Inconsistent Extrusion with Hermes / Hemera

Discussion in 'HotEnds & Extruders' started by Kraken 3D, Jan 25, 2020.

  1. Kraken 3D

    Kraken 3D Active Member

    Joined:
    Jan 25, 2020
    Messages:
    27
    Likes Received:
    0
    Thank you for chiming in - I will have to look over the information you linked. I have tried turning off stealth chop on the extruder and there was no noticeable improvement and it might have actually worsened.
     
  2. Thorinair

    Thorinair Well-Known Member

    Joined:
    Aug 8, 2017
    Messages:
    50
    Likes Received:
    7
    The salmon skin is not a Hemera issue. It is a physics issue. If you want to get rid of it, buy an extruder which isn't as precise.
     
  3. Kraken 3D

    Kraken 3D Active Member

    Joined:
    Jan 25, 2020
    Messages:
    27
    Likes Received:
    0
    I guess I have to slightly disagree there. I understand what you are implying however I don't think the Hemera is so precise the extrusions are rough and uneven along with the moire pattern and or salmon skin. Salmon skin is not really the issue here. There is a pattern but salmon skin is not linear. All of these inconsistencies are linear in nature. I would find it quite the opposite. Possibly the short filament path as you mentioned surely doesn't help with nozzle pressure but that is something that should be compensate-able via firmware I would hope; yet not sure how to address it.

    Some drivers suffer from salmon skin, TMC are not one that typically does. Adding TL smoothers on the old Ender boards helped that issue but again I don't see the correlation of salmon skin in these prints. This issue is mostly identical to the Prusa 602 plague. And as far as I know there was no fix for that and it went ignored. The only thing as I mentioned before was linearity correction in the Prusa firmware but I do not think that carries over to Marlin 2.0.
     
  4. Kraken 3D

    Kraken 3D Active Member

    Joined:
    Jan 25, 2020
    Messages:
    27
    Likes Received:
    0
    All good info- Not sure if I mentioned it before but when I tried to use SpreadCylce the prints were just as bad if not worse. However I only changed that value for the extruder only.

    With Marlin is there a way to set the StealthChop to not switch to spreadcycle? Or at least see where the current values are?
     
  5. Thorinair

    Thorinair Well-Known Member

    Joined:
    Aug 8, 2017
    Messages:
    50
    Likes Received:
    7
    The salmon skin absolutely is the source of inconsistency. The reason why there is no pattern is because of the infill and everything else that happens on every layer differently. Print a single walled cube (with curved edges) and no infill, and you will instantly see that this inconsistency suddenly aligns and creates a pattern.
     
  6. SchmartMaker

    SchmartMaker Active Member

    Joined:
    Jan 13, 2020
    Messages:
    44
    Likes Received:
    11
    I use a Duet with RepRapFirmware, but a search for Marlin options gave me M913 to set the hybrid threshold speed, M569 to set the TMC stepping mode and M906 to set the driver current. So with Marlin there should be some basic options to tinker with the drivers.

    I've been digging through some datasheets and application notes, and as far as I can tell, it seems that with TMC2208, TMC2209, TMC2224, TMC2660 and the likes, there's no actual microstepping table to adjust, just the chopper control registers. So the trick with Linear Correction for the TMC2130 will not carry over as is.

    The Prusa 620 issue essentially revolved around the inherent imperfection of a stepper motor, and extensive community research was done to improve the monotonic behavior i.e. microstepping uniformity of the standard Prusa LDO 1.8° stepper motors, but to also try various brands and step angles. I think the conclusion was that in general 0.9° steppers gave best results, if they didn’t feel too ‘notchy’. BTW, based on their recent Facebook posting, E3D might also be using an LDO 1.8° stepper on the Hemera, which may or may not contribute to the issue you’re seeing. Anyway, the stepper motor on the Hemera is a given for now.

    I must admit that I’m now also seeing some faint salmon skin on PETG prints, though less pronounced compared to yours. I use the Hemera with TMC2224 on 800-900mA. I recently replaced the stock Creality stepper on the X-axis with an unknown brand 0.9° stepper, which improved VFA quite a bit. That made the salmon skin more noticeable. I tried some stuff, and when I decrease current on the extruder to below 600mA, the pattern becomes more chaotic. When I increase current to above 1000mA the pattern does not change but amplifies VFA only on the right side of the hollow test cube I’m using. So increasing the Hemera stepper current somehow affects artifacts normally caused by idlers or the Y-axis stepper (still the original Creality 1.8°). Go figure that one... Crosstalk? Who knows. As I mentioned before, speeding up the print does help with the salmon skin a little in my case, but I did not find a setting yet that really makes a significant difference. I tried various stepper current settings, interpolation enabled/disabled, 1/4-1/256 microsteps, PSU voltage, StealthChop/SpreadCycle, belt tension, print temperature, etc. Most of it tested live during printing (the advantage of a Duet with RepRapFirmware). I may go down the rabbit hole and try some different chopper setting next and report back.

    Surprisingly enough, with the original Creality mainboard and its crude and noisy A4988 drivers, I never really noticed VFA. And with the 350mm Bowden tube and genuine BMG never saw salmon skin on my prints. But that setup had its downsides as well. Apparently one can’t have it all.
     
    #46 SchmartMaker, Feb 27, 2020
    Last edited: Feb 27, 2020
  7. Synapsis

    Synapsis Member

    Joined:
    Apr 12, 2018
    Messages:
    2
    Likes Received:
    0
    I have the same problem, I changed my Titan Aero hoping for an improvement but now its worse. I had to lower my e steps 390 tried different speeds, jerk acceleration etc. When I try to do a load filament it will start then it will whine and stop extruding ( the white is flush)
    but if I do extrudes like 5 mm consecutively there is no problem.
    Take a look!

    https://drive.google.com/drive/folders/1cCLrJYcrfZx8TYaLV3_dsPZuKszcblpc?usp=sharing
     
    #47 Synapsis, Feb 28, 2020
    Last edited: Feb 28, 2020
  8. SchmartMaker

    SchmartMaker Active Member

    Joined:
    Jan 13, 2020
    Messages:
    44
    Likes Received:
    11
    Thanks for sharing! Your pictures show the issues well, but I'm not confident that I'm seeing salmon skin or inconsistent extrusion patterns on your prints yet. The calibration cube you use changes print direction from CW to CCW a lot, so that's not ideal to bring out salmon skin. The best way to test for the issue the OP has, is to print this object.

    I think your prints do show some other issues, for example with overhangs, which could be mild overextrusion, printing too hot or poor part cooling if that's PLA you're using. There's also slight VFA present. Additionally some ringing/vibration from high speeds or high acceleration, so you could try printing slower. The elephant's foot could be from printing too hot or too close to the bed. And perhaps you haven't tuned retractions or linear/pressure advance fully yet, which could alleviate the deeper horizontal lines I see. I can't really explain the sagging above the 'Y' other than the plastic was too hot or didn't get sufficient time to cool or wasn't forcibly cooled.
     
  9. Synapsis

    Synapsis Member

    Joined:
    Apr 12, 2018
    Messages:
    2
    Likes Received:
    0
    Thanks SchmartMaker for the advise, will try printing the object in your link and checked the other things you said.
    I will try changing the extrusion in the slicer, I was using .90 multiplier width .40, speed is only 2800 mm/min. and retraction 1 mm.
    And yes it the first layer was too close to the bed.
     
  10. Uwe Heinritz

    Uwe Heinritz Member

    Joined:
    Mar 19, 2020
    Messages:
    6
    Likes Received:
    1
    Hello,
    I bought a Hemera in December and I have similar problems. On all printouts horizontal lines are visible.
    To eliminate these I have already made these changes to the printer:
    -installed higher quality LM8UU bearings and greased them correctly
    -Stabilized supports of the print bed
    -Replaced controller of the Tronxy by a SKR 1.1 Pro with TMC2209 drivers
    -Replaced belts and pulleys for X/Y with high-quality ones from Gates
    -installed Z-spindles with a smaller pitch (2mm)
    -Integrated anti-backlash nuts for Z-spindles
    -printed different types of filaments and with different speeds

    None of the changes have affected anything on the printouts.
    I also checked the gears of the Hemera and could not find any damage. The filament roller is mounted on ball bearings and the PTFE-tube to the extruder I left out for testing.

    Because I could not think of another cause for the problem, I created a test object (narrow stairs) for testing the extruder.
    Through the test I could clearly see that the distance between the lines increases as the volume per layer decreases.
    The increasing distances of the lines are visible in the attached picture (I have drawn them in black for better visibility).
    As a crosscheck I determined the volume per line/wave for each section of the staircase. For this purpose I divided the volume of a section by the number of waves of the section.
    The extruded volume of each wave is between 30-31mm³ in all sections. With a filament diameter of 1.7mm this volume corresponds to a filament length of 13.1-13.5mm. This should be approximately half a turn of the Hemera's filament gear.

    Any further ideas how I could get rid of this problem?
     

    Attached Files:

  11. Daniel Rock

    Daniel Rock Well-Known Member
    Staff Member

    Joined:
    Jul 9, 2018
    Messages:
    97
    Likes Received:
    10
    Can you remove the antibacklash nuts? and see if the problem goes away, just use standard POM nuts.
     
  12. Uwe Heinritz

    Uwe Heinritz Member

    Joined:
    Mar 19, 2020
    Messages:
    6
    Likes Received:
    1
    Hi,
    I did many prints without these nuts (the tronxy X5SA came with normal brass nuts and spindles with 8mm pitch.
    I tried to remove these lines with installing new spindles with 2mm pitch (did not help) and later with installing antibacklash nuts (did not help also).

    Did you look at my image? You can see that the distance between the lines increases as the volume per layer decreases. If the nuts should be the source for the problem, the distance between these lines would be even over the whole print or total randomly.
     
  13. Uwe Heinritz

    Uwe Heinritz Member

    Joined:
    Mar 19, 2020
    Messages:
    6
    Likes Received:
    1
    Hi,

    to exclude the remaining hardware of the printer as a source of error, I did a new test print last night.
    Using the data from the stair print I estimated how big the circumference of a cube has to be for the filament volume of a wave to be exactly one layer. This should make the stacked lines approximately equal.
    As volume I assumed 30mm³, which should correspond to a filament length of about 13.4mm (should be 1/2 turn of the extruder gear wheel).
    Like all test objects before, the cube was printed with 3 contour lines, no filling and 0.16mm layer height.
    The result shows much fewer lines, which are also much less pronounced. The reason for the remaining lines could be the only estimated print volume of a wave, or the Z-spindles.

    On the attached picture I have compared one edge of the stairs and one of the new test print. As a visual aid I have drawn a black line in each case. The better print result (bottom test part in the image) is clearly visible.
     

    Attached Files:

  14. SchmartMaker

    SchmartMaker Active Member

    Joined:
    Jan 13, 2020
    Messages:
    44
    Likes Received:
    11
    You've looked at the print volume as a possible cause of the repeating patterns. But in your tower, the higher it goes, the lower the layer time is. Did you try printing at a different speed to exclude this variable as a potential cause? E.g. printing at half speed should not show changes in the pattern, and the pattern spacing should be the same in each test print you've done.
     
  15. Uwe Heinritz

    Uwe Heinritz Member

    Joined:
    Mar 19, 2020
    Messages:
    6
    Likes Received:
    1
    Hi, yes I have tried different speeds and also different filaments (PETG, PLA). The pattern always looks the same.
    The pattern of the lines does only changes when I adjust parameters which affects the print volume per layer (Infill, number of perimeters, layer height, size of the object).
     
  16. SchmartMaker

    SchmartMaker Active Member

    Joined:
    Jan 13, 2020
    Messages:
    44
    Likes Received:
    11
    @Thorinair, just curious, do you have a multimeter to measure the resistance of both coils of the Hemera stepper motor? I'm seeing a 0.1 Ohm difference in resistance between coil A and B. I'm fairly sure that doesn't help either.
     
  17. SchmartMaker

    SchmartMaker Active Member

    Joined:
    Jan 13, 2020
    Messages:
    44
    Likes Received:
    11
    Okay, that's clear then. I've seen misaligned gears because the two bolts that secure the heatsink to the stepper motor weren't fastened sufficiently. But you've taken the Hemera apart already, so I assume you fitted the heatsink tight and flush. I know that E3D finds lubrication unnecessary, but I'm wondering if it would hurt to put some 15000 CPS differential oil on the gears... Your filament diameter is consistent? And did the bearings of the hob assembly turn without grinding or resistance?
     
  18. Uwe Heinritz

    Uwe Heinritz Member

    Joined:
    Mar 19, 2020
    Messages:
    6
    Likes Received:
    1
    I pressed the heat sink flush to the motor before I fastened the bolts. The filament diameters is consistent (meassured) and could not be the source of the Problem (way to even pattern). I did not see any problems with the gears and hobs when I reassembled the hemera. All gears are Looking good. The motor sound of the hemera is very even during the extrusion of a longer piece of filament (e.g. extrude 200mm filament with 200mm/min).
     
  19. SchmartMaker

    SchmartMaker Active Member

    Joined:
    Jan 13, 2020
    Messages:
    44
    Likes Received:
    11
    Uwe, you've probably only changed the extruder and wiring, but just to rule out artifacts from an expanding and contracting heatbed, have you tried turning it off halfway if you're using a heated bed at all?
     
  20. Uwe Heinritz

    Uwe Heinritz Member

    Joined:
    Mar 19, 2020
    Messages:
    6
    Likes Received:
    1
    Hi Schmartmaker. I just read your post.
    As it happens, I ran a test yesterday with the heatbed deactivated. There were no lines appearing. As a counter-test I printed again directly afterwards with the heatbed activated and observed the temperature curve (Octoprint). The waves in the temperature curve of the heatbed fit very exactly to the waves in the print. I think it has nothing to do with the expansion/contraction of the heatbed, because it only fluctuates around 0.5°C.
    I have recently tested a more powerful power supply (branded device with 20A) and also with 2 separate power supplies (15A for heated bed, 10A for controller/motor). In both tests the lines are present again. So it could be either the temperature fluctuations or just the controller (SKR 1.1 Pro). Maybe the controller has a problem with the switching of the heatbed.

    Tomorrow I will test an identical controller (I have another one at work).
     
    SchmartMaker likes this.

Share This Page