Exploring Extrusion Limits - Soft (64D) TPU

  • Morning All,

    After some confusing issues with extrusion multiplier I took things right back to basics and have printed a lot of 20 * 20mm, 10mm high squares with a single pass of the extruder and checked the wall thickness against what Sli3r was aiming for.

    For the sake for this discussion please assume my measurement of the filament is fine, and the extruder calibrated. My processes for both are evolving, but getting closer to something that is as consistent as possible.

    TPU processes slowly and seems to create a lot more load on the extrusion system than PLA/ABS/PET-G. From my limited experience on both this mirrors how it processes on industrial single screw extrusion systems.

    My process was to start with 10mm/s and adjust the extrusion multiplier until the wall thickness (measured 2 points on each side with an external micrometer) was within 0.05mm.

    10mm required a slight reduction to the extrusion multiplier and looked opaque .
    15mm required no extrusion multiplier and looked a little less opaque.
    20-30mm appeared very clear and required extrusion multipliers ranging from 1.17 through to 1.33.
    35mm started to look overheated or over extruded before reaching the target wall thickness.

    I decided to pick move to a wall thickness of 0.7mm and the extrusion multipliers from the previous 0.4mm wall all needed to be increased. I saw the same distortion on the walls trying to reach 0.7mm with 30mm/sec that I saw with the 0.4mm walls.

    So sticking with 20mm/s I decided to keep the extrusion multiplier at 1 and go up in temperature to see if the assumed reduction in melt viscosity would improve the situation, and it did. However the temperatures required to reach the 0.7mm wall thickness resulted in the part looking opaque again.

    Finally I came across the auto speed function on Slic3r, but found the volumetric rate required to get target wall thickness was low, at 2.25mm3/s but this was the limit where both the 0.4mm and 0.7mm walls were accurate.

    Have I missed anything? Tried tightening the extruder screw but the stepper started to skip. (current limit as high as possible). I've a bigger stepper on order!

    Seems like to get the best clear appearance on the parts you need to push into the zone where the polymer melt/extrusion system behaves in a non-linear manner. What are people's opinion? Should we be looking for a extrusion multiplier correction table based on requested volumetric extrusion?

  • David has already implemented non-linear extrusion via the M592 command which will be appearing in the next release. I am using it already and it works well. So far I have used it for PLA, PETG and TPU.

  • Cool, I'll await the wiki page before asking much more about it then. I presume what I was discussing was the driver for it's development?

  • Links for others who stumble across this post:

    G-Code Wiki
    Reprap Forum Post (also linked on above page)
    Duet Forum Post

  • Yes, good links. Here's what I discovered that made me interested in this topic and it explained why I was struggling to get uniform extrusion on walls/skin/etc.


    This guy was ahead of the curve.

  • Thanks for that. I've found TPU is a real slow material compared to the others. It's pushed me into the needing to process faster zone when I'd much rather be working with bog standard parameter sets.

    Didn't say in the original post but spec is:

    0.4mm nozzle.
    E3d V6 Hot end.
    250mm of bowden tube
    E3d titan extruder with E3d's standard motor (current high, temp 52C)

  • I love these threadsā€¦investigative 'tweaking' and pushing the limits to find the low/high limits.

    Thanks for doing the research and sharing...Keep it coming!

  • Changed spool and it threw out my figures. Foolishly thought diameter would be similar with the same batch number! I'm going to adopt the following procedure when setting up materials from now on. I checked the gcode and Slice3r was calling for extrusion widths varying from 0.4 to 0.7 for 0.2mm layer thickness.

    • Best effort extruder calibration

    • Measure filament with micrometer looking for light drag of the filament in the micrometer jaws

    • Run a 20mm square, 10mm high cuboid with 1 vertical shell and 0% infill to get a thin wall

    • Set all extrusion widths to 0.4mm and run a range of eight parts with increasing volumetric extrusion rate.

    • Do the same for 0.7mm

    • Measure wall thickness on all parts taking enough samples on each part to get an average.

    • Compare the calculated extrusion multiplier that would be required to correct the 0.4 and 0.7mm wall thickness to the correct width. Pick the volumetric rate that has the most similar required correction for the 0.4 and 0.7mm wall thickness. Apply the calculated extrusion multiplier

    • rebuild the 00.4 and 0.7mm walls with the chosen parameters.

    The 20x20x10 samples take minutes each. Normally takes me longer to measure two points on each side than it does the printer to print them!

  • Since doing the above I've noticed the muck that accumulated in the extruder has all but disappeared. In retrospect this was an obvious sign that I was pushing things far too hard.

  • I've just realised that I think I'm struggling with something else here.

    Yes, I can see a non linear relationship between requested volumetric output and a required correction on a 0.7mm wall.

    But; I'm currently using autospeed on Slic3R and the problem is at the same requested volumetric output 0.4mm walls need a greater correction than 0.7mm walls. The 592 feature won't help here. Because I am requesting the same volumetric rate the formula from the duet wiki will apply the same correction:

    (1 + min(L, Av + Bv^2))

    Any more hints apart from have a little filament bonfire in the car park! šŸ˜‰

    I'm getting tempted to force Slic3R to use the same extrusion thickness for all the different types of deposition (e.g. perimeter, infill, outer/inner, etc.). I can bodge the extrusion multiplier to get that right and it should make something thenā€¦

    Titan Extruder
    E3D V6 Hotend
    0.4mm Diameter Nozzle
    0.2mm Layer Thickness

  • I've been testing volumetric extrusion rates from 1.6 through to 3.4mm3/sec.

    I'm beginning to think the opaque appearance is the polymer overheating. I had a filament jam late last week that was cause by over extrusion on the first layer blocking the path out of the nozzle for the polymer on the second layer. With a bit of a shove it started feeding again for the third layer, but in a white colour and almost frothy.

    I'm currently thinking that;

    • Running slowly increases the time the polymer spends in the nozzle and it over heats and begins to bubble slightly giving an opaque appearance to the think walls, and white to the parts.

    • Running fast reduces the time in the hot end and so reduces the ultimate melt temperature, and makes clearer parts.

    I've been increasing the nozzle temperature to reduce polymer viscosity, but perhaps the polymer is just particularly crap at taking heat from the hot end. Maybe a longer melt zone at a lower temperature would have the same effect without risking over heating the melt?

    I had shear thinning floating around in the back of my brain but at the higher speeds, particularly on the thinner walls the parts didn't only look bad but there were holes in the walls, so if that was a thing for this polymer and I hadn't reached the extrusion rate where that occurs it was well in excess of where I am working.

    Edit: Recommended processing temps were 230-250Ā°C and the polymer does ooze out of the hot end at 230Ā°C.

    Edit 2: I think I may also be suffering with the polymer having next to no melt strength.

Log in to reply