Pinch-wheel extruder design

I know that conventional wisdom around here is that a pinch-wheel extruder won't provide enough force to work, but does anyone have a good reason for this?

I've done a bit of research on this concept, going way back to the MkI extruder spec'd at [staff.bath.ac.uk]. While an interesting read, the only explanation it gives for including such a gratuitous number of drive wheels is that "A single-wheel pair can only be tightened on the polymer rod being fed so much before they start to squash it permenantly, so the force they exert is limited by this and by the coefficient of friction between the wheels and the rod." That is certainly a valid concern, but is there actually any evidence that the a higher-friction roller won't be able to take care of it by itself, considering the more resilient plastics (eg. ABS) that the RepRap project has begun experimenting with since 2005?

For example, I'm looking at the RepStrap extruder drive mechanism being built by the fabr guy (picture at [www.ooeygui.com]). This seems much less complicated that the MkII extruding screw, while serving the same purpose and even allowing for non-flexible feed rods as a bonus. It is certainly easier to fabricate, whether you have a RepRap or just conventional shop tools! Related note: I also like how he opted to use a stepper motor to drive the extruder. I'm sure that cheap DC motors like the GM3 currently used are the way forward for this project, but for now, trying to control it seems like more of a hassle than it's worth. No matter how I implement the extruder, I'm going to consider upgrading the drive motor to a stepper.

Anyway, does anybody have an estimate for how much force the drive system actually needs to be able to exert on the filament? In other words, does anyone have an estimate of the pressure in the heating chamber during extrusion? That seems like the biggest potential deal-breaker for this idea.

Thanks,
Kyle

I did some rough calculations a long time ago and arrived at the notion that the extruder barrel pressure was running about 17-18 bar. I got at that number from rigging a scale gripped to the HDPE filament and seeing what force was required to push it into the heated barrel and get it to extrude.

That sounds like a lot, but it really isn't when you consider just how small the cross-section of the filament actually is.

Thanks Forrest.

Doing some napkin math, assuming 18 bar and a .15cm radius filament (and check me if I'm wrong):

18 bar * ((14.5lbs/in^2)/1 bar) * (4.45N/1 lb) * (1 in/2.54 cm)^2 * (.15cm)^2 ~= 4.05N of force - that's not that much.

Just for fun, I'll make some more warrantless assumptions and see how they turn out. Let's say we've got a 1cm diameter drive wheel and the extruder is only 50% efficient. I think this means we'd need a motor with 4.05N*2 * .5cm = 4.05 N*cm of torque - that's not very much.

The tin can motors that Forrest has been experimenting with might well work. It looks like with a 0.5cm drive wheel the smaller 26000 series motor may even be able to provide enough force (the datasheet claims it has 1.13N*cm of torque).

And the GM3 itself provides about 35N*cm of torque, making it overkill for a drive system like this!

Thoughts?

Forrest,

Thanks for the estimate of extruder pressure. -- I was wondering how high it was.

----
Kyle, I think you're missing a factor of Pi for the filament area.

18 bar * ((14.5lbs/in^2)/1 bar) * (4.45N/1 lb) *
(1 in/2.54 cm)^2 * PI *(.15cm)^2 ~= 12.7 Newtons -

Still not that much force.
----

For a pinch-wheel design, we would probably need to gear down, not for more torque, but to get good speed regulation at the desired feed speed.

Nophead reported extruding 0.5 mm diam (ABS) at 16 mm/sec. The feedstock is 6x bigger in diameter, so the in-feed speed should be only:
16 mm/sec * (3.0 / 0.5) ^2 = 0.44 mm/sec.
This needs either a small diameter wheel, slow rotation, or both. Small wheel diam. is likely to lead to more problems denting the filament. If we let the drivewheel be ~same as the filament (hand-waving arg. here re Hertzian contact stress), then we'd need that roller to turn at:
3*PI / 0.44 mm/sec = one rev every 21.4... sec, or ~3 PRM. Slower if the drivewheel were larger. Can a GM3 (or GM17) run reasonably steadily at such low rates? Perhaps with encoder feedback, as Nophead uses.

What were the problems with the pinch-wheel drive that prompted changeover to the screwdrive?

-- Larry

Larry_Pfeffer Wrote:
-------------------------------------------------------
>
> What were the problems with the pinch-wheel drive
> that prompted changeover to the screwdrive?
>


Adrian did this, but let me see if I can remember the reason he said that he abandoned that design.

IIRC, it just didn't work reliably enough to justify the complexity of the device. It certainly was big and complicated compared to the Mk II which he and Vik came up with late that same year.

The force required to extrude depends on the plastic. ABS extrudes very easily and is not very slippery, so maybe a pinch wheel would work but at the other extreme HDPE requires several Kg and is very slippery. PCL and PLA are probably in between.

I prefer to keep my options open rather than optimise for one plastic.

The GM3 certainly is not overkill. I operate it beyond it's max torque by locking the clutch.

I also plan to try a stepper but it may need some gearing as well. The big Keling motors should have enough torque without gears but they are very heavy to mount on a moving head machine. No problem on HydraRaptor though.

I am also going to try fitting one of the tiny tin can motors to a GM3 gearbox. The shaft size is the same. It will be way too slow with the standard gearing but you can reconfigure the gearbox I think.

Another mad idea I had was to feed the filament down stiff tube, like a you get round bike brake cables. That way the feed motor could be distant from the nozzle and stationary.

---

[www.hydraraptor.blogspot.com]

@Larry: Doh. Good catch with the "pi" thing... that was dumb.

I'm afraid the stiff tube idea may lead to more variability in the extrusion rate as the tube moves around - what advantage would it bring?

And for what it's worth, I emailed Lou from Fabr to ask about his design. I think it's ok if I quote him directly: "The extruder was a simple design, which seemed to match existing commercial ones. I attempted to build a screw based extruder, but had lots of problems getting the screw to 'bite' into the hdpe I'm using for testing. The rubberized pulley actually has amazing grip with little pressure. It works in test conditions, but am in the process of hooking it all up and driving it with the gcode interpreter."

I read that as saying that it may work even for the worst-case HDPE. Of course, he hasn't tested it in actual operation yet, but I'm cautiously optimistic based on that statement. The key may be in the choice of roller material - urethane is used pretty commonly when high friction is needed. I tried to find its CoF with these plastics online, but didn't have much luck. I'm sure it's high though.

I'm sure he'll continue blogging progress as it happens. This will be interesting to watch.

Quote

I'm afraid the stiff tube idea may lead to more variability in the extrusion rate as the tube moves around - what advantage would it bring?

If you think about how brakes on a bike work: you apply the pressure between the cable and the flexible metal tube that contains it. Moving the cable about does not activate the brakes, only a differential force between the cable and the sleeve.

The advantage is that you can use a big heavy stationary extruder motor and a very small and light heated nozzle that moves.

Might not work at all though, just an off the wall idea I had.

---

[www.hydraraptor.blogspot.com]

I am working on an alternative feeder design - described elswhere on this forum.
Its good to have some pressure data
I have HDPE on order so it looks like I will start tests there.

The bowden cable (bike brake) idea is interesting. My Brother did some calcualtions when we were re-designing our car's handbrake arrangement - avoid bends at all costs - massive friction occurs. Of course these are all in tension, I would assume under pressure the filament would tend to buckle and 'squirm' in the tube which may or may not cause problems.

I could only find the coefficient of friction for urethane on metals. For steel, it was about 0.5, 0.6 for aluminum and 0.7 for brass. This was with the highest grip urethane, "Cyclothane-A 83A". This was from a properties sheet at:
[www.durabelt.com]

Pinch wheels were also discussed on the "Cheapo Mouse as Shaft Encoder" thread at:
[forums.reprap.org]

In that thread Dylan showed a picture of his lego & plywood pinch wheel, and said it worked well.

In theory it looks like pinch wheels could handle at least ABS and metals.

nophead Wrote:
-------------------------------------------------------
> > I'm afraid the stiff tube idea may lead to more
> variability in the extrusion rate as the tube
> moves around - what advantage would it bring?
>
> If you think about how brakes on a bike work: you
> apply the pressure between the cable and the
> flexible metal tube that contains it. Moving the
> cable about does not activate the brakes, only a
> differential force between the cable and the
> sleeve.
>
> The advantage is that you can use a big heavy
> stationary extruder motor and a very small and
> light heated nozzle that moves.
>
> Might not work at all though, just an off the wall
> idea I had.
>
Wow! That's a really nice idea for Tommelise. I could make the gantry a LOT lighter with that kind of arrangement and use that car window gearmotor mounted on the printer base to drive just about any plastic filament that I please.

Beautiful!

McMaster has 1/8" and 3mm I.D. PTFE tubing. Looks to be about $2 per foot.