The "Flying Extruder" Mod

[aehM_Key]

Hello!

The Problem
A direct extruder (with the extruder motor at the nozzle) makes the end effector heavy. Due to the limited torque of the axis-motors and the limited stability of the linear bearings and the frame the printing speed is limited.

A bowden extruder (with the extruder motor at the frame, feeding the nozzle trough a tube) does not have these issues. The end effector is very light an can move with high speeds and high acceleration. However the bowden principle introduces many problems. This is mainly friction in the tube and a hysteresis due to the elasticity of the tube and play of the filament in it.
To fight stringing retraction is used. But because of this hysteresis you need more retraction. You'll also need more retraction if you are trying to print faster, because the pressure in the nozzle is higher and this pressure has to be eliminated before moving the nozzle to another position (otherwise you get strings). More retraction takes time an you get small blobs during the time the nozzle stays at one position.
Long story short: The fastest I could print with a decent quality was around 25mm/s with 5mm retraction.


The Solution
The Flying Extruder!

The extruder is suspended by rubber bands. These are running around several rolls to increase the length (each around 2m relaxed) to have more or less the same force over the whole vertical range. The extruder is connected with a very short tube (around 75mm) to the nozzle. This tube decouples the high frequencies between the nozzle and the extruder.
Long story short: I'm printing with no problems with 75m/s and 3mm retraction and still with the same acceleration of 800mm/s². But this was only a short test and I think I can bring it up with a bit optimization.
Another advantage is, that it's now easier to build an enclosure and that the printer is much quieter, because the extruder is decoupled.
(Cabling is not finished yet.)

Additional tips
Its important to cool the cold end even with ABS, because retraction is then more effective and reproducible.
I added some shrinking tube (with internal glue) to the tube and used cable ties around the tube fittings (to keep the plastic part away from the metal part) to minimize the play and make the retraction more effective.

Kind regards,
MK

PS: This is still in a prototyping state. I haven't tested everything end some optimizations must be made.

[Eaglezsoar]

Now that is a novel approach! Well done.
What is that "rubber band" actually called and a source for it?
Did you print those rollers the band uses?

[aehM_Key]

Thank you

I tried to print the small rollers in the first place, but couldn't get them to spin as smooth as they have to, so I found some rollers with steel shafts and printed only the flanges. I used ball bearings inside the large rollers:

I found the translation for the "rubber band": It's called "Braided Elastic". I also tried "Thera Band" (often used for slingshots) but it is too sticky and climbs the flanges.

[dbarrans]

Now that's an interesting idea. Do you use any Z lift when traveling? If so, you still have to move the extruder along with the platform, I think, and there'd be more stress on the Bowden fittings because of the momentum of the extruder.

I wonder if you could do this without the rollers by attaching much shorter elastic bands to the tops of the carriages?

- dan

[aehM_Key]

Yes, I use 0.5mm Z-Lift, but the extruder does not notice that, because of the elasticity of the tube.

The bowden fittings can handle it so far. I printed for some hours and the tube was also looking ok.

No, shorter elastic bands would not work, as I described in my first post. The relative stretching would be huge and the force at a low position to high, vice versa.

[JohnStack]

Looking at this, I'm wondering what you might come up with if you used the Ezstruder - which is significantly lighter.

Great work. You went quiet for a while....I was wondering!

[ApacheXMD]

Interesting! I wonder, though, since you're still essentially moving the extruder alongside the effector, arent you still moving much more additional mass? I realize the force of gravity is negated by the elastic bands but moving laterally youre still bound by F=MA where your mass (M) is still increased by the extruder.

Still this is awesome to see.

Can you post a video?

[aehM_Key]

Looking at this, I'm wondering what you might come up with if you used the Ezstruder - which is significantly lighter.

Could be good, but could also be bad, because the eigenfrequency would be lower. For now I'll go with what I have.

You went quiet for a while....I was wondering!

Figuring all this stuff out took a while (also it looks easy now ). I'm reading the forum regularly - had just nothing to post

I wonder, though, since you're still essentially moving the extruder alongside the effector, arent you still moving much more additional mass? I realize the force of gravity is negated by the elastic bands but moving laterally youre still bound by F=MA where your mass (M) is still increased by the extruder.

Yes, but the extruder mass is moved with a much lower acceleration (A) then the nozzle itself is moving.

Can you post a video?

Here you go: http://vimeo.com/68489641 .
Ask Kisslicer why it's producing such crazy moves
You can see the extruder not only moving, but also tilting, which requires also less force.

(BTW: I changed the tube fittings.)

[jesse]

Very cool idea and execution

[JohnStack]

That print is hauling a$$!

(Funny, my Rostock is laser-inscribed "My Wobble, Your Weeble" - perhaps that should have been your Rostock's name!)

You've given all of us a lot to think about!

[ApacheXMD]

Here you go: http://vimeo.com/68489641 .
Ask Kisslicer why it's producing such crazy moves
You can see the extruder not only moving, but also tilting, which requires also less force.

(BTW: I changed the tube fittings.)

Wow that's crazy! I can't argue with the tiny bowden tube. You must have retraction dialed in now.

[dbarrans]

No, I didn't mean shorter bands attached at the top of the frame. I was talking about attaching them to the carriages just above where the arms are attached. Not so much force difference there, because the carriages move along with the platform, and with the extruder.

Looking at the video, I see it does a great job of damping the smaller fast motions, but the extruder sure bobs around a lot. Have you ever seen it hit one of the arms during a long horizontal move?

- dan

[cambo3d]

I call it the bobble head extruder. better yet the "bobble-struder" or "bob-n-strude"

[aehM_Key]

No, I didn't mean shorter bands attached at the top of the frame. I was talking about attaching them to the carriages just above where the arms are attached. Not so much force difference there, because the carriages move along with the platform, and with the extruder.

Ah, sorry, I got you wrong.
I was thinking about this, but then the steppers would have to carry the whole mass.

Have you ever seen it hit one of the arms during a long horizontal move?

No, but this is one of the things I have to monitor. Especially when I try to print at the outside of the heat bed. I can make the tube a bit longer if this happens.

I call it the bobble head extruder.

Whatever makes you happy

[geneb]

Try print a stretchlet at high speed - I'm curious to see if it sets up an ugly resonance...

g.

[dbarrans]

It would take a bit of work to implement, but I'd think a separate set of arms with separate steppers to drive them would work out better. That would remove the possibility of resonances causing problems with the elastic interface between the extruder and the hot end. The idea is to move the extruder along with the hot end platform while filtering out the high-frequency movements of the hot end. Of course, it would also take a bit of programming to make it work.

I wonder if much of the benefit of this setup might be had by using the simpler approach of a fixed extruder mounting that would place it as low as possible in the center, just above the hot end in home position. This would allow a shorter Bowden tube, but not nearly as short as the flying extruder.

I guess I'm trying to solve theoretical problems that haven't been observed yet with the flying extruder. Maybe I should wait and see whether these are real problems first.

- dan

[dbarrans]

Of course, it would also be simpler to use more powerful steppers along with a stronger framework that can handle the acceleration with more weight on the platform. That's the brute force method, but might save a lot of trouble in the long run. Just don't get in the way while it's running.

- dan

[RichW]

I have to wonder if it wouldn't be better to use counter weights rather than elastic. If they were mounted inside a tube, they might even function as a damper to the "bobbling".

[JohnStack]

This might not belong here whatsoever and may already have been suggested - but going back to the problem that's trying to be solved - the imprecision of using the tube.

What if there were two cold extruders?

One with a very small stepper motor sitting directly on the head and another - the existing extruder sitting stationary.

The original would push the filament through and do most of the work. The second would work in tandem relieving the stress from the tube, evening out the feed and removing the variability that the tube introduces.

Questions: Could they work synchronously? Would the tube still introduce issues that the second would only exacerbate?

[RichW]

If you're using an extruder on the end effector, would you need a bowden tube at all? Maybe a feed device to ensure that there was always some slack in the filament leading to the direct drive extruder.

[JohnStack]

The problem with the stepper is the weight/speed relationship. The selling point of the delta is speed at hopefully +25mm/sec.

[joecnc2006]

Can you use three spring attached to the top which could act the same. This would free the space to remove parts from bed.

[626Pilot]

I like the idea of getting a small stepper motor with as many steps per degree as I can find, and then printing up some mounting hardware to put it on various hot-ends. That should do the trick. A lot of the mass in this mod is done away with if you start with an EZStruder rather than the "Steve's" extruder. I've seen videos of someone printing at decent speed with a QUBD head, which is a stepper motor and hot-end in one. He has some info and video here: http://www.kronosrobotics.com/3d/rostock/

[kbob]

I like the idea of getting a small stepper motor with as many steps per degree as I can find, and then printing up some mounting hardware to put it on various hot-ends. That should do the trick. A lot of the mass in this mod is done away with if you start with an EZStruder rather than the "Steve's" extruder. I've seen videos of someone printing at decent speed with a QUBD head, which is a stepper motor and hot-end in one. He has some info and video here: http://www.kronosrobotics.com/3d/rostock/

Why does the number of steps matter? It seems that what you want is a good torque to weight ratio.

[teoman]

Industrial robots have similar mechanisms to keep the wires hoses suspended lightly above the robot.

Something like this could be used as it "might" have costant pull for most of the travel range.
http://m.banggood.com/Stainless-Steel-T ... 24361.html" onclick="window.open(this.href);return false;