Improving the UM2 feeder mechanism for flexible filament

[gadgetfreak 7]

So, we've been trying different flexible filaments in the UM2 and it actually works pretty OK if the filament is not *too* flexible.

I started by reading Sanjays excellent thread where he tested three different filaments. So I ordered FlexPolyEster of grades Shore 45D (harder rubber) and 40D (softer rubber).

The 45D actually works very well to extrude in the UM2 unmodified. We've been printing at 0.2mm layers at 30mm/s with good results. We had to print on blue tape that was cleaned with isopropanol though, larger objects otherwise shook loose from the heated glass bed (we tried 60C, 90C, with and without gluestick, we tried hairspray, brim, taped brim etc....so far no luck directly on glass bed...).

However, the 45D is not as soft as we would like for some applications. The 40D has a softer rubbery touch and feel, close to NinjaFlex for example.

Although we managed to print a very small Octopus with 40D, when we tried the BMX handle at 30mm/s it failed and the feeder mechanism could not feed the material for very long at all. Like Sanjay writes in the thread referenced above, the rubbery filament wants to go here and there and goes to the side in the UM2 feeder instead of pushing upwards towards the bowden tube.

In the thread Sanjay therefore suggest a modification to the drive:

 

This would restrict the filaments' movement so that it could not bend sideways.

 

I'm now thinking of how this can be implemented into the UM2. Check these pictures to see the outside and innards of the UM2 feeder mechanism and where I've marked in red where "filament guides" would be needed:

 

 

(note: the filament in the picture is not flexible just a transparent one)

 

The problem as I see it is that this is a very precisely injection molded part and it seems to me very hard to actually be able to fasten any type of guides here that would work or last....perhaps epoxy?!

 

Perhaps if someone at UM (hrm...Daid wink wink ) could supply the CAD files then they could be modified with guides and then I actually don't think that FDM/FFF would suffice here but maybe with some nice SLA or DLP machines the precision would be good enough to make a transparent case (too also see and analyze the results) with modified guides?

 

Now it would be nice what others think of this....

 

 

 

[gr5 2,025]

These are the best photos yet of the inside of the extruder.

Your ideas are great. One of those red guide lines however will get hit by the idler. But the other 3 look good to me.

 

[SandervG 1,518]

Alright, looks good. I am going to read the report. What is your thought on controling the flexibility in the bowden tube itself, and perhaps hot end?

[gadgetfreak 7]

Alright, looks good. I am going to read the report. What is your thought on controling the flexibility in the bowden tube itself, and perhaps hot end?

 

Great! If you or David could supply the CAD drawing for the filament feeder housing it's easier for everyone to contribute.

I'm not really sure what can be done for controlling the flexibility in the bowden tube or the hot end. However, a lot can be done when choosing slicing parameters; pehaps a "flexible filament profile" would be good to have; example is to use a constant velocity, even for first layer (see below).

Also, there is a big difference on various flexible filaments...some we might get to work well with hw/sw modifications and some might just be too rubbery...only experimentation can tell!

Sanjay makes very good remarks on printing with flexible filament (although he has as far as I can tell not tested with a bowden-style printer like the Ultimaker but with direct driven feed...).

Summary of the most important things in my view:

"- print at a completely constant velocity, for perimeters, infill, everything. Make it quite slow - around 30mm/s all round - don't use a slower first layer. The reason for this is that we want to print this stuff in as much of a steady state system as possible. Changes in velocity and extrude rate affect the output flow rate of the extrudate in a drastic way. Travel moves should be made as quickly as your machine can handle them. Retraction should be aggressive and quite long.

- Printing too quickly, or having the slightest glitch in you extruder-hotend filament path will cause buckling.

- these filaments act like a spring when compressed and driven down to the hotend. If your hobbed bolt stops turning there is still stored pressure in the system and filament will continue to extrude/ooze."

 

[Daid 304]

Great! If you or David could supply the CAD drawing for the filament feeder housing it's easier for everyone to contribute.

 

I'll quickly ask and let you know on Tuesday. I have the "latest" files, which are injection molded ready. However, I also have an older version which is not injection molded ready, but can be 3D printed much easier. (It's what I have on my home machine) Might be of more use to share those. But before I do I have to ask if I can share them.

(I have a day off on Monday)

 

[gadgetfreak 7]

Thanks David!

[nick-foley 5]

Seeing the inside of the UM2 extruder for the first time - it looks very very printable. I bet you could even print the whole assembly in one piece, pausing the print to drop in the bearing... maybe even the spring. Probably best in something rigid though.

 

[chopmeister 0]

Seeing the inside of the UM2 extruder for the first time - it looks very very printable.

 

Just what I was thinking. What's more, who says you have to stick with the stock design?

 

[ian 32]

or the stock design is offically vastly improved and tweaked.. then you can stay with the stock design.. the new stock design :wink:

Ian

ps.i dont think people should consider under extrusion and the other faults with the existing extruder design a permanent problem !

 

[Daid 304]

Ok, a bit later then expected. But I got permission to share the printable files:

https://www.youmagine.com/designs/ultimaker2-feeder-printable

This is a slightly dated version that used the UM-Original feeder spring and a 5mm spacer. But other then that it functions the same. I printed a lot of these for testing goals, and quite a few people working at Ultimaker have one of these printed versions on their home UM-Origonal.

 

[takei-naodar 0]

Thank you Daid, you saved us all quite some time we would have otherwise needed to measure things out ^^

Could you also post the CAD files pls?

 

[Daid 304]

Could you also post the CAD files pls?

 

I do not have those. Sorry. If I had them, I would have shared the step files.

 

[drayson 74]

Great - interesting to know that some of the UM employees have this stuff installed... Will there be an official upgrade available??

Nevertheless, guess there is a way to install it - isnt??

 

[takei-naodar 0]

I do not have those. Sorry. If I had them, I would have shared the step files.

 

Back to measuring out the stuff it is then....

 

[ballanux 0]

Ok, a bit later then expected. But I got permission to share the printable files:

https://www.youmagine.com/designs/ultimaker2-feeder-printable

This is a slightly dated version that used the UM-Original feeder spring and a 5mm spacer. But other then that it functions the same. I printed a lot of these for testing goals, and quite a few people working at Ultimaker have one of these printed versions on their home UM-Origonal.

 

mmm, so this could be used in the UM Original without much problem? I think I will try it :-)

Is the spring in this feeder the same than in the UM-Original?

 

[gr5 2,025]

Won't work - you need a special knurled bolt thingy to go on the motor. Much sharper and pointier than the one on UM1. Plus a ball bearing.

 

[drayson 74]

Hmmm... An MK8 curled pulley should fit - isnt?

Would be great if somebody could organize the bearing spec...

@Daid, would this be possible?

Next, what else is to change? Marlin settings? Cura? I guess, any info would assist

[Daid 304]

mmm, so this could be used in the UM Original without much problem? I think I will try it :smile:

Is the spring in this feeder the same than in the UM-Original?

 

You need to drill extra holes to fit the motor, and the motor of the UM2 is stronger and more steps then the UM-Ori feeder. And you need the knurled sleeve.

The spring can be re-used, but you need to add a small spacer (5mm in my case). The actual spring in the feeder for the UM2 is a different one. The bearing is the same type as in the UM-Ori feeder that is holding the knurled bolt (so you have 2 of those)

Software wise you need to adjust the steps per mm for E to 141 or 282 depending on if you have a 200 step or 400 step per rotation motor.

http://daid.eu/~daid/IMG_20130405_155954.small.jpg parts needed

http://daid.eu/~daid/IMG_20130406_112324.small.jpg result on my home machine.

But your biggest problem would be the knurled wheel.

The home machines of UM employees are used quite a bit for testing UM2 parts. People have all kinds of machines in all kinds of states now. Many of us have the UM2 electronics fitted. Some have the feeder, some have the heated bed. But it all required custom tinkering or printed parts to get stuff fitted.

 

[ballanux 0]

You need to drill extra holes to fit the motor, and the motor of the UM2 is stronger and more steps then the UM-Ori feeder. And you need the knurled sleeve.

The spring can be re-used, but you need to add a small spacer (5mm in my case). The actual spring in the feeder for the UM2 is a different one. The bearing is the same type as in the UM-Ori feeder that is holding the knurled bolt (so you have 2 of those)

Software wise you need to adjust the steps per mm for E to 141 or 282 depending on if you have a 200 step or 400 step per rotation motor.

http://daid.eu/~daid/IMG_20130405_155954.small.jpg parts needed

http://daid.eu/~daid/IMG_20130406_112324.small.jpg result on my home machine.

But your biggest problem would be the knurled wheel.

The home machines of UM employees are used quite a bit for testing UM2 parts. People have all kinds of machines in all kinds of states now. Many of us have the UM2 electronics fitted. Some have the feeder, some have the heated bed. But it all required custom tinkering or printed parts to get stuff fitted.

 

Great, thanks for the info! I might try to find a similar knurled wheel and put it together...

What are the specs of the motor? would be one of 400 steps and about 48Nm of torque enough?

 

[gadgetfreak 7]

Great, thanks for the info! I might try to find a similar knurled wheel and put it together...

What are the specs of the motor? would be one of 400 steps and about 48Nm of torque enough?

 

Here's some related info from two weeks ago from Daid (numbers) and Joergen (sparkfun link):

> It has a holding torque of 44 N*cm, 400 steps per revolution. NEMA17 of

> 47mm. However, unless you are doing direct-drive it's way overpowered.

> I have no idea where to get them in small numbers.

 

google to the rescue: https://www.sparkfun.com/products/10846

 

[drayson 74]

Hi!

might this kurled wheel work?

http://www.ebay.at/itm/281102410932?ssPageName=STRK:MEWAX:IT&_trksid=p3984.m1423.l2649

and I found a similar stepper

http://www.ebay.at/itm/1-pcs-nema17-stepper-motor-0-9-2-4A-48mm-4leads-4200g-cm-for-3D-printer-CNC-/141055057398?pt=Motoren_Getriebe&hash=item20d7895df6

 

[gadgetfreak 7]

I'm planning to use this guide for inspiration:

http://www.spaceclaim.com/en/Support/Tutorials/Modules/SpaceClaim_STL.aspx?t=110#

to make a proper CAD part using SpaceClaim but it should be possible also to use the free version DesignSpark Mechanical:

http://designspark.com/eng/page/mechanical

The process looks really cool and pretty simple...I hope I'll have some time tomorrow but please do try this folks!

 

[ballanux 0]

Here's some related info from two weeks ago from Daid (numbers) and Joergen (sparkfun link):

> It has a holding torque of 44 N*cm, 400 steps per revolution. NEMA17 of

> 47mm. However, unless you are doing direct-drive it's way overpowered.

> I have no idea where to get them in small numbers.

 

google to the rescue: https://www.sparkfun.com/products/10846

 

Yes! I was thinking in buying that one :-)

 

[gr5 2,025]

I don't think you want to put one of these on the UM1 because there is no way to turn it by hand. On the UM2 there are special new procedures in Marlin for changing filament. There's no "release" button that you can push to then push and pull filament in and out of your machine.

I'd like to see a design modification with a release button that holds the ball bearing away from the knurled sleeve so you can shove filament in or out with the other hand.

Also the UM2 has a new menu item "move filament" that makes it easy to prime the nozzle a bit. You can move filament in or out of the head. It's much too slow though to remove it from the entire bowden - for that you have the "change filament" menu option.

 

[ballanux 0]

I don't think you want to put one of these on the UM1 because there is no way to turn it by hand. On the UM2 there are special new procedures in Marlin for changing filament. There's no "release" button that you can push to then push and pull filament in and out of your machine.

I'd like to see a design modification with a release button that holds the ball bearing away from the knurled sleeve so you can shove filament in or out with the other hand.

Also the UM2 has a new menu item "move filament" that makes it easy to prime the nozzle a bit. You can move filament in or out of the head. It's much too slow though to remove it from the entire bowden - for that you have the "change filament" menu option.

 

I didn't thought about not being able to push the filament by hand, but I don't see much problem about the "move filament" menu, the UM1 also has that option, maybe not so visible, but you can go to prepare, move axis, extuder.

Anyway I may try to put it as a second extruder, without dismantling the original, so I will have to find also a new spring and bearing