Fiberglass thread extruder

Among the many ideas I have already thrown out, and that other people have posted, I still think that the best second non-conductive material to extrude is fiberglass. ABS plastic is very good, but not as dimensionally stable as metal, and has a much higher coefficient of thermal expansion. It also is prone to creep from stress over a long period. But adding fiberglass to the material as it is deposited will help with all of these things, plus give the pieces even greater strength.

The extruder will have to be more carefully constructed because the thread will easily bend and jam, it should still be possible to make a pinch roller system with rubber or silicone rollers to push the thread out the opening. To cut off one section after laying it down a scissors type device would work best, actuated by a small servo motor. If the previous layer of ABS is still soft, it might be possible to press the thread against it to tack down each end. Or the mouth of the fiberglass extruder would have a hot surface to remelt the ABS, stick the fiberglass in quickly, let it re-harden before moving the head, then stop at the other end and repeat the process to tack down the other end. Or it might be best to extrude a little bit of ABS along with the fiberglass so that it immediately attaches to the plastic and is not pushed out of the way when the next layer of plastic is extruded.

Fiberglass should be very easy to lay down in X and Y dimensions, but I have no idea at this point in how to work it in the Z dimension and layers are being built up. I think the best way at this point to get three dimensional strength from fiberglass reinforced ABS is to build a cap or side piece that is rotated 90 degrees and laying down on the bed, then glued on later. Like the plastic itself, the fiberglass should be laid down in different directions for each layer to give greater strength and dimensional stability.

I have made a quick Google search for fiberglass thread, cord, yarn, string, and all I could find was sewing thread that is PTFE coated. This will not work well because the PTFE (Teflon) will not bond at all with the ABS, and the thread will just pull through the body of the plastic parts without giving them any extra strength.

If we are going to make a very large fraction of a reprap out of RP plastic parts, strengthening them and keeping them from bowing or bending under pressure will become critical. ABS is already one of the better plastics to use for structural strength, but I don't think that long structural pieces of it will remain straight for years. I have seen even fairly large diameter ABS pipes bow after awhile.


Mike

---

Team Open Air
Blog Team Open Air
rocket scientists think LIGHTYEARS outside the box!

Hey Mike,

I'm glad to hear that you're thinking along these lines too! Indeed, you seem to have come to almost exactly the same conclusions as team SpoolHead, which is very reassuring, because it means we're all probably on the right track.

After doing some research of what's on the market for pushing wires these days, it looks like a motor-driven wire drive would need to be two driven wheels (instead of one and an idler) of equal diameter, turning at the same rate. Otherwise, the wire will tend to curl up around the drive wheel, I think.

This is somewhat complicated to pursue (you'd probably have to use gears, belts, pulleys, or two motors to keep the wheels in sync), so for now we're going with a solenoid-driven mechanical pencil to feed the wires. It's a really simple, cheap solution, but the downside is that it can only take specific wire diameters... fibreglass might be much finer than we can handle (unless it's braided or something), but steel wire is also a good possibility that we're going to explore.

Like you suggest, we're also planning to use a heated tip to soften the surface of the ABS (or in our case, PLA) before attaching the wires. Whether or not this will work well is the big question that we have yet to answer.

We were hoping to do something like scissors, but that inevitably got in the way of the heated tip. So instead we're building a solenoid-driven rotating cutter, as seen on the wiki page. This might be a bit tricky to construct for people without good machining facilities, so we're open to better options.

Our project was initially targeting X,Y, and Z wire fabrication; we had a scheme worked out that would let us print wires in the Z direction (only one at a time though, of course, unlike in X and Y where you can have multiple wires on multiple paths). But this would have required the print head to have its own independent Z axis, to lift up while the plastics were printing (can't cut the wire!), or some other degrees of freedom to achieve a similar goal (like reverse-feeding the wire). It would also take some additional software considerations to keep the wire getting tangled around the plastic extruder - you'd never be able to trace circles around the wire, so you'd have to go back and forth. This seemed like too much work to get done before our project deadline, and so we abandoned it. A more promising option would probably be VDX's laser-wire-microwelding plan.

... i made some tests with diode-lasers and a 50Watt-fiber-laser - the fiber-laser has a spotsize of 20 to 5 micrometers and can cut/evaporate or melt/fuse the wire without problems, whereas diodelasers with 5 Watts of power have a much worser focussing and can get best spotsizes of 100 microns or even bigger.

So with a diodelaser you can heat and evaporate the plastic, but you can't cut the wire ... but you can melt a small blob of soldering paste - so you can solder the wire to a pad ...

Another idea is to combine more of the diodes with a bigger focussing lens, so you'll receive a spot around 0.3 to 0.5mm diameter, with 7 (or even more) lasers focussing on the same spot ... maybe with more than 20Watts you can melt the wire too

I'm estimating the costs of one diode-laser-head with drivers and optics between 300 and 600 Euros and a bundle-head with 7 diodes (one in the middle, 6 around as tightest array with 15mm diameter of the optical bundle) between 1000 and 1600 Euros - a comercial turnkey-10Watt-module is around 1000 Euros ...

The price for comercial available fibre-lasers with 20 to 50 Watts last time i checked was above 9000 Euros - the system i'm actually using will cost something around 30000 Euros, so not really suitable for DIY ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

ybayless,
If you look inside the box of a MIG or wire feed welder, you will find exactly the system you are talking about! They push 6 feet (2 meters) of wire down a narrow tube inside the welding cable AFTER it has gone through their pinch roller system! And they use only one driven wheel and one idler wheel. They do have channels cut in the idler wheel to max the diameter of the wire going out.

This works great for steel wire, but it is too difficult to push soft aluminum that far "up hill" through such a long thin pipe. But for your application, with the wire exiting just beyond the orifice that guides the wire away from the pinch rollers, you should have so problem.

However, that is where fiberglass differs from metal wire. Even stander cord and rope would push through the pinch rollers and out the opening with greater ease than fiberglass thread. On idea I have is to a liquid extruder (plunger and syringe, positive displacement pump, auger, etc) apply a very small amount of ABS cement to the fiberglass, then have it extend (and stick) onto a Teflon coated roller of 1/4 circle of spring steel. This would have the fiberglass extending out of the extruder parallel with the plastic already laid down. Pressing the roller or spring to the plastic and moving back slowly, while feeding with the pinch roller will _hopefully_ leave a trail of ABS cement and fiberglass behind. Once you reach the end of the line the scissors would cut the fiberglass between the pinch rollers and the bottom part, pulling back a little further would tamp done the loose end. You would need to rotate the head, or at least the lay-down roller or spring steel so that it was always following the fiberglass thread as it came out, no matter what direction the X-Y axes are going. You could also but a pre-heater in front to soften the plastic before the fiberglass and cement are laid down.


For your wire process, I would consider using the same rotating lay-down roller/spring steel to press your wires more firmly into the plastic. In your case, pre-heating the plastic may not be necessary. I think you would get the best results from pass a current through the wire from the pinch roller (make the idler wheel metal and use a brush of connect through the bearings) to the lay-down roller. Once the metal is hotter than the melting point of the plastic, it will cut its own way into the body of the plastic, only to cool quickly and come trapped and bonded into the plastic. Should work very well!

Mike

---

Team Open Air
Blog Team Open Air
rocket scientists think LIGHTYEARS outside the box!

Victor,
fascinating idea, using Lasers or power LEDs to cut, solder, and/or fuse the wires. What type of wire are you working with right now? Steel/iron wire has a moderate thermal conductivity, so it pulls heat away from the point you are trying to cut rather slowly. But steel has a very high melting point. Copper as a melting point a little of half that of steel, but it has such a high thermal conductivity that it will draw the heat away rapidly. The thinner the wire, the more slowly heat will be drawn away and the easier it will be to cut. Aluminum would be easier than copper because it is not _quite_ as thermally conductive, and has a much lower melting point. But once again, thin wire would work best. Try getting aluminum wire from a welding supply house. There it will be very thin, and not covered with insulation.

Mike

---

Team Open Air
Blog Team Open Air
rocket scientists think LIGHTYEARS outside the box!

Hi Mike,

... when working with wires i had to fix platinum wirs with 10 microns and 1 micron diameter on gold pads, what i made with eutectic gold-solder (80%gold, 20%tin) mixed with golddust and dexpanthenol as paste - so more brazing than fusing.

I made other tests with melting/cutting steel-sheets of 0.3mm thicknes, what's no problem with 40Watts@1070nm in a spot of 30 microns ... a diodelaser with some ten Watts power in a spot around 200 microns diameter wouldn't have any effect here

For experimenting with wire i have some steel-fibres with 25, 35 and 50 microns diameter and copper-wire with 70 or 100 microns - i'll try this when i have some diodelasers with 4Watts single or 30Watts in a bundle working ...

But diode-lasers, even with 4 Watts only should be berfect for fusing coloured plastic-wire on a surface, so i'll search some dark plastic-fibres and make some tests with drawing lines and dipping single drops ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

The thermal conductivity of the wire is proportional to its cross-sectional area. If the wire is very thin, its thermal conductivity probably becomes negligible.

The conductivity of stainless steel is about 20 W/mK and copper is about 390. They'll have equivalent conductivity when the copper's cross section is about 20 times smaller than the steel, so the diameter is a little less than a quarter the size.

That is, if it will work for 250 micron stainless, it will work for 100 micron copper. Except that the copper will actually be much easier, because of its lower melting point.

... another essential point is the thermal conductivity of the surface you want fuse the wire to - it's the same material as the wire, but with much more thermal capacity.

When the molten segment (only the tip or a continuous thread already fused to the surface and frawn further) touches the surface, hou have to deal with the much bigger thermal flow and cooling effects in the connecting area too ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

Right, I hadn't considered that!

One thing you might try looking for is spools of fiberglass thread for chop-guns. Fairly cheap stuff but it might be a bit too heavy for desktop application.
I have been mulling over fiberglass injection or composite filaments in my head for a while now; my long-term goal is to build a garage-size printer that can print an entire monocoque boat hull. I've considered the possibility of a powder-style build chamber where the "powder" is shredded glass fiber and the print head lays either plastic or resin, but in all practicality it would be much easier if one could just print the whole thing in plastic in order to avoid cure times with resin and also avoid moving tons of stuff on a lowering print bed.
On a large scale with a larger nozzle, it would be much easier, and you could probably feed the fiberglass thread into the extruder with no problem and just let the plastic flow drag the fiberglass rope out. Smaller scales, much more difficult.
The best concept I can think of at the moment is some sort of "venturi-style" nozzle where the thread feeds into the "suction" part of the nozzle, and is pushed along by friction from a wheel. The plastic and fiberglass meet as they exit the hot head. If it ever bound up, though, it would make a mess of your print. It would also be problematic (and a waste of glass fiber) if the head had to reposition itself across the print area. In this case you'd also need some sort of blade to cut the fiber.
The easiest way would just be to make a composite filament to begin with, but the small shopped fibers required to make that happen wouldn't give it much tensile strength.

I think chopped mat would make things much easier as its used regularly in plastic injection to improve mechanical strength. 30% chopped mat in nylon is used regularly for automotive engine parts.

A cheap filler material in plastic would also help to reduce the cost which is a limiting factor in printing large objects.

Perhaps printing plastic, then powder in altenative layers may be an option. You could experiment with different amounts of powder and in different patterns if inter layer adhesion was an issue.

Something I also wonder about, especially regarding the fiberglass-chop-impregnated plastic filament...
Just what happens, mechanically, between layers of plastic when they are laid down? Does the plastic actually heat enough between the new layer and old layer to form the equivalent of a chemical bond/actual fusion, or is it just a really tight mechanical bond?
If one were to use chop-impregnated plastic, and real layer fusion could occur, would the layers potentially melt enough for the chop fragments to cross layers a bit and form some tensile strength between those layers?

The general concensus here is that the layers are a mechanical bond "hot forge" rather than a melted fusion, but I dont think anyone knows for sure. This would be logical as the deposition speeds some people run at probably wouldn't give the lower layer time to melt. Using this type of joint doesn't mean that it is weaker than the parent material.

I've done hot forge operations in metal and when a sample is sliced up, polished and etched you can see the distinct separation line. The joint if done correctly has a tensile strength equal to the parent material (Its using the same interatomic forces that hold the parent metal together). I'm not sure whether this parting line would be visible in a plastic part. It might be interesting to produce a sample with two coloured layers and slice it up to see how it looks under a microscope.

This is slightly off topic, but I was thinking slightly beyond the Rep Rap compass in terms of being able to move from mere rapid prototyping to actual micro-manufacture. The difference, it would seem, would be the durability of the materials used, their rigidity versus their plascticity, and the accuracy with with they might be printed. Regarding the durability and the rigidity, the thought came to mind that within the compass of the Rep Rap project, that one could substitute certain rigid composites for metal cast bolts, nuts, etc. So I was thinking in terms of either using carbon fibers in a similar fashion to fiberglass, or perhaps even basalt fiber. One of the hottest topics right now is carbon fiber weaving, which is actually in use by Lexus for building some prototype vehicle they have designed. Rep Rap could certainly print its own "loom" and subsequently weave any composite filament to be later epoxy filled, but it would be even better to design a nozzle that could deposit, cut, and epoxy the filament in question. I think from looking at some of the experiments copper wire embedding into abs here on Rep Rap's page, that the most difficult part would be the cutting of the filament. I wish there were a better way to do composite plastics via 3d devices.

What if you had meshing gears that could snip the fiber. Then if the fiber was being threaded through some little notch or hole, this could be selectively moved to bring the fiber in between the gear teeth which would snip it. Or else use a linear oscillating motor to move some teeth, like on a hair-trimmer?

How do these people do it?

[machinedesign.com]

[www.performancematerials.com]

Edited 3 time(s). Last edit at 02/17/2013 02:58AM by sanman.

So funny, I thought I was being inventive thinking along these lines to make an extruder that would do this, and here is a whole forum dedicated to it.