Molding and PLA extrusion
I tried vaseline as an alternative release agent and the whole molding/casting process worked a lot better. The plastic removes cleanly and there is no discolouration. The method involved heating the vaseline to make it a thin liquid. It was then painted onto the surface. Heating the plaster of paris absorbs some of the vaseline, so another application was added to ensure some at the surface.

The plastic is very viscous when melted and extremely hard when set. This leads to two concerns I currently have. One is that it may not flow through a narrow diameter extrusion nozzle. The other is that it is so hard that the screw-based feed mechanism may not be able to reliably grip onto rods of PLA (unlike EVA, which is quite soft so easy to feed).

I've had a slightly different idea for extruding PLA. It probably won't be useful in a reprap situation but it may be useful as an interim method in the repstrap. The idea is to have a narrow tube (15mm copper tubing) that opens in a larger diameter tube above it, which serves as a hopper. A nozzle would be attached to the 15mm copper tubing. A 12mm auger bit running in reverse goes through the hopper and down into the tube. The whole thing is heated via nichrome wire to melt PLA. Because the PLA is quite viscous, rotating the auger downwards should push the melted plastic downwards, forcing it out of the nozzle. The advantage of this approach is that you can feed almost any form of PLA into the hopper and it will be extruded. This means you don't need to prepare rods and you can use granules, chips or powder. It also makes recycling easy because you just break up the original plastic part and drop it in the hopper.
- posted by Simon McAuliffe @ 10:22

 
PLA fidelity
I thought I'd try a little experiment to see how well PLA will take on small detailed shapes.

This evening I cast a gear into a plaster of paris mold of a spur gear. I put PLA granules into the mold. I heated everything to a little over 200°C and waited. The PLA melted and formed an amorphous blob but it wouldn't reach into the teeth of the gear. A little disappointing, so I removed it from the heat and left it to cool. Interestingly in the process of cooling, the plastic was drawn into the finest details of the gear. Something about the cooling process really helped. I don't know what this means for extrusion or if it helps at all.

The result is extremely accurate and detailed and you can see very fine details of the involute tooth shape (not very apparent from the photo). There were a couple of little bubbles but they were well away from the centre and the teeth and it seems very hard and strong. It meshes very nicely with matching gears with no slop.

I coated the plaster of paris with a cooking oil spray to act as a release agent for the mold, but it failed pretty abismally. It was too hot for the oil I guess, and it led to the brown discolouration as the oil denatured (along with an unpleasant smell that filled the house). The plaster didn't release as it should and I had to break it off and scrape and brush the remaining plaster off.

The mold making was probably fairly critical to the detailed result, but we're not really interested in molding parts right now, so I won't go into detail.
- posted by Simon McAuliffe @ 10:00

 
PLA
How exciting. After a mix-up with NZ Post, a sample of 210g of PLA arrived for me to play with today. Thanks Vik!

I thought I'd check out the behaviour of the PLA. It turns out it melts at a much higher temperature than I expected. I expected it would melt between 130°C and 180°C. Here's what I found:

155°C starts to become noticably glassy, but granules are still quite hard.

170°C starts to deform

180°C granules merge together very slowly (still too viscous to flow in any useful way)

185°C slowly flattens to a single continuous mixture, but still very viscous. Quite meldable at this stage though.

200°C still quite viscous

230°C still quite viscous. I'm not sure it will extrude easily at <1mm thicknesses.

At this point there was a little bit of a smell, so I thought I wouldn't push it too much futher. Beyond this point becomes less useful in any case.

As far as mechanical properties, this stuff is very tough. Should certainly do a good job for a lot of parts. It's a bit harder than I expected, but also more brittle. I think those are usually traded off against each other, so no big surprise. If we can make our own, we may be able to tailor that to our taste. The brittleness doesn't seem like a problem, you have to apply quite a lot of force to crack it.

I also discovered it adheres really well to things. The small sample I melted in a ceramic container wasn't coming out and reheating only gets some of it out. I thought I'd try breaking the container around it on the offchance it would come free. It's well and truly bonded however, so that sample is ruined (getting the fragments of ceramic out will be next to impossible now). Something not to try again...

Good adhesion is a promising sign really since it means objects we create will probably be strong (lots of extruded rows need to bond together).
- posted by Simon McAuliffe @ 08:56

 
Gear cutting
I thought I should note down an idea I had a few days ago before I forget. The turntable platform will have precise control of it's angle, so it would make for an ideal gear cutting tool. If a blank made of plastic or metal was mounted on a support above the centre of the turntable, a simple cutting tool could be used to produce custom gears. The cutting tool could just be on an adjustable up/down pivot and hand operated. The computer would set the angle, then you'd just swing the cutter up and down and proceed to the next tooth.

This might be particularly useful, for example, to produce bevel gears for the extruder and using only simple radial gears to construct the turntable.

Custom gear cutting would also make it plausible to construct a gear that can mesh with a worm gear that is made from threaded steel rod. It would need to be made of steel and have small teeth, but that might be one easy way to get worm gearing. Perhaps the larger pitch coach screws could even be used (?).

Of course, this could all potentially be automated in the future too, but I'm just thinking of construction of a basic machine for now.
- posted by Simon McAuliffe @ 07:54

 
Bushing support
The support nut for the bushing is created by drilling a 6mm hole through the middle of the pipe. The hole is widened to snugly accomodate an M6 nut. Then an M6x20 bolt is put through the hole and fastened with the nut.

The purpose of this design is to hold the bushing in an accurate position. It can be removed (which you need to do if you want to take the extruder assembly off), and then put back in exactly the same place without recalibrating anything (hopefully). The top point of the nut should be marked so that you can ensure you put it on in the same orientation when re-assembling.

The groove obviously only needs to be cut in a rectangular shape with a straight metal file because the sides of the hex nut are parallel (this means it slides up or down to disassemble). The depth isn't too important. It just needs to be deep enough to key the nut nice and accurately without any chance of slipping, but it can be quite deep too because the top and bottom of the tubing will support it even if you cut all the way through the copper. It should also only be on the inner side of the tubing, otherwise it can't be tightened. The filing should be done slowly and checked frequently so that there is no slop in the nut positioning.

Any excess bolt length can be cut/ground to sit roughly flush with the end of the bolt. The exact position of this is not critical.

The next step is to attach a larger M10 nut to the M6 nut, which will be the bushing housing. To begin with, you only want a temporary joint (rather than brazing them together) because you might need to adjust their relative positions to correct for manufacturing errors. I am just using a hot glue gun because the "glue" can be peeled off cleanly afterwards so it won't interfere with the brazing. Later on, once the final position is determined and glued in place the nuts can be removed and then clamped together. The glue can be peeled off and then the nuts cleaned and brazed.
- posted by Simon McAuliffe @ 07:30

 
Extruder feeder screws
I was picking up a few M6 nuts and bolts for the repstrap bushing connector and grabbed several of these at the same time.

They have a high pitch thread which means they can be used on quite a high angle while still resulting in the thread running more or less horizontally. That means a fairly short length such as these should be quite workable in the scheme outlined earlier.

These M8x75 galvanised coach screws are readily available from even a small hardware store.

With a little machining these should turn into quite good parts. It remains to be seen if they will bind with PLA/PHB rods sufficiently well to draw them downwards strongly enough.
- posted by Simon McAuliffe @ 10:05

 
Bushings
To keep things simple my concept of the head assembly uses threaded rod and nuts (see overview diagram at the start of the project). However this means the threaded rod must rotate smoothly. For simplicity's sake, I plan to use high density plastic bushings. This requires that

a. the plastic is not affected by oil; and
b. that the threaded rod is very smooth at the end.

To check the practicality of (b) I did a little experimenting today. Using just an electric drill I wanted to produce a high polish on the end of a threaded rod. This is based on the afghan lathe technique with minor variations.

The first step was to remove the threading. This was just done by lathing with a metal file held on a support (using the drill as a lathe). The shaft is then smoothed to a high polish by using sandpaper. Starting with a moderately coarse paper and reducing a little at a time down to a very fine paper. I finished with a #2000 paper, which produces a very smooth surface.

The end of the shaft can be cut next. It can be polished and the edge rounded too, to prevent any binding in the bushing. This is done in the lathe again.

The idea now (which I haven't tried yet) is to drill a hole in the copper pipe. On the inner side of the pipe the cut will be widened so that a nut can be slightly inset into the groove to prevent it from turning. I really need a deeper nut than a regular nut because the bolt will come in from the other side of the copper pipe and go into the nut, but I want some free space inside the nut cavity, so I'll be brazing two nuts together. Now you can bolt the nuts onto the copper pipe. Fit the polished end of the shaft into the gap, and ensure everything is in exactly the correct position. Fill the rest of the nut cavity with melted thermoplastic and let it set. Then just turn the shaft and it should come free. Oil it and you have a perfectly snug fitting bushing that will rotate without any slop. The main advantage of this approach is that any construction errors will be corrected for by the plastic bushing being set in the correct position -- provided the errors aren't so large that the shaft doesn't fit inside the nut cavity at all. In that case, you can braze your two nuts together offset slightly from each other to correct the majority of the error and use the plastic to fine-tune the final position.

That's the idea anyway. I considered literally more than 50 different ways of connecting the shaft to the pipe, but when this extremely simple idea eventually came to me (simple ideas are sometimes not easy to come by), it seemed like a good way to proceed.

If it doesn't turn out to work as well as I hope, I can always fall back to using bearings. I was in the scrap yard shop the other day and picked up a pair of old roller skates (not roller blades) for $2. These are great! Inside each of the wheels there are two bearings, so for a pair of skates, that's 8 wheels or 16 bearings. Great value at around 12c each.

So as a backup I thought I'd try to cut a slot into the shaft. I just did this by lathing with a hacksaw, which was held square by supports . A C-washer can be put into the slot as shown.

So with the C-washer clipped on, it fits nicely into the bearing (of course I actually cheated a little here because I knew the bearing size in advance). If I mount the bearing on the copper pipe, this will make an even better and smoother-running shaft. I haven't worked out a good way of mounting it yet, but it shouldn't be too hard. The general idea is to use three screws to hold the bearing in place so its exact position can be adjusted, or perhaps use thermoplastic again to hold it in just the right position. If I use bearings, the polish obviously isn't necessary, but it just looks so pretty :) Still I'd prefer the simplicity of the bushing. Perhaps not everybody can find old roller skates these days.

This was all just a test. For the final bushing, I think a smaller diameter would be better. For a bushing, the larger the diameter, the greater the friction. So you really want the smoothed shaft end to be as small as possible while still being sufficiently strong that it won't bend or wobble. That will depend on the eventual head assembly's weight which I don't know, but at least a bit smaller than pictured would make sense.

Something not obvious from these pictures is the use of liberal amounts of light lubricant during the cutting and polishing. I just cleaned it off before taking the photos.
- posted by Simon McAuliffe @ 04:17

 
Material feeder
While I remember, I was thinking about a possible enhancement to one of Vik's ideas for using a threaded rod (or bolt) to draw feedstock into the extruder.

The main purpose of this is to eliminate the need for bending the feedstock rod. In fact by slanting the thread by the same angle as its pitch, then the force drawing the feedstock downwards may be more efficiently applied (as in the diagram, the threads end up being completely horizontal). To get more angle, you could slant the threaded rod at twice the thread pitch and still be no worse off than having it completely vertical.

By also facing the low ends of the threads towards each other, it leaves more space for gearing to drive the threads. It will probably require bevel gears to make this work nicely. The downside is that this will also produce a sideways force, but the rod needs to be passed down a tube in any case so this will keep the feedstock moving vertically only. At the same time, with the thread in the right direction, there is also a sideways force in the opposite direction. [Do they come close to cancelling out?]

The points in the diagram marked with a * indicate a point where a bearing or bushing is required.
- posted by Simon McAuliffe @ 07:02

 
Levelling the platform
A few days ago I was taking apart a broken VCR for parts (I can get these for a couple of dollars at the scrap yard and they contain some useful parts like gearing etc.). One interesting idea they use for calibrating the spin axis of the head is to use three small adjustment screws.

Three obviously seems the best number. Two doesn't give it stability or the ability to adjust in one direction and four would require coordinated adjustment without any apparent benefit, and is likely to just damage the screws.

So what I'm planning on doing is attaching a triangle shaped piece of metal to the top of my turntable shaft. The corners of the triangle will be attached to the non-stick pizza tray (turntable) by screws. Small adjustments in the screws will allow me to get a nicely flat spinning turntable.

Similarly, the extruder platform will need some adjustments to ensure it is perfectly parallel to the turntable surface. I will do something similar to connect the top half of the frame to the bottom half so that it is also adjustable. To keep things simple, I will probably have to use 4 adjustable connectors.
- posted by Simon McAuliffe @ 10:20

 
Building a frame
I've been spending my recent time writing communications infrastructure for a PIC-based network of devices. That's gone pretty well and amongst other things, we now have a working distributed controller for a motor. That means I really need to make some progress building this contraption so I have a use for the new motor controllers.

I have recently realised that Vik's contributions to the reprap project parallel some of my intentions, and involve building a working prototyping machine to produce reprap parts. That makes this project rather superfluous, especially now that some of the experimental ideas (such as threaded rods and nuts for producing movement) are being used there too. That makes it less useful for me to experiment with the same ideas, but I'm going to continue anyway as there are still some slightly different ideas, and I still want a machine for myself.

With that in mind, this evening I brazed together the top part of the frame for my concept:



It was my first attempt at brazing and it turned out to be easier than I expected (certainly easier than welding). I was using an oxy-acetylene welder at a fairly low heat, but a small/cheap gas torch should work just as well.

The trick is to evenly heat the surrounding pipe until it has a bit of a red glow. This is still well below the melting point of the copper. Then just touch the brazing rod onto the surface and it flows quite easily and nicely into the gaps. The result seems really quite strong.

I used duct tape to hold the parts in place while I welded. Copper conducts heat very quickly, so I put a wet towel over parts with the duct tape to prevent it from burning. For the first joint, I taped the tube to the side of a square brick to provide a good right-angle. From there on the angles pretty much just work out provided the tube lengths and cuts are fairly accurate. The only thing to be careful of is ensuring the frame sits flat and isn't twisted at all.

If anybody repeats this, take some care cutting the pipes at 45 degree angles. The closer you are the easier the joint will be to make. I found that using a large protractor I could get the angle to within 2 degrees or less, and the joints are quite close. I just cut the angles by eye, which I found I could reliably do to within about 10 degrees after a few. After that, I used a grinder to fine tune the angle. A regular old metal file works fine too, but it's a bit slower. Spend the time to get the angles right in the first place and make sure the ends of a pipe slope in the same plane.
- posted by Simon McAuliffe @ 09:39