Hot end & cold end of extruder separate,.. disadvantages?

I just saw the photo's of the sumpod and noticed that the extruder is split with the cold end bolted to the back of the case. This seems like a very sensible idea to me,.. less weight being moved around, less strain on the axis motors, with possible improvements in speed & accuracy.

I'm guessing though that since there aren't many other printers that use this method that there must be a disadvantage or 2 that I'm missing,... please enlighten me!!

The disadvantage is the springiness of the filament. Each movement of the extruder side leads to a delayed reaction at the nozzle.
At different rates in either direction.
This is difficult to compensate.

I assume this problem would be minimised by using the shortest run possible, by using 3mm filament and by using a tight fitting sheath?

[reprap.org]
There are plenty of repraps that use a Bowden Extruder, like the Huxley, but it seems nearly impossible to overcome hysterisis, which causes oozing, and necessitates using retraction before non-printing moves of the hot end.

Njones Wrote:
-------------------------------------------------------
> I assume this problem would be minimised by using
> the shortest run possible, by using 3mm filament
> and by using a tight fitting sheath?


To some extent maybe.
The problem is: Thicker filament requires more pressure at the nozzle and the less play is between the bowden cable and the filament the more friction and thus transmission loss.
As it is now they simply retract the filament before extrusion should stop and verse visa. This does help but is more or less a weak compromise.
It should theoretically possible to compensate for the springiness if the movement of the axis would be taken into consideration as well as the extruder.
But that is something requiring some serious effort to figure out the math and develop the firmware, a tough task.... I know I can't do it and many other too...

Edited 1 time(s). Last edit at 07/25/2011 04:52PM by ElectricMucus.

The other disadvantage of Bowden is it puts more drag on the carriage (because it is thicker and stiffer than just a filament) so will increase backlash.

---

[www.hydraraptor.blogspot.com]

From my experience with both, the bowden extruder vs standard extruder boils down to a speed vs precision trade-off.

The carriage can be ridiculously light with a bowden extruder which allows you to print quite fast, but at the cost of blobbing at corners and other places where there are abrupt changes in speed/direction. Some objects will still turn out quite nicely, but there are others which will be a constant headache trying to make print well.

In my opinion, Bowden extruders should only be used by a machine with the Ultimaker style gantry. Otherwise, you are still limited by the weight of the bed and can't really achieve the extra speed needed to make the loss in precision worthwhile.

As a compromise, what if an extruder were set up with two motors - a large, torquey, stationary Nema-23 frame motor, and a small lightweight Nema 14 frame or smaller on the print head? Although it would be a more expensive setup with two motors, it might be an interesting option for a high performance system.

I'll take a look at the math and see if this actually is a sensible proposition or if it's just crazy.

Now small, powerful stepper motors are being used as direct drive extruders (like the Makerbot stepstruder Mk6 and the new Mendelparts v9), extruder/hot end size is coming down. Though the motor is the weightiest part, it is clear that this is not an issue for most machines that use NEMA17 size motors for driving the axes. Having a Bowden cable reprap myself, due to limitations of my reprap frame size, I think the case for Bowden cables is limited to where there are size/clearance issues, and then it's a compromise between build area size and print quality.

I'm not sure having two motor, especially two different size ones, would be very workable. But maybe some kind of servo clamp/lock which initiates at the start of a move would work.

If the extruder really can't be mounted on the hotend, another option would be remotely driving an extruder wheel. Flexible drive shafts have been considered in the past, but, if the hot end movement is limited to one axis (ie X) it could also be done with a stationary motor at the end of the axis driving a square rod along the line of the axis, turning an extruder wheel which is able to slide up and down the drive shaft with the carriage.

This topic is interesting but recently the best output I have seen is from the bowden fed machines. Emaker videos show good quality prints with no talk of problems printing. Neither does ultimaker. I don't understand why ultimaker a better implementation of a bowden extruder. Seems the same as the emaker.

Maybe theoretically direct is better but it doesn't seem to hold up given the quality of bowden fed.

In fact makerbot has only recently made a machine that seems to work consistently according to forums over the past year. Even the output of the new one isn't guaranteed.

I think its down to how much tweaking is done to the settings for the specific machine.

Perhaps we need someone with an ultimaker to step up so we can do a side by side comparison of direct drive vs bowden on a variety of parts.

I personally think that the bowden style extruder is the future. The only thing holding many ppl back from multiple nozzles on one head is the software,.. as soon as that is sorted there will most likely be one large nozzle for quick/rough work and one fine nozzle for the visible surfaces,... possibly a few pairs of nozzles to allow for multi colour printing. It'd be impossible to mount that many motors on the head, so I expect the bowden style will be the only choice.

You only need one motor, and some form of filament select. Look at the dimension printers. One motor, two filaments...

i think you could use 2 motors for multiple extruders, one motor for the pinchwheel and one motor moves a ring of nozzles around like a turret. and possibly when it moves mechanically lifts the idler bearing of the filament then clamps back down on the new when the nozzles have moved around.

like a microscope with changing lenses, or a projection turrent changing lens

the hardest would be stopping the nozzles from dribbling when they have done their job and also having them purged and ready to go, it would need a purge section for the nozzles or the ring mounted slightly on an angle or something so the other nozzles lift slightly up and into a very thin case which stops the flow keeping it ready to start again

Andrew Diehl Wrote:
-------------------------------------------------------
> You only need one motor, and some form of filament
> select. Look at the dimension printers. One motor,
> two filaments...


I like this idea.
Imagine a gearbox driven by a servo. Since the extruder is driven by a nowadays shifting could be done by detecting a stall of the servo, moving it back a little, turning the stepper 1/2 of a tooth, try again, turning 1/4 of a tooth in the other direction, 1/8, 1/16 and so on. Changing gears might take a while but could take place without introducing too much backlash. A servo controlled gearbox would mean additional bulk though which we are trying to avoid in the first place.

@Njones - sorry, but none of the top-end machines use Bowden cables. Arguably the best prints come from Stratasys printers, which don't (see the teardown of a Stratasys printer here [haveblue.org]), and possibly the best cheap commercial printer - the UP! - doesn't either. In the home-built world, Nophead's highly regarded hydraraptor doesn't use a Bowden cable. So I'd love to know what prints you've been looking at...
There is, perhaps, a speed advantage to using a Bowden to lighten the printhead, but high speed requires really accurate extrusion; using a Bowden means this can't be achieved due to the inherent properties of the plastic filament.
You also suggest using a Bowden to reduce the size/weight of the print head when multiple print heads are used. While this is true, the extruders for this would still have to be mounted somewhere, so machine size would still be large. Powerful stepper motors can easily overcome the weight penalty of either multiple extruders, or a system to move a single motor output between filaments.
Having built a repstrap with one, Bowden cables are a compromise, and should only be used where space is very limited; better quality AND (arguably) speed can be achieved with direct drive extruders.

From the video linked below the UP printer does seem to use a tube to hold the filament and a Bowden style extruder.

[www.youtube.com]

But you do have a point about extruder accuracy, responsiveness, etc. Pushing a fiber through a tube will always add some "slop" and friction in the system. But how much is too much?

That's just a small motor to feed the filament to the extruder which has the stepper motor to drive the filament. When the extruder pulls enough plastic, the outer tube pushed down on the switch on the small motor which causes the motor to run for a short time to give the filament some slack. So, it's not really a bowden extruder. l

ElectricMucus Wrote:
-------------------------------------------------------
> Andrew Diehl Wrote:
> --------------------------------------------------
> -----
> > You only need one motor, and some form of
> filament
> > select. Look at the dimension printers. One
> motor,
> > two filaments...
>
>
> I like this idea.
> Imagine a gearbox driven by a servo. Since the
> extruder is driven by a nowadays shifting could be
> done by detecting a stall of the servo, moving it
> back a little, turning the stepper 1/2 of a tooth,
> try again, turning 1/4 of a tooth in the other
> direction, 1/8, 1/16 and so on. Changing gears
> might take a while but could take place without
> introducing too much backlash. A servo controlled
> gearbox would mean additional bulk though which we
> are trying to avoid in the first place.

Stratasys came up with an ingenious solution to that. They bump a selector button on the head at the axis ends to select filament. No second motor, servos etc. Just pure mechanics to select filaments.

Andrew Diehl Wrote:
> Stratasys came up with an ingenious solution to
> that. They bump a selector button on the head at
> the axis ends to select filament. No second motor,
> servos etc. Just pure mechanics to select
> filaments.

From [haveblue.org] I'm not sure that's how it works, he says there are two extruder motors, one was removed in the 4th picture. I think the solenoid lifts the support extruder up a bit so it doesn't drag through the build.

With what seems capable to print the Ultimaker and the eMaker-Huxley, I think the bowden is a viable option (I bought an eMaker to experiment the bowden feeding, we will see :p).

But on the other hand, the simplicity of the direct drive is very appreciable too (tested on a Orca+V9).

Having one set of mechanics and a filament selector device is never going to work in the long term. The simple fact is that 3D printing is SLOW. Having multiple extruder's working simultaneously, perhaps using different colour PLA or different nozzle sizes is a simple way to speed up the whole process. Perhaps each of the multiple heads could have a selector system, so all could be laying down one colour if the design required it.

I expect the mechanical problems involved in a setup like this could be overcome very easily,.. the software however would probably be hugely complex.

Njones Wrote:
-------------------------------------------------------
> Having one set of mechanics and a filament
> selector device is never going to work in the long
> term. The simple fact is that 3D printing is SLOW.
> Having multiple extruder's working simultaneously,
> perhaps using different colour PLA or different
> nozzle sizes is a simple way to speed up the whole
> process. Perhaps each of the multiple heads could
> have a selector system, so all could be laying
> down one colour if the design required it.
>
> I expect the mechanical problems involved in a
> setup like this could be overcome very easily,..
> the software however would probably be hugely
> complex.
How?

I can think of no feasible option.

Cartesian robots would need the ability to avoid collision by lifting (jumping) over the lower axes.
SCARA would need to avoid certain movements, you couldn't print infill and outline at the same time.
Delta robots cannot even use the same space at all.

The other 2 options left would need excessive effort.
Separate spaces connected by a linear or radial conveyor.
A tree structure of robots.

what's left? did I miss something?

Let's be honest. In the long run FDM is going to be far surpassed by other technologies.

We're talking near-long-term of 3-7 years.

droftarts Wrote:
-------------------------------------------------------
> Now small, powerful stepper motors are being used
> as direct drive extruders (like the Makerbot
> stepstruder Mk6 and the new Mendelparts v9),
> extruder/hot end size is coming down.

I wasn't sure but it seems that on the v9 it's a classic nema17 as already used, with a brass insert like on the Adrian [flic.kr] (and I just realized that makerbot parts are more more expensive than I thought).

Don't know for 3mm filament but it worked fine with 1,75mm

The FDM series machines as shown on my blog and John Branlund's blog (http://3d4u.org/MyFDM/) had two motors. The much newer Dimension series machines have the toggle bar to shift drive power to either the support or model nozzle.

Brad Rigdon of printto3d shows how the toggle bar works on his Dimension at about 0:50 -
[www.youtube.com]


droftarts Wrote:
-------------------------------------------------------
> Andrew Diehl Wrote:
> > Stratasys came up with an ingenious solution to
> > that. They bump a selector button on the head
> at
> > the axis ends to select filament. No second
> motor,
> > servos etc. Just pure mechanics to select
> > filaments.
>
> From [haveblue.org] I'm not sure
> that's how it works, he says there are two
> extruder motors, one was removed in the 4th
> picture. I think the solenoid lifts the support
> extruder up a bit so it doesn't drag through the
> build.

---

[haveblue.org]

Apologies if I'm late on this topic, but anybody knows how to keep the bowden in place on both ends?
I used one for 3mm filament and 2 M5 nuts work perfectly, but what about the thinner bowden for 1.7mm filament?
Thanks
Syncra

Hi Syncra,

I used 4mm Pneumatic Push-fit connectors to hold the PTFE tube 4mm outer 2mm inner. That works really well.

More info on this page here - [richrap.blogspot.com]

---

[richrap.blogspot.com]

Looks really good! Where did you buy push-fit components and bowden cable? Is it PTFE?
Thanks again

They are just normal Pneumatic connectors, I had some 6mm ones and ordered 4mm ones from RS. The Tube is PTFE and was from Mendel-parts.
You can also buy from Adtech or one of their distributors.

---

[richrap.blogspot.com]