Extrusion calibration?



  • Okay so now I'm trying to tune the prints a bit more.

    This is a few of the pictures of my 20x20x20 cube and the corner dimensions are actually quite close to 20mm (19.94-20.04) and I used M579 to scale the Z down about half a mm.

    Quick question, M579 require an M500 or is that a one time command that the Duet remembers? I forget.

    https://postimg.org/image/qioair1hp/

    https://postimg.org/image/fj33770st/

    https://postimg.org/image/t001q3t4t/

    It looks half way decent from the images but there bits here and there that go over the corners but it almost looks like for most of the print it created a tiny radius rather than a sharp corner or maybe those are areas of over travel? It's not necessarily consistent across the features, however, is absolutely consistent across parts.

    https://postimg.org/image/npv2yptfh/

    Two different materials to get an idea of repeatability of dimensions, in case someone is curious.

    The corners are close to the desired number of 20 mm, however, when measured in the middle of the cube as opposed to the corners the number is up to 0.18 mm under.

    What should I do to tighten up these dimensions?

    I know this is referred to as shell, wall, or perimeter and the extrusion width should directly affect this, as will several other settings. Rather than dive in and ignorantly tweak and alter settings and waste time and material I prefer to approach this as intelligently as possible. I have a hard time finding advanced tuning guides, it seems as though it is still a bit of a black art.

    What could cause the issues (as I expect it is more than one) that I am experiencing?

    I dialed up the current to 1.5A despite not knowing my maximum rated current, I'll be honest…I don't care about the motors one bit. As long as I don't fry the drivers, I'm content. The additional amperage definitely helped, in fact it almost sounds quieter if you can believe that.

    Should I upload my slicing profile? Config file? What would give more insight to the cause of these issues?

    I briefly looked for advanced tuning tutorials/guides and did not come across much that looked beneficial to this, any and all input is welcomed. I'm not touching a setting until I hear from someone that knows more than I do. I'll read, I'll tinker, I'll measure and test whatever I need to...I just don't know what to do.

    Thank you in advance. Ya know, I spent so long learning how to build a 3D Printer only to realize I severely neglected to learn how to use it and tune it. ? Anyone else do that? ?


  • administrators

    To use M579 permanently you need to put the M579 command in config.g.



  • @dc42:

    To use M579 permanently you need to put the M579 command in config.g.

    Ooo…Okay, that is good to know. I added it to it. That's probably something that should be added to the section on G-Code documentation, that might be a little useful in that section:)



  • 20mm is a bit on the small side to try and calibrate dimensional accuracy, i would say 100mm at the min but the more the better, you also need to be careful that you don't include any print defect/ringing, bed squash or over/under extrusion in your measurement.

    For example if your part is 19.5mm but needs to be 20 and you adjust the M579 scale to make it 20mm you may end up with a 200mm part that is off by 5mm. It could be that if you had not changed any setting the part would still be off my only 0.5mm if it was 200mm long.
    To put it another way it could be that your nozzle is moving the correct distance but its just the printed part that is not perfect due to other reasons not related to the XY scale or steps/mm.

    My current DIY CoreXY printer is using spectra line and 3d printed spools so it hard to calculate step/mm, the way i did so was to measure the actual carriage movement using calipers.

    1. Power on motors and move carriage close to a edge.
    2. Measure distance and zero out the calipers.
    3. Command a 100mm move.
    4. Measure new distance.
    5. Take Commanded distance divided by measured distance then multiply that by my steps/mm. The output is the new steps/mm to be used.

    So if my commanded move is 100mm measured move is 103mm and my steps/mm started as 150.35 it would look like this:
    100/103=0.9708 Then 0.9708*150.35=145.97. So the new corrected steps/mm is 145.97.
    Again this is measuring the carriage movement and not the part, it can be done by measuring a printed part you just need to print it as big as you can and correct for any print defects. One way to correct for this is to print a small part say 10mm then most of the error you see should be the part itself and not the move. (unless its way off) Then subtract that error from the big part.



  • I'm curious how the stepper driver handles fractional steps. It would seem like eventually something gets rounded off?



  • Has to be doesn't it you can't really move 13.5 microsteps. But then at 1/16th with 100 steps/mm a half step rounded up or down is only 0.005mm (5 microns). So 13.97 for example is 14 all day long.



  • @DjDemonD:

    Has to be doesn't it you can't really move 13.5 microsteps. But then at 1/16th with 100 steps/mm a half step rounded up or down is only 0.005mm (5 microns). So 13.97 for example is 14 all day long.

    Yes, you're right. But why is the middle of the cube 19.85 when the edge is 19.97?
    I'm thinking the perimeter extrusion might not be the right size/amount or something but the corners are more or less fine, I don't get that.



  • @Kezat:

    measure the actual carriage movement using calipers.

    1. Power on motors and move carriage close to a edge.
    2. Measure distance and zero out the calipers.
    3. Command a 100mm move.
    4. Measure new distance.
    5. Take Commanded distance divided by measured distance then multiply that by my steps/mm. The output is the new steps/mm to be used.

    So if my commanded move is 100mm measured move is 103mm and my steps/mm started as 150.35 it would look like this:
    100/103=0.9708 Then 0.9708*150.35=145.97. So the new corrected steps/mm is 145.97.

    I've been meaning to do this for each individual axis. I am using linear rails so I suspect this will be quite easy, except for the distance. I could actually do this at school and use a very precise machine to measure the distance between probe points to a few tenths (in inches, I don't recall the metric last place, maybe thousandths?).

    My problem with that is that their Wi-Fi is locked down and I'd have to do it through a screen that I don't have. After 1" my calipers tend to drop off a thousandth or two after 2"

    There is a big vertical measurement device at school on an expensive granite slab, I should buy a screen and go to school to use that.

    That solves the movement problem.

    Now if the extrusion lines such as each width the nozzle lays down is not what it should be….What in earth do I do about that?? I want to go through every setting to get accurate dimensions and I'm not entirely sure how to go about verifying things are accurate. Ugh. This might take a while.


  • administrators

    @NoSkillzEngineer:

    But why is the middle of the cube 19.85 when the edge is 19.97?

    Two possible explanations:

    1. You have over-extrusion at the corners of the cube. Using pressure advance may help.

    2. Your printer is a delta and either the tower steps/mm or the rod length configured in firmware isn't quite right.



  • @dc42:

    @NoSkillzEngineer:

    But why is the middle of the cube 19.85 when the edge is 19.97?

    Two possible explanations:

    1. You have over-extrusion at the corners of the cube. Using pressure advance may help.

    2. Your printer is a delta and either the tower steps/mm or the rod length configured in firmware isn't quite right.

    Okay, is that in any particular order?

    I do suspect over extrusion on the corners, pressure advance still confuses me, so I will try the recommended S value between 0.1 and 0.25 and see what happens.

    Gage pins, (not sure if every one knows these, I didn't before last year) which are precision machined then precision ground to be a specific labeled diameter to verify if a drilled, bored, or reamed hole is the correct size…or to simply figure out the size of an unknown hole. I used gage pins to line up my carbon fiber rods with ball joint style ends and they were all the same length (within about 0.002" at the maximum, which hopefully is not where my error is coming from). I put my rods all together (they weren't perfect but could be persuaded to go onto the same pin) and measured the outside of the pins with a 12" Certified Fowler dial caliper and converted by multiplying by 25.4 (taking into account the pin dia/2).

    I am uncomfortable changing my delta arm length as I feel I measured it within about 0.002" at the maximum but I think I'm within 0.001" of an accurate measurement. My mindset is to change this if all else fails.

    I would believe the tower steps per mm are incorrect, the extruder steps per mm were off more than I expected using a 60 mm Nema 17 with an E3D titan, which should be pretty close but it ended up being several mm off.

    I left the machine printing unattended and something happened and I came back to one belt broken, the linear guide rail all the way off (nothing preventing it from coming off but the belts and the software) and the majority of the ball bearings all over the place. I still haven't touched it since it happened Thursday night.

    Once I get it back together, grease it, and verify each axis carriage moves across the linear rails the amount it should (I think there's a guide for this in the documentation) then I'll calibrate it and try printing something again.

    Thank you, I'll post my results later today.
    I almost suspect over extrusion regardless of my Steps/mm but one fix at a time! 🙂



  • Given your measurement technique, I would also not change the arm length.

    Also, regarding "A screen I don't have" to fine tune at school: Don't forget USB and a laptop!!



  • You did measure the centre line of the Balls didn't you when measuring the Rod Lengths? (Are we meaning Traxass type ball joints or Magnetic Ball joints as per Haydn's Mag Ball Arm's)

    Doug



  • @Danal:

    Given your measurement technique, I would also not change the arm length.

    Also, regarding "A screen I don't have" to fine tune at school: Don't forget USB and a laptop!!

    Actually…..I did completely forget about this! I only ever use the USB connector on the Duet Wifi when I mess up and hit the erase instead of reset (whoever changed the location of the erase button for the newer boards, thank you...wish it happened sooner haha), when the firmware gets corrupted and I have to reflash it and initially set up Wifi, and to power the printer.

    Since I mainly use the USB connector to power the board, I completely forget that I can legitimately connect to it like this. Thank you, I did absolutely forget about this ha!



  • @Dougal1957:

    You did measure the centre line of the Balls didn't you when measuring the Rod Lengths? (Are we meaning Traxass type ball joints or Magnetic Ball joints as per Haydn's Mag Ball Arm's)

    Doug

    This is the type of connection I have on my Kossel: https://www.robotdigg.com/product/449/Tie-Rod-Ends-for-Kossel

    I was hesitant to use the mag ball arms and thought a system like this would be more precise and easier to measure correctly. I put every rod on two gage pins, which are these:
    http://www.starrett.com/metrology/product-detail/metrology/metrology-products/precision-measuring-tools/fixed-gage-standards/Pin-Gages/S4010-832

    And pushed them together so they were touching each other. Since I knew exactly how large each pin is, down to the thousandth of an inch, I simply measured the length of the outside of both pins, as well as the inside length to verify my numbers were accurate, (that were located in the ball joint holes) and subtracted the (diameter of each pin/ 2) to find the radius to get the distance between centers. This was a little weird to do, it was the most precise way I could think of measuring the distance without machining my own jig (which I really really don't want to do, but can if I absolutely have to).



  • and you measured at both sides of the stack tp eliminate any possible tilt in the gauges (I do know what type of Gauges your referring to I used to use them a long time ago and I would normally call the go-nogo hole gauges as normally they are double ended so one should fit and the other not.

    What I mean is are you certain that the gauges where parrallel rather than like \ / (Exaggerated)


 

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