I'm working on an extruder based on a hopper/auger concept using readily available parts. I used the extruder diagram at [en.wikipedia.org] as the reference for my design. The variable size and thread spacing of the auger isn't quite possible for me without machining a feed-drive. This goes against my design principle of using all readily available parts.

With the parts I could find at the hardware store, here's the design as far as it is currently (pictures to come later tonight)

The feed auger will be a 5/8 x 4" wood drill. (I haven't measured it yet, but I believe that a 5/8" wood drill has a body that is fractionally less than 1/2, making it ideal for 1/2" fittings.)

The heater barrel is a 1/2" by 2" brass tube, called a "nipple" in the plumbing industry (fits a 5/8" bit perfectly) .

The extruder head is composed of three parts: A female-female 1/2-1/4" reduction fitting, a male-female 1/4" elbow and a 1/4" square head plug is the nozzle. All of these parts need to be part of the heating loop that also contains the heater barrel.

The hopper is connected with a PVC threaded 1/2" T fitting.

Here are my build instructions as they stand:
1) Prep the plug the same way as an acorn nut to create a nozzle, then mount it in the elbow. Mount the elbow to the reduction fitting.

2) Drop the drill bit through the brass tube backwards (the head of the bit is a little too big to fit through the pipe, but we'll fix that later). Now screw the pipe into the PVC T. Connect a drill to the bit and drill out the PVC T using the brass tube as a guide.

My outstanding issues are:
1) How to I make the drill bit fit in the brass tube?
I haven't decided whether to modify the bit with an angle grinder, dremel tool, or to ream out the brass tube to create

2) What do I use for a hopper? Some sort of funnel seems ideal, but I need a mechanism to prevent jams in the feed before the T.

3) What do I use as an insulator between heater barrel and the T and how do I mount the insulator? My current thinking is either glass tube with an inner diameter of 1/2" or some kind of ceramic washer.

4) What do I use for a bearing to keep the drill bit's shaft in place?

5) How do I create a reliable way to mount this whole thing to a reprap frame.

6) How do I couple the drill shaft into the motor drive without interfering with the object under construction?

7) will the drill shaft transmit too much heat and cause the plastic T to melt? Is a steel T more advisable? (A steel T might take more machining if the actual inner diameter is less than 1/2")

I don't think these issues are insurmountable, but I'd sure like some ideas on where to look next!

This extruder could, potentially, take less tooling or less complicated tooling to build than the darwin extruder.

My plan for controlling plastic flow is to use the motor current and rpm to determine torque, or use a force transducer on the drive chain of the drill bit shaft. I'll combine this applied force with rotation to determine/control flow rate.

Here's what I've got so far:
The parts:

Preparing to ream the T:


I couldn't quite hold on to the T as I drilled it, so here is my makeshift handle:


My progress so far:


This is where my questions above come in. I'm most concerned about the insulator right now.

You may have the basis of an extruder for ABS filament from granule. The drill ideally wants to be used in reverse as that has the sharp edges to form a seal. I tried a large bit in some of my oil press experiments. Unless you have a very close fitting metal tube the auger can bend and shear under load, especially at higher temperatures take care. I will re look again at my notes.

I had to machine the ends to allow the use of a plain bearing for support and make a drive dog to power the auger in reverse. Use earthed metal components, as plastic parts and funnels quickly build up static, gripping the granules.

Joules

That PVC fitting isn't going to do the job for insulating. No way on Earth is it going to work. PVC isn't even approved for carrying hot water, never mind what you're trying.

Edited 1 time(s). Last edit at 01/16/2009 08:08AM by Forrest Higgs.

---

-------------------------------------------------------

Hell, there are no rules here - we're trying to accomplish something.

Opportunity is missed by most people because it is dressed in overalls and looks like work.

Thomas A. Edison

Yes PVC has a high melting point but a low glass transition so it won't hold its shape above 81C even though it does not melt (become a fluid) till ~200C.

---

[www.hydraraptor.blogspot.com]

Do you suppose that a stainless steel fitting would resist heat flow enough to do the job, Nop?

---

-------------------------------------------------------

Hell, there are no rules here - we're trying to accomplish something.

Opportunity is missed by most people because it is dressed in overalls and looks like work.

Thomas A. Edison

I expect you could have a stainless steel section preceeded by a substantial heatsink.

If we could get brass, stainless and aluminium fittings that all mate with each other that would be ideal.

---

[www.hydraraptor.blogspot.com]

Perhaps a brass nozzle and melt zone, a short section of stainless pipe leadin to block of aluminium with a big heatsink attached.

You need to keep the SS section short because there is always a problem point where the plastic has softened but not melted. It just expands to plug the tube. The thermal gradient has to be steep enough that the problem area is short enough to not cause a jam.

---

[www.hydraraptor.blogspot.com]

Wasn't there an experiment like this done a while ago? I have a vague recollection of there being a problem due to a metal auger conducting heat back toward the granules, causing clogging.

I didn't expect the PVC to work at high temperature; I knew I'd need an insulator between the brass and the PVC. Steel would be fine (and there are mating couplings in steel) but if granules are melted before they hit the auger, it's game over.

I'm using MIP couplings and in MIP, steel, brass, and plastic all couple together.

If the problem is the auger conducting heat, I can get a longer auger.

An alternative to heatsinking: A peltier device, applied properly, could guarantee the temperature of the entry to the T-fitting.

Peltier devices are not very efficient, IIRC <30%, and generally need a heatsink themselves. I think in this case a simple heatsink and fan would work a lot better.

---

[www.hydraraptor.blogspot.com]

Stainless steel has about 1/3 the conductivity of steel.

---

[www.hydraraptor.blogspot.com]

Sadly stainless steel isn't widely available in MIP fittings. The design ethic behind this granule extruder is parts you can get anywhere.

Peltier devices can be run at 300% efficiency. It all depends on how big the temperature differential is, so keeping the hot side cool is really important. A water jacket is another option, but it has some significant drawbacks.

I looked up the relevant thermal conductivities. Copper is 380W/mK, stainless is anywhere between 12W/mK and 45W/mK. Glass is 1.1 W/mK. I stand by my original assertion that I should use glass pipe as a thermal break. I still need to find a way to limit how much heat the auger conducts.

I was thinking about something like that too - and my general conclusion is that there's no point in making an actual granule-fed extruder, because it will be heavy (making overshoots a lot worse due to increased inertia, especially with cheap motors or those "pseudo-steppers" meant for use with Mendel), big and possibly highly unreliable during bursts of short, rapid head movements. Better to keep improving the extruder as it is and make a dedicated, stationary device for filament production. Here are some more detailed ideas:

- The nozzle diameter and extrusion speed don't have to be matched very precisely and possibly should even be a bit too high, to produce a filament that's initially always at least 3mm in diameter. The reasoning behind this is explained later on.

- The filament should be extruded downwards, to let it form without any "sideways" force distorting the cross-section while still soft, but cooled immediately with a fan mounted just below the nozzle, so it doesn't stretch too much (or, alternatively, expand too much, depending on the material)

- There should be a receiving spool with some rather weedy motor attached to it, strong enough to make it actually wind the filament on, but not to pull on it too much. Additionally, the motor should be resistant to damage from stalling, because it would really stall most of the time.

- Between the extruder and the spool go several pairs of brass rollers, with semi-circular grooves in their sides, heated above the glass transition temperature but below the melting temperature of the material being used, and held together with moderate force. The first pair should form a circular passage somewhat bigger than 3mm in diameter to eliminate any substantial kinks in the raw filament and the last one should give something around 2,9mm.

- The only really tricky part here is eliminating material build-up at the rollers, but that might be solved by putting an additional pair of cold, gripping rollers just below the extruder (so that the first heated pair can safely pull on the filament) and adjusting speed of the heated rollers with some filament diameter sensing feedback loop. The sensor could be made using - that's right, you guessed it! - another pair of rollers, this time, a lot smaller, free-rotating and made of PTFE, held *very* gently around the filament using a small spring, with one of them mounted on a tensioner arm attached to a precision potentiometer or a high-resolution optical encoder shaft. OK, I promise, no more rollers from now on.

- As for the auger, just heat up the whole damn thing including the tube (making a temperature-resistant tube mount and motor coupling should be a whole lot easier than trying to maintain a short thermal gradient in the auger), leave a wide hole at the side on the top of the tube and position some kind of a secondary granule-feeding mechanism just over this hole, so it doesn't come in contact with the hot tube but the granules make it into the auger. It won't be enclosed, but that's probably not a problem, as long as no one sticks anything inside. That's how they do this in some industrial machines to keep things simple.

- The secondary feed should not let the auger run empty, even at the cost of some overflow. Just add a tray below to catch the overflowing granules and put them back wherever they came from every once in a while.

- Starting the extrusion process would probably have to be done manually, to pass the filament through all the rollers correctly. Then, it could work as long as there's a steady supply of granules and some space on the spool left.

Feel free to comment, or even to try this design - I'm afraid I can't, no lathe here to make the rollers, so I'll be happy to see it evaluated, and even happier if it actually works.

(Added after a while - some random thoughts)

In a stationary machine, when weight is not a problem, the ideal auger motor could be a windshield wiper motor from an old car. They're *really* powerful (2 to 5HP - no joking, that's the force you need to swing a winshield wiper reliably, especially in a bus or truck), almost always come with an integrated worm drive and some really tough steel casing, use 12VDC and, most importantly, they can be obtained for free or close enough at most car repair workshops.

The feed hole in the tube can be supplemented with a half-tube guide attached at the lower edge, steep enough to not let the plastic melt and stick to it before getting into the auger.

Edited 1 time(s). Last edit at 01/16/2009 01:20PM by Enleth.

Annirak, do you have a source for these 300% efficient Peltier devices? With something like that, who needs a RepRap? Thermodynamics be damned!



On a more serious note, that extruder looks nice. A plug in power drill might even have enough power to mechanically melt the plastic without an external heater, just run it in reverse and let it go! It'd be too big to fit on the RepRap, but like Enleth says you could have it off to the side producing filament to feed a Reprap.

Wade

Wade Wrote:
-------------------------------------------------------
> Annirak, do you have a source for these 300%
> efficient Peltier devices? With something like
> that, who needs a RepRap? Thermodynamics be
> damned!
>
Well, that's how heat pumps work. Efficiency numbers are virtually *always* over 100%. To get a peltier device to run as high as I've stated, you need to power it with about 40% or rated voltage and have less than a -20C difference across the sides (that is, the hot side cannot be more than about 20C hotter than the cold side). However, with a situation like we've got, the temperature differential actually improves efficiency. Consider this: when there is no power applied to peltier device, and it has a temperature differential across it, does heat still flow? Doesn't that make it infinitely efficient? So putting some power through it, doesn't it make sense that it could still be over 100% efficient? After all, we're just moving heat. This is where the carnot cycle efficiency comes into play: it's the practical limit energy generation from heat differentials. The reason you can't make a perpetual motion machine with a heat pump that has 300% efficiency is that you can't recover more than 3/10 of the energy it moves at the temperature differential it creates.

For our purposes, the cold side will be about 200C, and the hot side will be about 50C. That will make for very efficient heat transfer indeed.

Well, enough of a lesson on practical thermodynamics for now. Back to the subject at hand.
>
> On a more serious note, that extruder looks nice.
> A plug in power drill might even have enough power
> to mechanically melt the plastic without an
> external heater, just run it in reverse and let it
> go! It'd be too big to fit on the RepRap, but
> like Enleth says you could have it off to the side
> producing filament to feed a Reprap.
>
> Wade

I need a granule extruder for testing a new kind of plastic which I'll be getting samples of in the next few weeks. Whether my extruder fits on a reprap or fabs the filament to run a reprap, I don't care. The point is to extrude from granules so that I can test the properties of the plastic.


Brendan

degroof Wrote:
-------------------------------------------------------
> Wasn't there an experiment like this done a while
> ago? I have a vague recollection of there being a
> problem due to a metal auger conducting heat back
> toward the granules, causing clogging.

degroof Wrote:
-------------------------------------------------------
> Wasn't there an experiment like this done a while
> ago? I have a vague recollection of there being a
> problem due to a metal auger conducting heat back
> toward the granules, causing clogging.


There was this a while back: [blog.reprap.org]

But I think it's a little different. Possibly different enough for my variant to be worthwhile.

nophead Wrote:
-------------------------------------------------------
> Perhaps a brass nozzle and melt zone, a short
> section of stainless pipe leadin to block of
> aluminium with a big heatsink attached.
>
> You need to keep the SS section short because
> there is always a problem point where the plastic
> has softened but not melted. It just expands to
> plug the tube. The thermal gradient has to be
> steep enough that the problem area is short enough
> to not cause a jam.


I still think that glass is a better from an insulation perspective than stainless, but finding stainless pipe that mates with MIP fittings is probably easier. Might even find a suitable die so I can just cut the threads myself if it's not available.

Brass is pretty close to aluminum in thermal conductivity so it might make more sense to just use brass. I think the biggest problem is the thermal conductivity of the auger. Maybe it would make sense to drill a few small holes in the solids transport section of the extruder and force cold air through it?

Thanks for the thermodynamics lesson there Annirak. Have a look here:

[en.wikipedia.org]

Your 300% value is most likely a coefficient of performance (COP), which does not refer to efficiency in the usual sense. It just compares the heat moved relative to the heat input; thermodynamic efficiency is something else again, and should be around Nophead's 30% figure. Plain old "efficiency" is somewhat undefined when you're talking heat pumps.

I was just poking fun at the perpetual motion implications of efficiencies greater than one. Sorry about that.

Heat will still flow from hot to cold even if you're peltier is off, you're just loosing power to entropy. The question is whether your peltier equipped cooler will pump more heat than a cooler without the peltier element. Unless you've got a truly powerful Peltier, it's most likely going to just slow down the rate of heat transfer.

I think that unless you're trying very hard to get temperatures below ambient, you're far better off with just a heatsink, rather than a peltier element and a heatsink. Feel free to prove me wrong though!

Incidentally, peltier elements are expensive and power hungry; I had a peltier element beer cooler in my truck in the desert, and it could only get the beer a few deg below ambient (which was 40 deg C!) over many hours of use. And it used so much power it would kill the truck battery in minutes if left on when the truck wasn't running. Not terribly useful for me.


Wade

Yes I have a peltier beer fridge and was very disappointed to find it is a lot less efficient and more noisy than a conventional fridge.

Forrest tried something more like this a long time ago: [www.google.co.uk]

Enleth,
I don't think you need such a complicated system to get the right diameter. The filament comes out bigger than the nozzle aperture due to die swell. The faster the flow, the more the pressure and more die swell. I think all we need is to have an aperture a little smaller than 3mm and adjust the pressure with feedback to keep the diameter accurate.

Good idea to let it fall vertically. If I run my extruder at the top of the z-axis it produces pretty consistent filament and it cools and coils itself neatly on the table. If we could scale that effect up we would have a very simply system. The problem I anticipate is that the height required for cooling might go up with the square or cube of the filament diameter so it might not scale.

---

[www.hydraraptor.blogspot.com]

Actually if we make our own filament 2mm is probably a better size. When using it it will require less force and more speed to extrude making the machine's extruder smaller and lighter.

It will also cool faster and coil more easily. It might be feasible to go even smaller.

Edited 1 time(s). Last edit at 01/16/2009 03:10PM by nophead.

---

[www.hydraraptor.blogspot.com]

A granule extruder does not need to be large or heavy; in fact, I have yet to see a design proposed that is either, certainly in comparison to the current filament extruder (or the experimental pinch-wheel extruders with their large motors, although I suspect the final versions will be lighter). The auger concept leads to a relatively simple design, and no granule extruder needs to have a very large hopper at all when a fixed hopper to reload from after every N layers would work fine. This would offer the added bonus of enabling one to refill the fixed hopper partway through a build, while I believe it's hard, if not impossible, to do the equivalent with filament.

Hi there, I see this topic has come to somewhat of a standstill...

I have been building a machine over the last couple of months following loosely on the reprap principles. And I decided early to endeavor to get the granule extruder right.

Today I successfully extruded POM which is an acetyl, ie a very hard plastic. Consistently and without overstressing extruder components. (previous versions resulted in nozzles splitting and motors dramatically failing).

This was achieved using a step down gear assembly from a photocopier (all smaller motors failed) and a 6mm wood auger. Haven't had the inclination to update drawings of this.

Problems of excess heat melting material in the hopper is avoided by feeding material from a vented or cooled side.

My understanding is that ABS will melt much easier and I will move onto that once I have the machine assembled with this print head in it. For now I thought I'd share a bit of development and give back to the REPRAP community

Great to hear! Could you maybe post some more details (here a picture might indeed equal a thousand words) on how you got things working?

-Geert