The extruder motor and changes needed in extruder robustness

I am in the process of incrementing the extruder design. This makes it both easier to manufacture and more robust.

But we need to make it stronger yet, and also for one design to be able to handle (at least) PCL, HDPE and PLA.

Of course, this may be the time to go for broke and switch to a granule extruder design. The principal problem there is the easy (i.e. no lathe) manufacture of a robust auger that will withstand (at least) 200 C.

The 12v Solarbotics GM3 motor works OK as a driver for the current design (especially if you lubricate its gears with silicone grease - it's supplied dry). But do we need to switch to something more meaty?

So I'm starting this thread to invite thoughts and suggestions.

Edited 1 time(s). Last edit at 03/26/2008 05:18PM by Adrian Bowyer.

---

best wishes

Adrian

[reprap.org]
[reprapltd.com]

Some random thoughts based on my experiences so far:

- In the current design, the flexible shaft and PTFE barrel appear to be the primary weak spots.
- I've got a piece of 3/8" diameter bisque alumina on order, just to see how it holds up to pressure and temperature. I have fantasies of machining a one-piece barrel out of the stuff.
- It's difficult to tighten the extruder body so that the drive screw doesn't slip but the motor can still turn it effectively.
- The less distance (or volume) between the drive mechanism and the nozzle, the easier it is to control flow.
- I, personally, would like an extruder that could take filament, granules, cut-up bits and/or wadded-up grocery bags but that's probably a tall order.
- Maybe an auger could be assembled from graduated sections of threaded rod (i.e. a stepped cone).

Should shift to a GM17 motor which has a LOT more torque and is considerably quieter.

[www.solarbotics.com]

"Of course, this may be the time to go for broke and switch to a granule extruder design."

If you can make one that doesn't weight half a ton, fine. The extruder shouldn't need NEMA 23's to fling it around the xy working plane, though. Making heavier extruders is the opposite of the direction we should be traveling in, imo.

I don't think that you can do it, though. As well, the whole issue of how you insure flow continuity on a granule extruder operating at such low extrusion rates is one that I don't think that anybody has seriously thought about.

... i tried with milling talk (soap-stone/limestone, several density-grades) as heatresistant (until 1500

The flexible drive shaft lasts less than 20 hours so it has to go.

I have shown that the PTFE barrel can be replaced with stainless steel and a modest heatsink. That makes it both mechanically robust and high temperature.

The bearings are not the problem with GM3, it's the brushes. I got through two before switching to the 12V version. It runs quieter but the motor looks the same except it has a black end cap so the jury is out on that one.

GM17 looks a lot better so I think I will try that next, that fact that it is quieter is a big plus as that is the noisiest bit of my machine.

I think the answer to the drive screw grip problem is to reduce the friction in the polymer guide. Viks oily rag mod seems to work well with PCL as well as PLA. HDPE is more slippery so less of a problem. Perhaps a PTFE liner in the channel will solve the issue.

---

[www.hydraraptor.blogspot.com]

For particularly hard plastics like PLA the small electric gear motor is underpowered. I tried them and blew up two. They overheat.

Using the recovered gearbox and a 12V motor (I have heaps of them if you want one) I've been printing PLA quite happily and the temperature of the motor only goes up a few degrees. But I've no idea of the specs of the motor!

Vik :v)

Motors: I might have to pick up one of those GM17 motors. The GM3 I've got seems very unhappy.

Polymer guide friction: I was toying with the idea of adding some sort of roller bearings. First thing that comes to mind is getting some pieces of metal tubing or standoffs and mounting them on pins or bolts driven through the extruder body.

I agree that stainless steel is the best material so far for the barrel. It has a great combination of low thermal conductivity, mechanical toughness, and corrosion resistance. It is hard to machine, but it'll last a lifetime so you only have to machine or buy it once.

For pushing the thread, we should consider pinch wheels. To provide the compression without putting pressure on the bearings, you can put magnets on each wheel, so the wheels are pulled together, but no force is applied to the bearings. The bearings would only have to handle the small thread pushing force, not the large pinch compression force.

Instead of using a gear train to send power to both wheels from one motor, you could use a motor for each wheel. This would be until we can reprap our own gear trains.

If the brush motors wear out too soon, 70$ brushless motors rated for 20,000 hours are available at:
www.anaheimautomation.com/BLWRPG11_Brushless_DC_Planetary_Gearmotors.aspx

the entire listing is at:
www.anaheimautomation.com/Brushless_DC_Motors_With_Gearboxes.aspx

Cheers,
E

We tried pinch wheels in the MkI design. See the Wiki for why that didn't work.

Vik :v)

Steve,
Have you tried passing the filament through an oily rag to reduce the friction? I found it makes a big difference. So does PTFE spray but you have to top it up occasionally.

I also think simply making the pump body out of HDPE will reduce the friction compared to ABS, PCL and PU.

---

[www.hydraraptor.blogspot.com]

re: roller bearings

check out the rolamite

Hi Vik,

Thanks for pointing me to the wiki, the downsides I found for the pinch wheel are:

"My Version 1 RepRap polymorph extruder head has the great virtue that it works, but it also has some shortcomings. The main ones are that it is quite expensive to make, that it has a lot of moving parts, and that some of the components are rather hard to obtain."

Them's unfortunate shortcomings, but it did say that it worked. So if the drive screw extruder doesn't last long enough or has trouble with hard materials, it's nice to know there is an alternative even though it's expensive and complicated.

By the way, the link "Version 1 RepRap polymorph extruder head" goes to a page:
[reprap.org]

which is not found. So I changed the link to point to:
[staff.bath.ac.uk]

Cheers,
Enrique

"I don't think that you can do it, though. As well, the whole issue of how you insure flow continuity on a granule extruder operating at such low extrusion rates is one that I don't think that anybody has seriously thought about."

We have an existence proof: those Australian students who did it for Fab@home.

"Polymer guide friction: I was toying with the idea of adding some sort of roller bearings. First thing that comes to mind is getting some pieces of metal tubing or standoffs and mounting them on pins or bolts driven through the extruder body."

One of my earliest experiments was to make a recirculating roller track: the rollers rolled down with the filament then circulated up the back in a channel. It kept jamming and kept working anyway, so I deduced that the rollers weren't needed. The rollers were cut lengths of the filament, incidentally. We could revivify this idea.

The pinch wheels did not work that well - they kept slipping, almost no matter how hard I pinched. Vik's thread drive is brilliant because it bites into the filament and applies a steady positive force along its entire length.

Can you get PTFE sheet about 0.5mm thick? A rectangle of that bent into a half-round at the back of the extruder might fix the friction problem for good.

---

best wishes

Adrian

[reprap.org]
[reprapltd.com]

My extruder works pretty well using polished epoxy metal composite with some lubrication. I think HDPE would work even better. If we had to go to PTFE I was thinking of drilling a rod and then slicing it in half lengthwise to get a U shaped channel.

---

[www.hydraraptor.blogspot.com]

Adrian,

Thanks for pointing me to the extruder from the Australian group. Their report at:
[fabathome.org]

is excellent. Starting in section 3.5.3.6 Screw Extruder Shaft on page 66, they describe their granule extruder. From the description they use a stainless steel screw extruder shaft driven directly from a stepper motor. Since stainless steel is durable and there are no gears or brushes to wear out, this extruder will run for a long, long time.

> I've got a piece of 3/8" diameter bisque alumina on order

Just got notification that the package is "out for delivery". I should have it in time for the weekend. Let's see how many tool bits I can break on it...


> Can you get PTFE sheet about 0.5mm thick?

Would fusing a bunch of strips of PTFE tape together be viable?

"We have an existence proof: those Australian students who did it for Fab@home."

Um, Adriaan, they're extruding packing glue as best as I can see. As well, they have a big, clunky NEMA stepper driving it which means that the extruder assembly is going to weigh the better part of a kilogram.

You might also want to take a look at their accuracy measurements and pause for thought.

I must commend them, though, they write a very useful report on what they did. It is very complete and well done.

If a granule extruder ends up too heavy or doesnt work at slow speed you could use it to create the filament?

On a moving table machine a heavy head only affects the z-axis which moves slowly anyway.

---

[www.hydraraptor.blogspot.com]

Darwin ain't a moving table machine.

ah, but future versions could be...

I'll try with a bag of sugar on the head and see what effect it has I think the Darwin frame will take that level of load but experimentation will be the proof.

---

Ian
[www.bitsfrombytes.com]

Come to think of it my machine throws 9.2Kg (the mass of the table and upper axis) around at 16mm/s with 5V 1A NEMA 23 motors so a bag of sugar should be a piece of cake for this size of motor. Its maximum load capacity is 125Kg!

---

[www.hydraraptor.blogspot.com]

My concern is not the motors as you state they will breeze it, it's the frame rigidity especially with a slowly change mass will this lead to a gradual change in Z?

OK quick test done with a 1.5kg bag of flour (which weighs more 1 tonnes of feather or 1 tonne of steel? ) maybe a defection of 0.1mm but probably less that was the maximum error over five readings and there will be measurement error so i would say a 1kg head (or more) will be no problem.

---

Ian
[www.bitsfrombytes.com]

Hi,

I think it might be wise to go for a modular approach:
1. Make a system to create thread from granulate (I read Adrian's article about rolling threads by hand, but I guess this is limited to the lower-temp 'friendly plastics')
2. Create an extruder head that uses this thread
3. Work on compatibility of these systems

This way the thread creation can be improved independently of the current extruder design. The efforts on the extruder design can be focussed on making it more robust. This could reduce complexity through modularity. Moreover, those that prefer to just buy 3mm thread can just skip the granulate system.

I'd prefer the granulate system to include a process to recycle all kinds of waste plastics that we generate around the house. This involves creation of granulate (would a hacksaw -> blender -> granulate2thread -> RepRap). But I guess anything blends ( [www.willitblend.com] ) It would be a personal goal to give waste another useful life. Besides that, you could reuse test objects over and over and refine the system without any net-consumption... that would decrease cost and also increase the speed of development because it's more worthwhile to do this!

Perhaps in the meantime a feasibility study should be done on how to incorporate a system as illustrated by Nixon and Tan (the Austrialian group):
[fabathome.org]
To what degree is this fab-able in a RepRap (it needs many higher-melting point materials)?
I really thing this should be a side-track project. Not to say that these aren't extremely important, but we need to have both a stable version and beta development version(s). E.g. if I can't create a proper RepRap because it breaks down all the time, I will not be able to develop my own improvements on different aspects/parts.

The way I see it, there's a problem inherent in the way we wish to achieve self-replication:
1. We want to fab most of the extruder
2. The process of heated extrusion requires at least some parts not to be extruded in the first place.

It reminds me of articles about the torus/tokamak type fusion reactors. There are currently no materials that stay stable at the temperatures needed for fusion of the plasmas. Current efforts (as far as I know) are directed at creating a dynamic equilibrium that requires an acceptable amount of maintenance (down-time trade off).

I don't think we should go to 'too great lengths' to push the envelope on minimizing the amount of non-fabbed components. The only constraint that is reasonable to me is:
1. component has to be very very simple
2. parts need to be available anywhere.

One small recommendation on inertia of the axes:
Why not remove the PCB from the extruder head (X carriage) and XY carriage) and place it more remotely (on the frame) and wire it? A moderate length of ribbon cable will not weigh as much as those (3?) PCBs, right? It would reduce the stepper requirements.

--
Erik de Bruijn
[blog.erikdebruijn.nl]

Thanks Ian for testing Darwin.

The Australian report does not have the mass of all the components, but they do have detailed descriptions, so the rough component masses follow:

Deposition stepper: 110g; from Deposition Tool Stepper Motor Specification Sheets in Appendix H on page 161 of the pdf
Screw extruder: 31g; from drawing in section 3.5.3.6 Screw Extruder Shaft on page 67, pi / 4 * ( 1.5 * 1.5 * 1.5 + 6 * 0.5 * 0.5 ) cm3 * 8 g / cm3
Coupling and screws: 100g rough heavy case;
Barrel and structure: 100g rough heavy case;
Granules: 100g rough heavy case; 10 hours of granules for overnight operation at a deposition rate of 10 cc / hour and a density close to water, 10 hours * 10 cc / hour * 1 g / cc
Stuff I forgot: 100g rough heavy case;

Total: 541 g, roughly 0.5 kg

Since Ian showed Darwin can take a kilo, there is 0.5 kg leftover which could be used to add an extra 50 hours of granules for a weekend of unattended operation.

The highest melting temperature used was 125 C, and they highest tried was 145 C. The extruder could handle 145 C fine, they did not use it in operation because the viscosity was too low. This is all from their experiments with Technomelt Supra 100 in section 3.4 Material Selection on page 45 of the document.

They were very conservative in their temperature tests. They did not come close to melting or degrading components and since everything is made of metal, the extruder should be at least able to handle HDPE. The only thermal limit I can see is that the stainless steel shaft will transmit some heat to the stepper motor. It will not transfer much because stainless steel has very low thermal conductance for a metal, but it still has more than cement or ceramic. I speculate that the current design will go to a few hundred Celcius before the motor gets too hot and with a cement or ceramic screw extruder or a very long metal shaft, it could go to several hundred Celcius.

Erik wrote:
"Why not remove the PCB from the extruder head (X carriage) and XY carriage) and place it more remotely (on the frame) and wire it? A moderate length of ribbon cable will not weigh as much as those (3?) PCBs, right? It would reduce the stepper requirements."

I agree with Erik that the boards should be moved from the extruder head to the frame. For his reason of reducing inertia and also because the boards would be away from the extruder heat, which even if they can survive for a while, will certainly reduce their longevity.

Cheers,
Enrique

Enrique Wrote:
-------------------------------------------------------

> The highest melting temperature used was 125 C,
> and they highest tried was 145 C. The extruder
> could handle 145 C fine, they did not use it in
> operation because the viscosity was too low. This
> is all from their experiments with Technomelt
> Supra 100 in section 3.4 Material Selection on
> page 45 of the document.

I believe the tests in section 3.4 were all mock-ups that did not actually involve their extruder. Although I haven't finished the document (I'm on page 72), if section 3.4 is the only place where they mention temperatures of 145

Enrique Wrote---

"The only thermal limit I can see is that the stainless steel shaft will transmit some heat to the stepper motor. It will not transfer much because stainless steel has very low thermal conductance for a metal, but it still has more than cement or ceramic. I speculate that the current design will go to a few hundred Celcius before the motor gets too hot and with a cement or ceramic screw extruder or a very long metal shaft, it could go to several hundred Celcius"
------

I think that if one were to use a simple 1:1 gear exchange from the motor to the drive shaft, the amount of thermal energy transferred to the motor would be reduced drastically.

---

Jay

Jay wrote:
"I think that if one were to use a simple 1:1 gear exchange from the motor to the drive shaft, the amount of thermal energy transferred to the motor would be reduced drastically."

Quite true, it would reduce thermal transfer and raise the operating temperature limit. A downside of gears is that they wear out and add backlash.

Your implicit point of adding a thermal break between screw and motor is a great one and another way of doing it is by using a cement or ceramic tube for the coupling between screw and shaft and fill it with furnace cement or somehow cast the cement in place around both shafts or somehow reprap a cement coupling. Furnace cement is 5$ / pint and is supposed to stand 1650 C, according to the description at:
[doitbest.com]

Cheers,
E

Furnace cement also needs to be fired to get any form of structural integrity.

Vik :v)