Ultimaker2 Feeder System - Improvements and Ideas

[chrisp 10]

I finally got around to printing lRobertl's feeder/extruder. I am wondering how to set the tension properly? Should I go by how far the teeth are biting into the filament, or how far the 30mm screw is threaded into the nut, or how compressed the spring is? I had a few issues with my first prints and it seems like there is too much tension between the bearing and the knurled fitting.

 

Any advice from others who have been using it awhile would be much appreciated.

 

I also moved the knurled fitting back slightly to align the filament directly in the middle.

 

[pm_dude 27]

I have the sprint compressed to about 13mm long when the arm is loaded and a filament is present.

Try printing a small cylinder (like a coin) about 25mm wide and 5mm tall. While printing you should be able to tweak the tension untill you are happy with the result.

 

[donmilne 2]

+1 to what Pm_dude said, plus I'd say to use 100% fill. You want it doing the same thing on every layer so that the only variable is your tweaks. I would start off fairly light then tighten until you can just see indentations being pressed into the filament as it feeds. Then tighten a quarter turn at a time until you like the printed result.

I've found that I can tell by the knurled indentations on the filament if it's being squeezed the right amount.

 

[pm_dude 27]

Haaa yes! Thanks for completing my thoughts DonMilne

 

[gr5 2,167]

Oh, I know about the flex drive. my thought is something a lot different

@muttstang: Are you sure? You seem to be describing flexidrive. It uses a speedometer cable to drive a head mounted extruder. What is different about your idea?

 

[gr5 2,167]

A question for all you other people who read *everything* on the forum...

There was discussion in the last 2 months about a material that is VERY expensive but not too bad in small quantites that can be machined to the shape of the isolator. It is not a metal material (molten PLA sticks to steel for example). It is not PEEK. It is some new material I never heard of. Where is that thread? I can't find it anymore.

 

[IRobertI 520]

http://umforum.ultimaker.com/index.php?/topic/7024-teflon-spacer-replacement/page-6

 

[gr5 2,167]

Found the answer. Macor.

 

[muttstang 8]

@muttstang: Are you sure? You seem to be describing flexidrive. It uses a speedometer cable to drive a head mounted extruder. What is different about your idea?

 

My thought is more that the rods that the print head slides on would be splined shafts and they would rotate to drive an extruder mounted in the print head. You don't have a seperate component going into the print head to provide rotary motion. It is part of the bearing/rod system.

The thomson version is probably a lot more spendy, but there are others out there that might not be to bad. It would add to the cost but it would put the extruder in the head without the added mass of a motor. You could put one on the X and one on the Y and rotate them independently in the event of dual extrusion.

http://www.thomsonlinear.com/website/com/eng/products/ball_screws_and_lead_screws/ball_splines.php

or

http://www.nbcorporation.com/product/lineup/ballspline/ball_spline.html

 

[ultiarjan 1,223]

Aha... now i also get what you mean

[IRobertI 520]

I might have missed it but how would you provide the rotational force on the rods?

 

[jens 9]

I might have missed it but how would you provide the rotational force on the rods?

 

... without increasing the inertia of the printhead or the beltdriven sliders (which would mean pretty much the same)

Yes, I missed that, too.

 

[muttstang 8]

the bearings for the carriage would have to have the splined shafts on one side as well. They would transfer through a gear set within the carriage.

Maybe too much added complexity and it might be simpler to go with a small stepper on the print head with a gear reduction.

 

[willysnowman 0]

I looked at some of the designs here. Very interesting.

Here are my thoughts:

If I had to explain this from a mechanical standpoint, I would say that you want to grab a piece of flexible filament radially and apply a force and movement axially in two directions. The diameter of 2.85mm varies in some range. The cross sectional profiles varies as well. The hardness of the material and the modulus of elasticity (stiffness) of the material will vary. I don't have numbers for anything. I just know there will be a range the feeder will need to work within.

I don't think you want complete control of the material. You want some built in clutch action or slip. This would be if things go wrong; such as too much feed. The material will slip in the roller rather then break something else. I saw this in a http://www.tridimake.com/2013/03/which-hobbed-bolt-for-filament-feeder.html.

There seems to be some sort of software clutch working as in video below. How does the software know when to click back like in this video? Does the motor over amp and reset?

 

Can anyone explain what was happening from a controls standpoint? If I could measure the actual feed of the filament, could the clutch be software driven?

 

[donmilne 2]

That was the same bad assumption I made: the software doesn't cause the feeder stepper to skip back, that's entirely the stepper motor itself. It hits its current limit which means it simply doesn't have the torque to complete the step, when the active poles changes it flips to the nearest stable position - which looks like a backwards step.

Basically, unless someone in the know contradicts me, my understanding is that there was a deliberate design decision to give the UM2 a weaker feeder motor than the UM1 had, because the UM1 was perceived to have a problematic feeder mechanism - the problem being that when the filament stopped moving the motor ground a divot into it, causing the print to fail.

Basically, I think this might have been a mistake. Print failures due to underextrusion are no more useful than print failures due to grinding. In hindsight a better solution might be to keep the stronger motor but distribute the load on the filament, e.g. with two drive cogs. Of course we don't want to make the feeder so strong that it can actually damage the hotend or the bowden tube.

 

[willysnowman 0]

Ahh. It kind of seems like there is a mechanical slip (spring) and electrical (but not electronic) slip.

My thought was something like a Stratasys mechananism, only different.

I would use two drive wheels, but only drive one. The second one would be coupled to a simple encoder for feedback.

Mechanically, I can do this without a problem. I would machine the parts, wheels can be hardened toolsteel, housing aluminum.

I don't really know how to implement this into the controls. I would suppose the controls would want to go a certain number of steps. The encoder would let you know if you made it. A set slip would fire an error and pause the print.

This is still pushing the noodle rather than pulling, but I think it will work for +80% of what I do. I ideally you would want to pull the noodle and push a minimum distance.

 

[donmilne 2]

My vision was basically just a doubling up of Robert's feeder: two drive cogs on the same side of the filament, opposed by one or two (not sure) spring loaded idler bearings on an arm that can be moved aside to change filaments. The two drive cogs would themselves be driven by... something, a worm drive perhaps. Somehow they need to be driven in the same direction. These are just vague thoughts, I've haven't had time to think it through.

 

[fabi580 0]

I have found something quite similar :

Eric's Wade's Collaborative Dual Feed Extruder

But I do not think that this is the best solution. I think that the position on ether side is better.

This is an idea, the distance is fixed to keep it simple.

 

[akurtz 0]

Hi All,

I have been following this thread for quite a while and have read most of the posts and seen a lot of awesome engineering!

I got my Ultimaker2 back in September and have not started having under extrusion problems until now (I've only gone through 3/4th of a reel of filament (too much school work..)).

After thinking about this problem for quite a while and reviewing the results from other extruder designs I it is clear the problem is more caused by friction in the bowden tube than anything else. Additionally I have noticed that the filament conforms to the shape of the reel that it is on and even when unspooled returns to a tightly coiled shape (which gets more tightly coiled the further down the spool you go).

I believe if we created some type of device that could remove this coiling from the filament (basically just straighten it out like this machine:

) that we would be able to greatly reduce the amount of work the extruder thus (hopefully) solving under extrusion and grinding.

Anyone know if something like this has been implemented or have any thoughts?

Thanks,

Andrew

 

[pm_dude 27]

You would need to heat up your filament to make sure it doesn't whiten due to stress. That would ease the process too. At least... I think...

 

[donmilne 2]

I think that the position on ether side is better.

 

I understand that intuition - symmetry is always appealing. But, putting drive cogs on both sides of the filament means you can't cope with variable filament diameters, and filament changes are cumbersome. I recommend that you read this entire thread, because the complete list of design requirements includes some quite subtle points.

 

[fabi580 0]

Do you mean the list of chopmeister?

This is a fedder that is tested (they say it works almost perfect) without being able to adjust the distance. And it was an Idea it wasn't meant to be build...

So hear is version 2:

now I can cope with variable filament diameters. and it is possible to do make a hand extrusion. The filament glides are missing, but will be added. The helical gears have big teeth so they can operate without having an big backlash.

The planetary gearbox is from Kevin Lee.

The costs are ~ 22,50€:

UM drive cog: 12€

Metal: 5,50€

bearings + skews etc. ~ 5€

 

[ultiarjan 1,223]

interesting;

 

and the article;

http://3dprint.com/23569/printrbot-extruders-drumm/

 

[donmilne 2]

Hmm. I'm dubious. I don't like the loss of idler bearings. I'd want to wait until this thing acquires a decent track record, not just be swayed by a nice presentation.

 

[meduza 191]

Without spring tension, i'll think they will have a hard time getting it to work consistently with different filaments, can propably work great with some but not with all...