Detailed Documentation of our hardware and software mods for reliable printing

[Jamezracer 10]

In April 2015 we took delivery of two brand new Ultimaker 2 desktop 3D printers to complement our overworked large-format printer. Our first impressions were positive. We enjoyed the clean user interface, solid construction, and especially the low-inertia X & Y axis mechanism. It is now October 2015 and we have put over 2000 hours of printing on each of them along with countless rebuilds. Along the way we have come to appreciate the many strengths of this design which have no doubt been a result of the extensive testing and hacking of the original Ultimaker design by the opensource and DIY community. However, there are a couple of minor design faults which render the Ultimaker 2 unreliable and frustrating in a professional setting. Luckily these faults are cheap or free to remedy. In this paper we outline the various quibbles we have with the design of this printer and the steps we’ve taken to improve the performance and reliability. Along the way we provide empirical test data to demonstrate the improvements along with discussion on our own failures along the way.

The main paper is here

The complete repository with CAD files and software changes is here

Edited October 15, 2015 by Guest

[neotko 1,417]

I agree that the Hobbed Wheel Upgrade

it's nice. But the method used to adjust the esteps it's explained poorly (sorry). A more scientific approach would be to measure the inner radio of the bolt (since specially there are literally hundreds of models) and do a proper mathematical analysis. At least, use the reprap extruder feeder calculator...

A propper calibration guide:

http://reprap.org/wiki/Triffid_Hunter's_Calibration_Guide/en

An pseudoautomatic calculator for esteps and more:

http://prusaprinters.org/calculator/

Changing the esteps to the correct number it's an easy process that on um2 can be done by gcode.

Edited October 15, 2015 by Guest

[neotko 1,417]

Also you describe the 'atomic' method without properly pointing to the source of this that was MostlyAtomicBob...

[avogra 68]

Your modifications look really simple and effective! I didn't hear about the spacer for the hotend fan before. Will certainly give it a try!

You could print Roberts Feeder mirrored in order to mount it reversed. What else I have seen is, that people don't mount their spools to the spoolholder but put it on a stand to the left. This should have the same effect as reverse mounting and you also gain easier access for when you want to change filaments.

My printer passed the extrusion test up to 10mm^3 out of the box. I have it for 4 weeks now, so maybe Ultimaker has improved some of your points in more recent units? Or I just had luck with my unit. I already heard, that Ultimaker is improving their machines quietly over time. Are those changes publicly available somewhere?

[eldrick 65]

The hobbed wheel listed is described on Amazon as designed for .175 filament.

[lilmaniac2 0]

Nice read, enjoyed the thorough technical explanation for each modification.

[frederiekpascal 193]

Very interesting read, thanks for sharing.

You should give this spool holder a try. It works wonderful on my UM2, I love it.

Also, you did many upgrades to your machine to improve the performance and reliability like you said, for that I don't understand why you didn't give the Olsson block a try. I bought some E3D nozzles to fit in the block and now a dirty clogged nozzle goes right in the dumpster and a new one is on in 30 seconds. Also the feature to be able to use hardened steel nozzles and other sizes-nozzles to experiment even more with your machine and different filaments.

I was also thinking about changing my fan shroud and the filament feeder but read as much positive as negative experiences from users with those 2 upgrades that I'm skipping these upgrades for now.

[BrayChristopher 36]

You could print Roberts Feeder mirrored in order to mount it reversed. What else I have seen is, that people don't mount their spools to the spoolholder but put it on a stand to the left. This should have the same effect as reverse mounting and you also gain easier access for when you want to change filaments.

 

I did almost the same thing. My UM2 sits on a cart. I made a spool holder (with bearings) that attaches to the side of the cart, so the spool holder is below the UM2.

Feeder problems gone. I think it is a combination of a lot of things. Straight line from the spool to the feeder. Free spinning bearings. Fewer newby mistakes on my part. Good reliable filament.

I had the Robert Feeder all printed out and ready to install, but I haven't had the need to since I move the spool holder.

Nice report.

[Jamezracer 10]

Also you describe the 'atomic' method without properly pointing to the source of this that was MostlyAtomicBob...

 

I've seen this described most places as the cold pull method. I will add credit to Atomic Bob. Most of these mods are actually community mods so I want to give credit wherever possible.

I did gloss over the esteps calculation since I had to adjust it by testing after setting it based on diameter. The slip ratio actually came out to about 4%. I'll add this in.

eldrick-  You are right in that it is advertised as for 1.75mm but the diameter of the hobbed section is 3.2mm and fits around 2.85mm filament spectacularly.  

avogra- very lucky indeed! I found someone else at our university with an ultimaker 2 and he has all the same issues that I have. That was purchased around the same time as us though so they may be making improvements without telling us.

Thank you for the feedback guys, I'm going to keep adding this.

[neotko 1,417]

eldrick-  You are right in that it is advertised as for 1.75mm but the diameter of the hobbed section is 3.2mm and fits around 2.85mm filament spectacularly.  

 

Since it fits for 2.85mm did you observe any 'flattening' of the filament from the bearing force? I'm wondering about changing my bearing to one with a inner curve (concave?) so the force it's focused on the grip area and the filament suffers a bit less and guides better the filament for flexibles (and for weak filaments like wood, etc). Interesting the 4% slipping ratio. Btw with all my criticism I forgot to say that it's indeed a great recollection of tunnings. Good job

[Jamezracer 10]

 

eldrick-  You are right in that it is advertised as for 1.75mm but the diameter of the hobbed section is 3.2mm and fits around 2.85mm filament spectacularly.  

 

Since it fits for 2.85mm did you observe any 'flattening' of the filament from the bearing force? I'm wondering about changing my bearing to one with a inner curve (concave?) so the force it's focused on the grip area and the filament suffers a bit less and guides better the filament for flexibles (and for weak filaments like wood, etc). Interesting the 4% slipping ratio. Btw with all my criticism I forgot to say that it's indeed a great recollection of tunnings. Good job

 

I too worried about flatenning when running high tension on Robert's feeder. I actually have spare bearings with built in curves from our 3DP Unlimited printer so I threw them on for fun. It didn't make any noticeable difference....

criticism is more useful than "good job" so thank you!

[izzy 33]

I agree with the other replies, a great little paper.

Love the Hobbed gear and will be looking more into this, colud you also give more details a picture etc on your heatsink fan duct.

I use Roberts feeder which I have offset from the back plate by 3mm and added a knurled head to the adjuster screw so that I can easily adjust the tension (all the bits on my page on YouMagine

[Jamezracer 10]

I agree with the other replies, a great little paper.

Love the Hobbed gear and will be looking more into this, colud you also give more details a picture etc on your heatsink fan duct.

I use Roberts feeder which I have offset from the back plate by 3mm and added a knurled head to the adjuster screw so that I can easily adjust the tension (all the bits on my page on YouMagine

 

Hi Izzy, I just added a CAD render of the duct to the github and inline with the text. Thank you for the clip design. I also use your nozzle size indicator on our big printer!

neotko- I put in the due credit to MostlyAtomicBob and my original calculation for estimated steps/mm. based purely on filament diameter it should be 332.5 but I found 327 worked best with my slicer multiplier set to 1.0

[neotko 1,417]

I would really recommend you the @ultiarjan Rail System, it allows you to rotate the spool avoiding the tension of the filament from the unnatural rotation when it goes up.

https://www.youmagine.com/designs/ultimaker2-rail-system-u2rs

I did something very sloopy and rushed, but works, to change the angle also for my umo+

https://www.youmagine.com/designs/umo-some-friction-spool-holder

Ultiarjan solution allows to add tools and stuff, it's really well planned.

[izzy 33]

Hi Jamezs,

Could you put the duct on YouMagine, I am unable to download the STL file from GITHub, I also tried the SloidWorks files but got nowhere. i doubt if I would be able to use them as I only have SW2102, Cheers Izzy

[Jamezracer 10]

Hi Jamezs,

Could you put the duct on YouMagine, I am unable to download the STL file from GITHub, I also tried the SloidWorks files but got nowhere. i doubt if I would be able to use them as I only have SW2102, Cheers Izzy

 

You can right click on the github "raw" button to download. But I should put these on youmagine anyway so here you go

https://www.youmagine.com/designs/ultimaker-2-heatsink-fan-duct

[izzy 33]

Hi James,

Just a quick question re the Heat Sink Fan Duct, How have you routed your wires for the fans and heaters?

I printed off your Part but then made a modified part as I have my Cooling fans on a non standard Fan Duct that swivels on the rear Print head screws to allow very easy access to the nozzle (Design Work in Progress before i make available), but when fitted there is no room for the wires to exit as standard fit, did you modify the bottom plastic section of the UM2 Print Head to allow the wires to exit?

[eldrick 65]

I've installed the hobbed wheel and the PTFE Bowden tubing recommended.

The longer PTFE Bowden tube reduces friction considerably, although it is not as transparent as the OEM. It's a perfect fit.

I haven't experimented with tension on my |Roberts| feeder with the hobbed wheel yet, and it seems to need to be tightened at least a couple of full turns of the tension screw compared to the stock wheel. My quick use so far did show less mashing and distortion of the filament, so that's in the right direction too.

Good tweaks!

[Jamezracer 10]

Hi James,

Just a quick question re the Heat Sink Fan Duct, How have you routed your wires for the fans and heaters?

I printed off your Part but then made a modified part as I have my Cooling fans on a non standard Fan Duct that swivels on the rear Print head screws  to allow very easy access to the nozzle (Design Work in Progress before i make available), but when fitted there is no room for the wires to exit as standard fit, did you modify the bottom plastic section of the UM2 Print Head to allow the wires to exit?

 

You need to trim a little black plastic off right above the fan.

Eldrick- i found you can crank up the tension up very high without issues.

[izzy 33]

Does anybody have an accurate model of the bottom part of the print head?

I don't fancy removing a small notch or two from my origional part so would rather print a modified replacement.

The drawing on GitHub ia but poor, if nobody has an accurate drawing I'll strip mine apart to measure it.

Also for the heat sink fan, the Sunon fan spec shows a mounting hole pitch of 20mm +- 0.3.

On GitHub the drawing for the bottom plate is indicating that the pitch is 20.4mm.

Unless the Drawing is from Ultimaker I'd stick with the 20mm as this is a std pitch for 25mmSQ fans

[ultiarjan 1,223]

Not sure I understand the issue.. did you check the .step file of the unit? I never noticed it's not ok?

https://github.com/Ultimaker/Ultimaker2/blob/master/HeadAssembly.step

[izzy 33]

hi Jan, I can't seem to download the step file, even right clicking doesn't give me an option to download the file, even though I'm signed in!

[ultiarjan 1,223]

There should be a "zip" button on the right side of the page to download all. That should work.

[izzy 33]

I've managed to download the step file and checked the dimension for mounting the hest sink fan, the dimension on the bottom plate is measured at 20.4mm although the produced part on the actual machine appears to be 20.0mm which does corrolate to the dimension for the mounting holes for the fans of this size.

[eldrick 65]

The heatsink fan fit nicely on my UM2, and probably flows a bit better than a similar design posted earlier on youmagine.

 

I agree that the Hobbed Wheel Upgrade

it's nice. But the method used to adjust the esteps it's explained poorly (sorry). A more scientific approach would be to measure the inner radio of the bolt (since specially there are literally hundreds of models) and do a proper mathematical analysis. At least, use the reprap extruder feeder calculator...

A propper calibration guide:

http://reprap.org/wiki/Triffid_Hunter's_Calibration_Guide/en

An pseudoautomatic calculator for esteps and more:

http://prusaprinters.org/calculator/

Changing the esteps to the correct number it's an easy process that on um2 can be done by gcode.

 

I'd suggest an alternative method, which takes the compression of the filament by the hobbed wheel and thus the effective radius (varies with spring tension) into account, is as follows (I don't recall where I got it, so can't attribute it):

- Mark the filament before it goes into the feeder - 60-70mm from the feeder.

- Measure the length of the filament from your mark to the feeder entry.

- Feed 50mm of filament from the UM or pronterface.

- Remeasure the remaining length.

- Subtract to see how far the filament actually moved.

- Divide actual/50, to get a ratio between the current esteps and the actual filament moved.

- Multiply the ratio times current esteps and change the esteps to the result.

Repeat until you are satisfied as to accuracy.

Edited October 26, 2015 by Guest