[SOLVED] Underextrusion at 0.1mm but not at 0.2mm layer height

  • Hello gentlepeople!

    I'm having a hard time understanding this issue. I wanted to print something at 0.1mm layer height and noticed that some layers seemed under extruded and the overall structure of the object was very weak and broke easily.

    I decided to do some calibration prints. I printed a 2-wall cube. Each wall was 0.8mm thick, so all good. When I did the retraction calibration test print something strange happened: the two cyllinders seemed more like spiral skeletons than cylinders. The same test print works fine with a 0.2mm layer height.

    I tried changing the following settings (one at a time). Here are the results:

    • retraction: print got worse when I disabled it or reduced it from 3mm to 1mm due to lots of stringing and oozing, I think.
    • Combing: no effect
    • Seam: no effect
    • Temperature from 200 -> 220: no effect
    • z-hop disabled/enabled: no effect
    • acceleration: print improved when I went from 800 to 3000 mm/s2 for the outer wall
    • jerk: not much difference between 15 and 25 mm/s for the outerwall jerk
    • replaced nozzle with new one: no difference
    • replaced stock extruder with titan extruder: no difference
    • different brand PLA: no difference

    My theory is that the filament is under extruded right after a retraction + move + re-prime. I use cura as slicer and all my print speeds are 40mm/s, travel speed is 150mm/s. Accelerations are 3000 mm/s2 and jerk is set to 25mm/s for all features. Retraction is set to 3mm at 20mm/s (20 mm/s because higher than that the motor starts singing when retracting with the titan extruder).

    Below are some picture of the prints. The first one is printed with 2walls. The second one is printed with 3 walls.


    My config file: config.g

    I'm open to all ideas. Again, this only happens when I print at 0.1mm layer height. The printer used to be an Anycubic Mega i3

  • Have you forced the extrusion thicknesses or is the slicer basing extrusion thickness on layer thickness and it falling below or very close to nozzle diameter?

    I'd see if you get the same effect on a simpler single wall geometry with no base if possible. Very quick to build check and repeat. Check what the extrusion thickness is of the part vs what the slicer was aiming for.

    I've had trouble with extrusion thickness dropping below nozzle diameter, but to be fair that was user error. Thought I had a 0.4 nozzle and it was 0.5!

  • @doctrucker Extrusion thickness is set at 0.4mm. I did a 2-wall hollow cube and measured the wall thickness. It's 0.79-0.80mm, so close enough.

    Since yesterday, I've been messing with "extra prime amount" and I got one good print. I'm printing something else now to see if it does the trick.

    Thanks for your input.

  • On the squares I do the walls each always have a consistent thickness, but each are different, ie the one may be 0.55 +/- 0.05 and another may be say 0.6 with a similar tollerance. Build two in sequence and difference in the thickness of wall 1 between both parts will be smaller than the range in thicknesses of each wall on one part. Does that make sense, do you see something similar?

  • What are your E steps/mm set at? I suspect it is low, due to an ungeared extruder. Is this the case?

    The segment length combined with the layer height and extrusion width might be too low for the resolution of your E axis.

  • @doctrucker I will test that. So I'll print 2 hollow cubes next to each other and label the respective walls on each cube (i.e. left wall on cube 1 with left wall on cube 2 will be labeled "A" for example).

    Then after I print them at the same time, i measure the thickness of the wall A on cube "1" and the wall A on cube "2". The thicknesses should be within acceptable limits. Do I understand that correctly?

    I normaly do 2-layer walls for the hollow cube test. Should I try a 1-layer wall as well?

  • @bot it's 407 steps/mm. You are saying that the g-code asks for so little filament that the motor doesn't even move?

    If yes, then I think that's not the case. The motor visibly spins and at 45mm/s retraction speed it sounds like a zipper when you close it very fast. I lowered the retraction speed to 20mm/s because it got annoying.

  • Can you post the gcode file that showed the problem? I'd be curious to see what the minimum print move distance is. If the move is less than 1 microstep of the extruder you can get missed steps. The extruder just ends up not extruding. I've notice this when printing at 0.05 layer height.

    You can combat this a little bit by setting the extruder to a higher microstepping. I find anything over 64 causes hiccups. You don't have to calculate new steps if you're already using 16. Just add a new microsteps setting command for the extruder below the existing one. The firmware will calculate the new value for you.

  • @llamazen 407 doesn't seem terribly low to me, unless you're printing with a .25 mm extrusion width or smaller.

    For reference: to be able to reliably print small details down to 0.08 mm layers with a 0.25 mm extrusion width, I've found that more than 1000 steps/mm is necessary. With a 0.4 mm nozzle, the requirements are much less.

  • I did a test print today with 0.075mm^3 extra prime after retraction and it printed OK.

    It's weird though because I never had to use it before. I got the number from the E3D nozzle diagram for my 0.4mm nozzle.

  • @phaedrux said in Underextrusion at 0.1mm but not at 0.2mm layer height:

    You can combat this a little bit by setting the extruder to a higher microstepping. I find anything over 64 causes hiccups. .........................

    Just a "for info". The extruder moves much faster when doing retract/unretract than it does during normal print moves and that's what causes the hiccups. So if you are able to retract at slower speed, then you can push the micro stepping up a bit further. The step pulse frequency limit is about 200,000 so if you take the steps per mm, multiply that by the micro-stepping then divide the result into 200,000 it'll give you maximum extruder speed that you can use.

  • @deckingman thanks for that Ian.👍

  • @deckingman @Phaedrux Sorry guys, I don't quite understand what you're saying. I currently set the microstepping of the extruder to 16 interpolated.

    As far as I know, this sets the microstepping 256 virtual steps. I have tried setting removing interpolation but the motor gets noisy and whiny like it did with the cheap drivers.

  • @llamazen So the idea is that if the print move asks the extruder to move a distance shorter than the minimum distance of a single microstep the extruder motor won't move at all.

  • @llamazen I see where you are coming from and why. Unfortunately in this instance the term "interpolation" is a bit misleading (in fact, it's a lot misleading). ☺

    All it means is that when it is used, the driver chip will divide the micro step pulse into 16 smaller step pulses. But, it's always 16 smaller step pulses. The step pulse from the Duet is an on or an off, a 1 or a zero, a high or a low. There is no in between. I'm not sure what happens inside the firmware but I'd imagine there will be some sort of rounding happening. So if the calculated movement amount required say 10.6 micro-steps, then it would be rounded up to 11 pulses. Conversely, if the calculation resulted in 10.4 micro-steps then it gets rounded down to 10 pulses. Whatever happens, it's always a whole number of micro step pulses. So "interpolation" doesn't give any better positional accuracy. From a positioning point of view 16x micro-stepping with interpolation is exactly the same as 16X without interpolation and the only way to get finer resolution is to use higher micro-stepping.

  • @deckingman OK got it. I'll do a print with 64 and 256.

    Thanks guys for the input.

  • @llamazen said in Underextrusion at 0.1mm but not at 0.2mm layer height:

    @deckingman OK got it. I'll do a print with 64 and 256.

    Thanks guys for the input.

    Stick with 64 to start with. At 256x you'll likely run into the maximum step pulse frequency of 200,000 Hz that I was talking about (unless you use very low retraction speed). Run M122 during the print and check for any reported hiccups which should zero or near zero.

  • If you let me know the steps per mm @16x, I can calculate the maximum extruder speed that you can use for retraction at any other micro-stepping (or you can do that calculation yourself using the info I posted earlier).

  • @llamazen Sorry, I've diverted from the original subject to something that I had measured on a recent test. I think it is probably best that I spin that off to a another post that I will write now. I doubt what I am seeing will directly effect your work, but I suspect it may have a little to do with retraction and advance settings.


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