but change rod length manually (don't forget config_override.g) to adjust xy dimensions. Only two factors (directly) affect your objects z height, over extrusion or incorrect steps/mm. Use the calculated steps/mm for your titan but fine tune using a printed object (plenty on thingiverse) as the pinch wheel pressure and softness of filament affects the effective gearing.

    I totally agree that it should be motors, pulleys and belts that determines tower steps/mm but if you're extruding the right amount of filament and objects are always the same % too tall you need to change steps/mm. Incidentally a lot of people swear by 80.5 steps/mm with 1.8 motors, 20t pulleys and gt2 belts. But in your case you might (once extrusion is ruled out) need to lower yours.



  • Duet Web Control



  • Okay X and Y seem to be consistently +/- .05 and Z is 20.50-20.60 depending on how hard I push my calipers together. E steps /mm are what I'm supposed to change to shorten z?



  • Lower E steps per mm will extrude slightly less filament and therefore its a test to see if you're over extruding. Reduce it by 1% and then print it again. You can do so from paneldue/web interface by reducing extrusion /extruder to 99.



  • If this doesn't solve the problem or improve it at all try lowering your x y and z steps/mm by the % that the part is too tall, then autocalibrate again and print it.

    If you want a more precise method try this object I designed. http://www.thingiverse.com/thing:1619976
    Split the model in slic3r and print one cube at 15% infill and the other at 95%. The 15% one shouldn't have stringy infill and the 95% one should have tiny gaps between the infill lines after layer 6-7, theres detailed instructions.



  • @NoSkillzEngineer:

    @Dougal1957:

    I would do a 6 or 8 factor Calibration and do a test print with L set to as close as you can measure the centre to centre of your Rod Pivot points.

    Then do the Measurements.

    Z is affected by Steps/mm and x and y by the rod length.

    So as Dj says if after you do this you side are short then reduce the L value and try again.

    Bare in mind tho that the Plastic does Shrink when it cools and different plastics do so at different rates so you have to decide what is important to you as you can allow for the Shrinkage rates in the Slicing software. (ie if you know the different shrink rates you can allow for it in the Slicer so you need to get it as close as possible taking shrinkage into account)

    I Would not allow the Firmware to adjust the Rod Length for you as that is better done by Careful measurement if possible or by calibration as above.

    Hope this makes some sense

    Doug

    What Slicer do you use that takes shrinkage into account? I use Slic3r and have never seen that setting.

    Simplify 3D allows you to adjust the Horizontal Dimensions % for each individual type of filament or even by brand if you set up a process profile for it!


  • administrators

    You can also use the M579 command to scale the axes, see https://duet3d.com/wiki/G-code#M579:_Scale_Cartesian_axes. However, if your diagonal rod length or steps/mm is off, you are likely to find that when you print long straight lines, they come out slightly bent.



  • With slicer I often factor around 0.7% dimension change for ABS.



  • Hello,

    I have a kossel xl with duetwifi board. If I print a square that has a square hole in it the outside dimensions are perfect but the internal square is undersize by .4mm. If I then print a round disc with a hole in it the outside and inside dimensions are undersize by .4mm. Can the M579 command be used to fix this issue?



  • If I print a square that has a square hole in it the outside dimensions are perfect but the internal square is undersize by .4mm. If I then print a round disc with a hole in it the outside and inside dimensions are undersize by .4mm.

    Maybe this can help you:



  • I print everything with an extrusion multiplier of 90% which gives me the best dimensional accuracy and the best surface finish. No idea why it works but for me it does. My ageing Mendel prints everything slightly small compared to my newer CoreXY - same gcode files, same print settings, same filament and hot end end design. Again, no idea why. I recently discovered that with two parts which dovetail together for a sliding fit with negligible play, making them both in black worked well but making the male part in gold coloured PLA gave a loose sloppy fit. Changing the clearances so that printing the male part in gold PLA gave a nice smooth sliding fit, then printing the same male part in black PLA resulted in a too tight interference fit. Both filaments measured spot on at 1.75mm and of course, this was the same gcode file printed on the same machine with exactly the same settings, so yet again no idea why. I think that where dimensional accuracy is critical, it's just a matter of trial and error to get a perfect fit and what works on one machine, may not work on another.


Log in to reply