Inconsistent Extrusion?



  • Hi!

    I have almost finished my printer based on Duet 3D Wifi. Everything is almost perfect! Except one problem which I think, may be a problem with Inconsistent Extrusion, but I'm really not sure at the moment.

    alt text
    alt text
    alt text
    alt text
    alt text

    It's a CoreXY printer with falling bed. Everything is moving on V-slots. Extruder: Titan Aero (chinese) with e3d pancake 0.9 deg. motor. Hotend temperature is stable, I'm using PT100 sensor which works great.

    What i have tried:

    • Changing Extruder current (from low to high)
    • Changing Z motors current (from low to high)
    • Changing filament tension (from low to high)
    • Loosing threaded rod backlash clearance screw
    • Trying different flexible coupler mounting ways
    • Changing belt tension (from low to high)
    • Disabling retraction
    • Disabling Z-hop
    • Different microstepping modes on extruder
    • Different microstepping modes on Z axis motors
    • Changing Extruder Jerk (from low to high)
    • Changing Z Axis Jerk (from low to high)

    Unfortunately none of above attempts make any change in the surface quality.

    I'm out of ideas at the moment. I hope that You guys can give me some ideas where I should be looking for the solution.



  • I have the exact same problem on my CoreXY (not that much, though). I didn't find yet how to fix it... I will follow this thread to see if there is an obvious reason, of if it is a design issue.



  • I had a quite similar problem (although on a delta machine).
    Solution was to lower the extrusion factor using
    M221 S95 D0
    (in this case to 95% for Extruder 0)



  • @themelle Thanks, I will give it a try, but I think that it's not the case. I have calibrated everything (flow, e-steps etc.)



  • Same for me: I have a perfect top layer.



  • @themelle I think that reducing flow rate made amplitude of this noise smaller. But it's still there. REducing it more will give me underextrusion :X



  • i had to go to 87%.



  • Please test forcing the layer seam to a single point in the slicer and re-print.



  • This is my default setup...



  • @phaedrux Tried it while ago, but it looks the same. It's a bit wors where the seam is starting.



  • Yes, but do the walls look more consistent when the seam is started at the same location?



  • @phaedrux A bit. But it's not a huge difference.



  • alt text

    If you consider this part right circled in red, where a perimeter wall started in these two places, especially in the upper one you can see the line a little thicker coming into the seam and a little thinner coming out of it. If the nozzle is moving from left to right in this photo (I don't know if it is or not) this looks like a classic example of where pressure advance may help. Assuming the nozzle is moving left to right at these seams, it looks like pressure could be a little too high as it decelerates into the end of the line, and too low as it speeds up out of the start of the line. Have you tried experimenting with pressure advance?

    If the slicer has different layers moving the nozzle in different directions (two layers left to right, two layers right to left, etc.) and there are effects that pressure advance could help with that could explain the alternating patterns you see in the first of the photos you showed.

    You could try creating a single hollow cube, with seams aligned, and print it, and every few layers input a higher pressure advance figure and see if your corners and the start/end seam point look different. Example pressure advance command would be "M572 D0 S0.15", with pressure advance set to 0.15. You might try printing the single-walled cube starting with pressure advance at 0 and then every few lines manually run the g-code to set pressure advance, incrementing by 0.05 or so, say from 0 - .20, and see how that effects the corner blobbing and the seam start/end points.



  • @sethipus Thank You for professional answer. To be honest I was experimenting with pressure advance. I have been testing a few different tests to achieve best corners etc. also I was using two different tests to set proper value. If I remember right, my value for it was S0.46, or S0.42. Take a look on the last photo there are muuch thicker lines visible which excludes PA problem.

    This is really good attempt but I think there is some mechanical problem with the extruder. Yesterday I have found that it was leaking and my titanium heat break was bent! :C I'm going to rebuild it with some higher quality parts from e3d which I have ordered today. Also, I will assemble it which muuuch more caution.

    I will give You info guys how the results are looking after reassembling.



  • Good luck with your rebuild. Note that 0.4 is a very high value for pressure advance. If you read this Duet FAQ article on pressure advance it says that moderate bowden systems might use 0.1, long bowden systems using 0.2, and direct systems as little as 0.05. Have you testing with settings in the 0.05-0.2 range?

    Also, I and many others prefer to align the seams so that most of the print looks good and fairly uniform, and only that one seam looks bad. In S3D I'll usually position a print so that either a corner or the back of an object is oriented with point I specify as the one to align on. Also makes it easy to run tests like pressure advance on an object like a single-walled cube, with the seam on one corner.

    Btw, I'm running some extremely challenging prints right now for a 3D-printed radio-controlled airplane, with complex parts using single-walled printing. After many test runs I've had to crank my pressure advance up to 0.25, which is higher than I've ever used before. I've got a very long (perhaps 650-700mm) long bowden tube. The bowden system is so "flexy" inside, given how much pressure is required for the filament to make the required flow rate, that I have to use a full 5.4mm of retraction. If you consider that a direct drive extruder would probably use maybe 1mm under similar conditions, that implies I get over 4mm of filament compression in my bowden tube, and all of that has to be retracted/unretracted with every nozzle move, complicating the start and ending of lines, etc. I've got my extruder speeds cranked up really high to help with this, and thankfully the BMG clone's dual extruder gears and my stepper seem up to the task. I really wonder how much easier (and better) my prints would be if I just went direct drive.



  • @sethipus My mistake, I mean S0.046 or S0.042 - I'm on direct setup 😉 Have You tried Capricorn bowden tube? It has 1.9 mm inner diameter and it's much more precisely made than standard white bowden tube. So Your system will be less flexible. I was using it above 1 year ago when I was using bowden 3D printer and it worked a bit better. But the bowden was short, about 40 cm, so on larger tube like Yours it might help You more than me.

    That's the reason why I have build my printer with direct extruder. It's more accurate, less pressure problems etc.



  • @petrzmax yes I'm using Capricorn tubing in fact. I recently replaced my single-drive extruder with dual drive, and replaced the bowden tube at the same time because the one I was using before was so short that when the print carriage was in certain regions of the build plate I got very tight radius curves, which causes other problems. When I replaced the tube I went deliberately long, with the plan to cut it down a little after I've got some other things taken care of. I think I can cut down the tube by a good 100-200mm without getting too-small curve radius, since I also re-oriented the tubing a little. Hopefully when I cut the tubing down my bowden problems will reduce a little. Btw, I just had to stop a print of this wing panel and start a new one. I thought that the 0.25 pressure advance had solved my problem, but the problem was back. It's a combination of factors. What I'm seeing specifically is this single-walled print having underextrusion in the first few centimeters after the start of a new layer, leading to the layers in that region being incomplete. It's not a problem I see in my solid prints, but these single-wall complicated sparse parts are really tough to print well.



  • This is a stab in the dark, but can you please try setting M566 P1 ? This sets the jerk policy to mimic a more Marlin like way of doing things, that applies the jerk value to more move types. At least that's my understanding.

    I assume you're running firmware 2.03.



  • @sethipus said in Inconsistent Extrusion?:

    Good luck with your rebuild. Note that 0.4 is a very high value for pressure advance. If you read this Duet FAQ article on pressure advance it says that moderate bowden systems might use 0.1, long bowden systems using 0.2, and direct systems as little as 0.05. Have you testing with settings in the 0.05-0.2 range?

    You do know the values on the wiki are too conservative right? This has been stated very often.
    For example, for my 100mm Bowden volcano for ABS i need about 0.25, and for the same thing in PC i need 0.68
    I suggest you tune the values according to this thread, not what the wiki states for some empirical values.
    https://forum.duet3d.com/topic/6698/pressure-advance-calibration



  • @phaedrux I'm using this command M566 P1. Models on photo were printed with this setting. Sure I'm on 2.03 😉


Log in to reply