Bowden extruder control

Hello everyone,

Has anybody successfully printed with a bowden fed extruder?

My repstrap is now up and running, and can physically print, but I am having a real headache controlling the flow from by extruder nozzle. I am feeding the 3mm diameter ABS filament to the hot end via approx 1m long PTFE tube, where it is heated to extrude at 225 Celsius through a 0.5mm diameter nozzle.

I am finding the hysteresis from feeding the filament through the bowden tubing to be hard to manage. Long XY moves are fine because things are in a fairly steady state, however, during a Z move between layers, the ABS just keeps coming out of the nozzle giving me large splodges of material.

I have tried adding a 20mm reversal move to the extruder before or after the Z move (using RepSnapper), to no avail.

I am considering repositioning my filament feed mechanism to allow me to shorten the bowden cable hopefully down to around 500mm, to reduce the hysteresis.

I would be interested to know if anyone else has managed to successfully control a bowden extruder, and the method employed to achieve good prints with it.

Thanks,
jmg

---

[emaker.io]

[emaker.limited]

Havent tried and not on my to do list, but if i would try that, and run into this problem, then i would go for an item i dont really know how to describe:

Some cars have a "tube" to mechanically conduct a steel wire from one place to another with across a small drive gap: it is basically a high tensile spring covered in plastic (spring in the inside), usually long like 50 cm - 1 meter, and usually used in older cars, in things like opening the ventilation shafts / use for "shock" button to open the carburator feed for cold starts on petrol engines / sometimes on acceleration from pedal to carburator / maybe on older speedmeter drive / etc.

And i think it should be fairly easy to find one like that to take a 3 mm filament inside (tho not sure on the diameter but i know there is a good variety of these). And the steel springs covered in platic would ensure a minimal filament and tube bend, if any. Actually some of these items are usually pretty hard to bend even by hand, because of the solid spring they normally use.

The friction inside should be acceptable on the spring inside and abs is stiff enough to work with this in my imagination. This would be my first try on something like this, just mentioning maybe it can give an inspiration. Its actually a tube that was designed to do this - just normally with a steel wire instead of abs.

I'm not using a bowden, but i'd suggest you make the move to skeinforge. it's got a steep learning curve, but it's worth it in the end. i just started getting it working and it's much better than repsnapper at producing gcode. repsnapper produces 'adequate' gcode, but my results with skeinforge are looking professional.

I've got a theory that large, fast retraction moves might not be the most ideal. I think there might be a fine balance between pulling out the plug and creating negative pressure, (in turn retracting the filament into the nozzle) and just retracting the unmelted feedstock, creating a spikey end. Try playing around in repsnapper with the manual extrude control till you get a good speed/distance combo that will suck the filament back into the nozzle.

Just an addition, when using Skeinforge to produce 5D gcodes (you have to enable the Dimension feature) there is an option for entering how many mm of feedstock to push/pull at the beginning and end of extrusion moves.

In my setup, my bowden cable is only about 400mm, but I have this value set to 200 mm. (This value is proportional to he extruded output, not the feedstock input to the extruder)

This works, however there are small blobs at the beginning and end of extrusion. The result of this is that each layer usually has a small bump at varied positions on the outline. Tuning my firmware for faster extruder speeds has improved this... but quadrupling the speed of the push/pull only halved the size of the blobs.

Just for comparison though, the printed objects are FAR cleaner when done in Skeinforge than anything I've seen Repsnapper produce.

***Edit the cable length is 400 mm not 40

Edited 1 time(s). Last edit at 02/04/2011 01:15PM by AphoticJezter.

Hey guys,

I took the bowden off my repstrap in the end, but I am using one on my eMAKER Huxley.

This is feeding 1.75mm filament through a 250mm long bowden tube, (AphoticJester: how is your bowden tube only 40mm long?!), and after a few hours of tuning I have almost got rid of all blobs. I'm having to use quite large reversal moves (35mm).

My gcode isn't straight skeinforge either. I use skeinforge to produce minimal gcode just to give me the print path. I then process this using a python script to produce the gcode I print with.

I have it setup so that I can tune how far before the end of a thread my extruder starts reversing, and how much I move along at the start of a thread whilst bringing the filament back to where it was before reversal. My thinking is that there will always be some time delay between what my extruder stepper motor does, and what actually happens out of the nozzle.

I also use the script to join any threads which end and start within 2mm of each other, to remove unnecessary reversal moves, and I have skeinforge setup to do Loops->Perimeter->Fill with 2 shells so it doesn't move up a layer at the edge of the component (so if I do have a blob, it won't be visible on the finished part).

I am still tuning though, so I expect to have to tweak things even more.

---

[emaker.io]

[emaker.limited]

Bowden Cable Alternatives.

I'm thinking of changing the location of the Extrude feeder motor to where the X-Axis Motor is.

A uniform Cross-Section shaft could be used to transmit the motion of the motor to the first stage pinion of the extruder (logic: I think this will transmit hi speed/low torque better than it will transmit the lower speed/higher torque). Either a round shaft with a flat or a "trifoil" arrangement with three BB's within the hollow shaft of the Extruder Pinion could do the job.

The Uniform Cross-Sec would allow the pinion to be moved along the shaft without the "X-Co-ordinates" of the print head affecting the amount being extruded.*

(*with the premise of large polar moment of area and/or modulus to reduce the amount of torsion in the bar at the extruder).

I've not done any calcs on this yet, but think it may be a better way of controling feedrate relaiably than allowing a Bowden cable arrangement (where the filament is under buckling/compressive strain).

I'm as yet to build my "Rep/Strap", but I just thought I'd pitch this and ask if anyone has tried this config on an extruder head yet?

(n.b. This arrangement is not compatible with the cartesian configuration of my RepStrap, but will hopefully work with more conventional Mendels).

Points and pointers are welcome.

-AS
(openly Nube).

I can't visualize this. Could you do up a napkin drawing?

Andysuth,

This sounds like an idea I had and then shot down in another thread. It might be more appropriate here, though.

For all the torque it will carry, a piece of aluminum square tubing should work for your "shaft of uniform cross section". (Look up "spline shaft" for a commercial equivalent). For the square tubing "shaft", the pinion would have a (probably) printable matching square hole in it. If a triangular extrusion is commonly available, that might be even better. I know the quare is pretty commonly available. For either profile, I don't think they will introduce appreciable backlash. If backlash is a problem, the fit could be adjusted by filing or sanding. I don't think ball bearings will be necessary.

What is your goal for this configuration? Lower moving mass in the print head?

HTH

jmgiacalone Wrote:
-------------------------------------------------------

> long bowden tube, (AphoticJester: how is your
> bowden tube only 40mm long?!)

My bad, it's 400 mm, not 40 (I've edited my original post to reflect this)

What about having an additional tiny stepper on the carriage, like a NEMA 11?

the NEMA 17 would provide the main pushing force, but the secondary stepper would control the retraction and help overcome hysteresis issues.

I guess you would need equal gearing on each stepper so that they both moved the same distance, which somewhat defeats the weight/space savings inherent in bowden designs. And i doubt that you can gang two dissimilar steppers to the same driver, so you would need to add another stepper driver, and edit the firmware.

Buback Wrote:
-------------------------------------------------------
> I can't visualize this. Could you do up a napkin
> drawing?


Buback,

Are you talking to me, or the person who's thread I rudely hijacked? ;p

I'll throw together a few calcs on the torsion, I'll do a few ProE models (or is a different format more acceptable?) and post it somewhere in the next few days.

Should we move this topic to its own thread?

HTH,

Yes, similar, but I'd have possibly used PTFE sleeve over S/Steel pair of shafts or profiled shaft.

Two round shaftsnearly touching would be able to transmit torque with no need for machining the shaft.

-Andy

yep andy that was directed at you. sometimes a picture, even badly drawn, says a thousand words.

too true!

I'll throw one on the site in the next few days!

-AS

[www.youtube.com]

Here are some very quick 3d Models I did. Nothing too fancy, no bearing configs. etc.

The Animation is on YouTube for discussion.

The concept is to remove from the print head the Stepper used for driving the ABS strip through the nozzle.

This will remove the mass from the print head and give faster acceleration, smoother motion and hopefully stop overshoot.
This only works if the force to overcome the friction along this shaft is less than that needed to accelerate the motor.

This is not to scale, has had no engineering calcs performed and is meant for a discussion piece.

Three h6 S/Steel tubes could be used for the rotating shaft and go into a reprap'd pinion with PTFE sleeves in it. I've not said Aluminium, because of the temptation to use extruded Al Tube, which has a fair bit of mold jitter fromt he etruder head.

Either this lot has to be smaller than the 8mm ID of standard bearings used (608RS) or new, larger bearings have to be used.

The Three Shaft config has been used as little machining is required, whereas if a shaft with a keyway or other profile were to be used, more precise machining would be required.

I'll keep you informed as to how this concept develops.

Next step is calculation, so I'll have to look into the spec's of existing Wade/Adrian Geared Extruders.

-Andy

EDIT: I'VE TRIED TO SHIFT THIS TO ANOTHER PART OF THE SITE AND NOT HIJACK THIS THREAD.

-AS

Edited 3 time(s). Last edit at 02/08/2011 04:12AM by andysuth.

How do you intend to keep the small gear (the one on the shafts) coming off the big gear when you move the carriage back and forth?

Ideas:
Print a lip on both sides of the big gear. The issue is that you need a printer that can do overhang for one of them, or you have to print the gear in 2 or more parts (eg: like the Mendel Z-axis Pulley).
Use a herringbone gear, which will not go sideways due to the herringbone teeth. The issue here is that they tend to need fairly fine-tuned printing capabilities.

Why not just put the gears on the motor and drive the pinch wheel directly with the shaft(s)? If the gear is on the printhead, then it will use up more room there and reduce the maximum build height.

Cefiar Wrote:
-------------------------------------------------------
> How do you intend to keep the small gear (the one
> on the shafts) coming off the big gear when you
> move the carriage back and forth?
>
> Ideas:
> Print a lip on both sides of the big gear. The
> issue is that you need a printer that can do
> overhang for one of them, or you have to print the
> gear in 2 or more parts (eg: like the Mendel
> Z-axis Pulley).
> Use a herringbone gear, which will not go
> sideways due to the herringbone teeth. The issue
> here is that they tend to need fairly fine-tuned
> printing capabilities.

Yes they both sound great. I was going to contain the shaft with two 608 bearings. Whilst they are not supposed to take side loads, they should be able to cope with the small loads from friction (the whole concept here relies on friction being minimum).

Thanks for your interest.

-Andy.

andysuth Wrote:
> Yes they both sound great. I was going to contain
> the shaft with two 608 bearings. Whilst they are
> not supposed to take side loads, they should be
> able to cope with the small loads from friction
> (the whole concept here relies on friction being
> minimum).

I't no so bad. 608 is deep grove, so from [www.skf.com]

If deep groove ball bearings are subjected to purely axial load, this axial load should generally not exceed the value of 0,5 C0. Small bearings (bore diameter up to approx. 12 mm) and light series bearings (Diameter Series 8, 9, 0, and 1) should not be subjected to an axial load greater than 0,25 C0. Excessive axial loads can lead to a considerable reduction in bearing service life.

ledvinap Wrote:
-------------------------------------------------------
> andysuth Wrote:
> > Yes they both sound great. I was going to
> contain
> > the shaft with two 608 bearings. Whilst they
> are
> > not supposed to take side loads, they should be
> > able to cope with the small loads from friction
> > (the whole concept here relies on friction
> being
> > minimum).
>
> I't no so bad. 608 is deep grove, so from
> [www.skf.com]
> n&maincatalogue=1&newlink=1_1_13
>
> If deep groove ball bearings are subjected to
> purely axial load, this axial load should
> generally not exceed the value of 0,5 C0. Small
> bearings (bore diameter up to approx. 12 mm) and
> light series bearings (Diameter Series 8, 9, 0,
> and 1) should not be subjected to an axial load
> greater than 0,25 C0. Excessive axial loads can
> lead to a considerable reduction in bearing
> service life.

Nice find on the datasheet there Ledvinap, though a lot of the bearings used on repraps are non-SKF bearings, eBay specials, so I'd be a little cautious as SKF is one of the higher quality bearings and may be a little more capable than the 608rs that I'm using.

Do you mind if I just relocate this reply to the other thread I've started, when I put my first post here I didn't know how to start the new thread?

Cheers,

-Andy