BMG + Mosquito + Tool Changer

Discussion in 'Tool heads & ToolChanger' started by Ntesla66, Oct 24, 2019.

  1. garethky

    garethky Well-Known Member

    Joined:
    Oct 8, 2019
    Messages:
    49
    Likes Received:
    28
    I tried to reproduce my success with the benchy and had failures where the bridging for the roof happens. Filament is curling up at the corners of the "house" on the benchy and eventually there is a collision with the hot end that causes the print to unstick from the bed. A few things are going on:
    • Cooling is not adequate for good bridging
    • Cooling air seems to be hitting the hot end and possibly cooling it
    A re-design of the nozzle is required to address both. I'm going to change it so it fits around the opening of the fan and can be printed with the opening down on the bed. I also think that the 4010 fan doesn't blow evenly across its opening so the new design will narrow and provide asymmetric flow channels. But development is going to be slower now that I'm back from vacation.
     
  2. Brian Peterson

    Joined:
    Jun 26, 2018
    Messages:
    7
    Likes Received:
    1

    Please keep up the good work! Looking forward to trying this out.
     
    garethky likes this.
  3. garethky

    garethky Well-Known Member

    Joined:
    Oct 8, 2019
    Messages:
    49
    Likes Received:
    28
    I have to admit defeat on the under-slung cooling fan. Its just sucking hot air up from the bed and cooking the part. Testing with higher heatbed temps the results are much worse.

    The fan needs to be moved up onto the side of the extruder somewhere. Behind the stepper or under the dock.

    Behind the stepper has downsides. If you don't use the same length stepper as me its not going to work. Also the duct would have to be really long and the width of the extruder would go up a bit and its already kind of wide. Under the dock would push the extruder out further and require 2 x 90 degree bends to get airflow pointed at the nozzle.

    I'm starting to think that @Jai Stanley has it right over in the "Tall BMG" thread: https://forum.e3d-online.com/threads/prototype-bmg-based-tall-direct-drive-extruder-tool.3479/
    Thats a lot more compact in X/Y and there is room in the Y direction for a larger 50mm cooling fan. Plus there is space to fully remove the idler gear and check for filament grinding and space on the other side to add a filament viewing port. My design has no space for either.

    Currently thinking about starting over with that "Tall" design for the mosquito...o_O
     
  4. garethky

    garethky Well-Known Member

    Joined:
    Oct 8, 2019
    Messages:
    49
    Likes Received:
    28
    I'm continuing to test my "tall" extruder variant. So far its very promising. I have discovered lots of issues that really have nothing to do with the extruder. Primarily the tool changer GCode sets the part cooling fan to 50% speed and my slicer (Prusa Slicer) didn't set it to 0 for the first 3 layers. This was leading to absolute fits getting a first layer to stick. The next issue is part cooling, the new design has too much cooling! I've had to dial back on the PCF speed by at least 25% vs the stock Prusament settings to stop the parts from warping off the bed.

    The last issue in trying to find the source of some over extrusion. I found this really interesting line of GCode that Prusa Slicer puts into every print for the MK3S (which uses the same Bondtech gear system I am using):

    M221 S{if layer_height<0.075}100{else}95{endif}

    This sets the Extrusion Factor to 95% for pretty much all prints. You would have to do it with a Dx parameter in Duet like this to have it apply to all of the tool heads:

    M221 S{if layer_height<0.075}100{else}95{endif} D0
    M221 S{if layer_height<0.075}100{else}95{endif} D1
    M221 S{if layer_height<0.075}100{else}95{endif} D2
    M221 S{if layer_height<0.075}100{else}95{endif} D3

    So that's one source. The other thing the slicer does is set linear advance in the filament settings custom GCode with an M900. As far as I can tell Duet doesn't support M900 but it does support Pressure Advance with M572. This value needs to be tuned in for each kind of filament.

    So lots of messing around still to go but I'm getting there.
     
    #44 garethky, Feb 12, 2020
    Last edited: Feb 19, 2020
  5. dc42

    dc42 Well-Known Member

    Joined:
    Aug 16, 2016
    Messages:
    403
    Likes Received:
    190
    That's right, for some reason the Marlin developers ignored the fact that we already had M572 for several years when they chose M900 to set pressure advance. Maybe it's because M900 originally specified M900 in a way that depended on the motor steps/mm. Later they changed to seconds like RRF.
     
  6. garethky

    garethky Well-Known Member

    Joined:
    Oct 8, 2019
    Messages:
    49
    Likes Received:
    28
    Honestly, the one test I ran, setting pressure advance to 0 seemed to produce the best results. I think my retractions need turning.

    And Z-Jerk is apparently an issue as well, I'm going to start a separate thread for that.
     
  7. garethky

    garethky Well-Known Member

    Joined:
    Oct 8, 2019
    Messages:
    49
    Likes Received:
    28
    I am learning so much about turning! Prusa really saves you from having to dial in each material. But this is no different that a cut recipe on a CNC or dialing in an Espresso machine for new beans. Lots of trial and error.

    Washing the PEI sheet with soap and water really seems to have solved most of the adhesion issues I was having. This allowed me to reduce extruder temps by 5 degrees and run the same 30%/50% fan settings and the MK3S PETG preset without any first layer adhesion issues or part curling/warping. This was the root cause of most of the print artifacts.

    I also ran the PID tuning of the heat bed so it no longer temp cycles during the print which could have caused some issues previously with inconsistent first layer heights. I'm using a BuildTak magnetic base with a textured PEI sheet. The mag base really doesn't want to conduct heat into the build plate. To get to 85C measured at the build surface I'm heating the bed to 115C.

    Retractions were also a bit of a surprise. I'm printing this with 0.4mm retractions. That's the value recommended by Slice Engineering and it seems to work really well. The Prusa setting for PETG is 1.8mm!

    The last issue I'm seeing is some buildup of plastic on the nozzle which causes occasional artifacts. Not sure if this is nozzle specific or an over-extrusion artifact.

    I'm working on having extruder #2 print out the parts for extruder #3. If it can Rep-Rap then its good enough to ship. Here is some first layer porn: IMG_5037.jpg
    IMG_5040.jpg
     
    Ntesla66, Greg Holloway and Nibbels like this.
  8. Ben Kay

    Ben Kay Well-Known Member

    Joined:
    Sep 9, 2019
    Messages:
    47
    Likes Received:
    14
    Nice, where'd you get a textured sheet that large?
     
    Nibbels likes this.
  9. garethky

    garethky Well-Known Member

    Joined:
    Oct 8, 2019
    Messages:
    49
    Likes Received:
    28
    I don't know of a source for the exact size of the print bed, but I know of 2 that are larger and can be cut down:
    1. LayerLocker by Matter Hackers: https://www.matterhackers.com/store...oated-pei-build-plate-steel-sheet/sk/MDP3RPRD (this is the one I have but I'm not thrilled)
    2. Ultistik https://www.filamentone.com/collect...powder-coated-ultem-pei-build-plate-300-x-300 (getting one of these to see if its any more study than the LayerLock)
     
  10. Wiley

    Wiley New Member

    Joined:
    Dec 28, 2016
    Messages:
    1
    Likes Received:
    0
    This looks really exciting! Did your "tall" redesign get to a point where you're comfortable sharing some STLs? I'd like to play with this as well.
     
  11. Mikee

    Mikee Member

    Joined:
    Mar 16, 2020
    Messages:
    6
    Likes Received:
    4
    garethky likes this.
  12. garethky

    garethky Well-Known Member

    Joined:
    Oct 8, 2019
    Messages:
    49
    Likes Received:
    28
    I found out that I had some collision issues with the cooling fan. I got side tracked on the heated bed, wanting a more consistent first layer without the glass plate. I need to print some alterations and check them before sharing STLs. World current events have been intervening on 3D printing projects.

    The link to the CAD project is here, and it allows for STL export if you are brave: https://cad.onshape.com/documents/2...bedc38a640e7a5f738/e/08338b0153cc1fb757b761de
     
  13. garethky

    garethky Well-Known Member

    Joined:
    Oct 8, 2019
    Messages:
    49
    Likes Received:
    28
    Ok, at long last I have something you can print and try out that I'm pretty confident will work well

    extruder-pair-back.jpg extruder-pair-left.jpg printed-parts.jpg required-hardware.jpg

    Two big reasons Im more confident:
    1. The Idler has been completely re-designed. This was the weak point. The idler shaft is now specified to be 24mm long (same as the Prusa MK3 extruder) and it has significantly more material supporting it. This was tested with a couple Kg of filament and it shows no signs of wear like previous iterations did.
      insert-idler-shaft.jpg
    2. I now have a much better understanding of what acceleration and jerk values to use to get good prints. The weak infill, over extrusion, blobs on the nozzle, gaps at seams, pressure advance slowing the printer down... all of that is 100% gone and I'm getting some beautiful prints in PETG.

    GitHub has the latest STLs and instructions: https://github.com/garethky/Slice-Mosquito-Direct-Tool-Head

    Please let me know how it goes if you build one!
     
    #53 garethky, May 3, 2020
    Last edited: May 7, 2020
  14. Spoon Unit

    Spoon Unit Well-Known Member

    Joined:
    Sep 6, 2015
    Messages:
    1,546
    Likes Received:
    478
    Awesome work. The level of documentation is superb.
     
    garethky likes this.
  15. garethky

    garethky Well-Known Member

    Joined:
    Oct 8, 2019
    Messages:
    49
    Likes Received:
    28
  16. garethky

    garethky Well-Known Member

    Joined:
    Oct 8, 2019
    Messages:
    49
    Likes Received:
    28
    After upgrading for Duet 3 I had hoped that I could quiet down that Mosquito part cooling fan. It did not work, it does not like PWM. So this led me down a rabbit hole of fan testing. I testing 2 x 25mm fans, 2 x 30mm fans and the 2 x 40mm Noctua fans. I tested in the dock, in free air and with the fans blowing in 2 different directions. I tested 5 different duct designs and I took a belt grinder to the extruder to get them to fit. I took the measurements from 3 different places and settled on inside the heat break being the most accurate. I focused on temps at 220C in the dock, we really don't want the heat break to get above PLA's glass transition temp and cause a clog.

    IMG_5239.jpg

    Upshot is, that 40mm x 10mm Noctua actually does quite well, just 2C higher than the stock fan in the dock and 3C lower than stock fan in free air. Its also a known quantity and very quiet. Bad news is every plastic part of the extruder needs tweaks to fit it on there and get the most out of the fan. So that was not what I though I was going to do for V3 but it seems like the next step on the road to a quiet.

    How do you all feel about switching to this fan?
     
  17. dc42

    dc42 Well-Known Member

    Joined:
    Aug 16, 2016
    Messages:
    403
    Likes Received:
    190
    I use the 3-wire version of that fan on my Hemera tool. No issues with it so far.

    BTW it's possible to use high frequency PWM and an LC network to convert PWM to smoothed DC, to drive a fan that doesn't like PWM.
     
    garethky likes this.
  18. garethky

    garethky Well-Known Member

    Joined:
    Oct 8, 2019
    Messages:
    49
    Likes Received:
    28
    I thought about getting a buck converter and hooking into the jumper in the Duet 3 but these fans are just too noisy to stay. The whole point of the 5015 part cooling fan was low noise, so its silly not to make the same decision about the hot end fan.

    I had heard that some people have had issues with the Noctua in hotter environments, but this was with PLA and a V6. The Mosquito and the Hemera are supposed to have better thermal performance and need less cooling so this shouldn't be an issue. Lots of people are doing enclosures with the MK3 to print ABS/ASA and that works out OK. I think that's where my build is headed.
     
  19. ByteSized

    ByteSized Member

    Joined:
    Apr 5, 2020
    Messages:
    17
    Likes Received:
    5
    Why oh why do they have to be brown? o_O

    With Hemera still unavailable this is on my shortlist for a direct drive extruder, replacing that Slice fan would be my first priority, so yes I think this is the way to go.
     
    garethky likes this.
  20. garethky

    garethky Well-Known Member

    Joined:
    Oct 8, 2019
    Messages:
    49
    Likes Received:
    28
    You can get black ones from Prusa but only in 5V. https://shop.prusa3d.com/en/mk3mk3s/199-hotend-fan-noctua.html
     

Share This Page