RepRap: Blog

Labels: Printed electronics

We've talked for some time about using low-melting point alloys like Wood's metal (65^oC) and Field's metal (40^oC) in a RepRap write head to allow us to build electrical circuitry, and such a system is being planned for RepRap 2.0 "Mendel". We have also talked of trying them in exactly the same extrude head that we use for polymers, for which we'd need them in the form of a 3mm rod.

I've always thought that the problem with this might be that the (admittedly small) melt cavity inside the head would simply drain out the bottom uncontrollably.




But the other day my student who's working on this, Mike Samuel, had a brilliant idea: why not replace the cavity with a small heated button with a fine extrusion hole through it, then just touch the metal on the back when you want it to melt and extrude? The extrusion system already has a mechanism to back the rod off; this would remove it from the hot-spot and stop the flow.

The head would be like the above. The brass part would have a heating coil, like the ordinary extruder, and a thermistor to measure its temperature. The Wood's metal rod would be fed into the top by the existing transport mechanism that's used for the 3 mm polymer rod we use. And all this would use exactly the same electronics and software to drive it that we already have.

So I tried to make some 3 mm Wood's metal rod.

My first experiment was rubbish: I had a small crucible with a 3mm hole in the bottom heated by a soldering iron. The hole led through an insulating block with a 3 mm hole to a cooled section with another 3mm hole, where I thought the metal would solidify and be able to be pulled through - pulltrusion, as it's called. The whole thing locked up frozen solid, and nothing would move anywhere.

Then I looked at the silicone tubes I was using to carry the cooling water. Dang me if they weren't 3mm internal diameter. I put one on a funnel with a clothes-peg at the other end, heated them all in a beaker of boiling water, took them out and quickly hung them up, then poured in the Wood's metal:



After it had cooled I just sliced away the silicone tube with a scalpel, which left a perfect shiny rod:



I measured it: 2.95 mm, all the way down :-))

Labels: Field's metal, RepRap electronics, Wood's metal

I ran more experiments today using my 1/4^th inch (6.35 mm) ID screw pump. I began with the metal drill bit that I hadn’t used in yesterday’s experiments.

Metal drill bit

Maple syrup

As you recall, water did not pump well at all in the rig. Adrian suggested that syrup should pump better than water. He was correct. Maple syrup did pump rather well in this configuration.

Semolina grains

I tested the configuration with uncooked semolina grains. The metal drill bit pumped that media wonderfully. I tried putting my fingertip over the end of the barrel. The semolina collected behind my finger and the drill bit was thrust back out of the rear of the barrel with considerable force. A very solid plug of semolina grains were packed in the barrel. I had to literally drill that plug out.

CAPA 6800 pellets.

I fed polymer pellets into the “t” junction. The metal drill bit tended to chop these up and pump them to the end of the barrel. Most, however, of the pellets were thrown back out of the hopper. The pellets that emerged from the barrel were mangled rather than chopped into proper bits.

Visual inspection indicated that the pellets were considerably larger than the helical passage in the metal drill bit. The pellets had to be deformed or chopped until they could be fitted into that passage.

Wood auger bit

I repeated the experiments with semolina grains and polymer pellets. The wood auger afforded a considerably larger passage than the metal drill bit did. Both semolina grains and polymer pellets were pumped effectively. The polymer pellets were not mangled as they had been with the metal drill bit.

Observations

Considerable torque forces are being developed when tough polymer pellets are being cut and mangled. Hanging on to the barrel with a spanner is not practical. A frame holding both the barrel and the drill is needed.

Galacticroot over at CNC created a screw pump for wax that was my initial model for how to do things.

I built up a rig like the one that Galacticroot did. I used 1/4 inch ID fittings to do it which makes for a much cheaper and more dimunitive pump that Galacticroot had.

Metal bits

The first thing I tried was a 1/4 inch metal bit. I used my electric drill for power and initially tried to pump water with it. You can see it and the barrel that housed it at the top of the picture. The first thing I noticed was that the metal bit pump would not self-prime. I primed the pump and ran it again. At that point I observed that the metal bit would not pump water under any circumstances. Indeed, it seemed to slow down the natural flow of water through the pump. Reversing the rotation of the bit made no difference save that a bit of water began to flow out of the pump at the end closest to the drill and the flow out of the extruder hole pretty much stopped.

At that point I decided that the fit of the drill in the barrel was too loose. I greased the barrel heavily with petroleum grease, reprimed the pump and tried again. There was no useful pumping being done. I observed a tiny bit of pumping in the reverse direction but nothing significant.

Having failed I went back to the literature. The first thing I noticed is that the diagrams I had showed the flight (thread) length on such drawings as I had pretty much identical to the gross diameter of the screw. On our 1/4 inch metal drill the flight length was twice (1/2 inch) the diameter. As well, the proportional depth of the threads was much, much larger than what you encountered on a polymer pumping screw. Indeed, there was little core shaft as such in a metal drill.

Wood augers

Next, I tried a wood auger. The flight length of the wood auger was 3/8th inch for the 1/4 inch diameter as opposed to 1/2 inch found in the metal bit. The wood auger was also much longer with many more flights over its length. I acquired a longer barrel and rebuilt the pump. That configuration can be seen in the picture above.

I installed the wood auger, greased the assembly, primed the pump and began again. With the wood auger I got an appreciable pumping action, but nothing useful. Looking more closely at the wood auger it was easy to see that the depth of the flight was proportionally huge compared to a polymer pump screw. There was again almost no core to this auger.

Masonry bits

There are several types of masonry bits. Revisiting the local hardware store I encountered one proportionally almost identical in form to a polymer pump. Sadly, it was 3/16ths inch rather than 1/4. All of the other masonry bits were simple twisted rods and apparently useless.

In the morning I will visit a much larger hardware store and see if I can acquire proper masonry bit to continue the experiments.

Apologies for the delay in posting this....

A few weeks ago I attempted to produce some bearings using Field's metal and RP moulds. The aim was to prove that it was possible for a RepRap machine to produce its own bearings, as Field's metal melts at approximately 70 degrees celcius......

I tried 2 methods
1. Casting the metal in a mould with the 8mm steel bar in situ, using silicone grease as a spacer.


2. Casting the metal in a mould with a plug that could be hammered through once the metal had solidified.


Method 1 failed as a bearing (but would be good as a locating/fixing mechanism - ie calibrate
machine, then set all joints in metal)

Method 2 produced some nice bearings..... (Only tested by hand so far)


The best results were achieved with a centre plug diameter of 8.2mm (to fit an 8mm bar)


Will be performing some basic tests on them this week
Fulll report available if anyone is interested

I had a pretty busy weekend. I managed to:
* Progress the virtual comms and testing infrastructure (see svn branch), which will soon allow me to complete the remainder of the comms infrastructure. This will eliminate those occasional pesky errors.
* Updated my old graycode wheel perl script and put it into the reprap subversion repository. See reprap/misc/graycode in subversion.
* Using new graycode script, produced a template for a bi-directional position sensor for a DC motor controller.
* Tried out my previously imaginery process for turning PC artwork into plastic objects, pre-reprap. Potentially a useful process for "repstrapping" with kitchen level technology.
* Some work on reprap application to eliminate most of the hard coded magic values and make them customisable via the preferences screen.
* Tested my old thermoplastic bushing idea for my old repstrap design. Mixed success. It fits so perfectly that it doesn't turn very easily, but it certainly removes any play in the threaded rod. Needs more work...


This is what my 2-bit graycode bi-directional position sensor currently looks like. It's so simple it could just as easily have been created by hand, but using the script allows for very simple changes to diameter, spacings, etc. In fact, these are the parameters for the graycode script. It's fairly versatile:
* Number of bits (arbitrary number)
* Wheel dimension (mm)
* Inner dimension (mm)
* Strut size (mm)
* Span (degrees). For angle sensors, it's often useful to use less than the full 360 degrees.
* Stagger (~mm). Stagger between sensor positions. This allows bulkier sensors to be used and the graycode rings are offset. The application also produces a cutting/placement guide that should where the sensors should be placed, which is pretty essential when staggering.
* Resolution (dpi)
* Portion. Fraction of wheel to include (entire wheel is not necessary if used for angle measurement and using less than 360 degree span)
* Reps. The number of repeats of the graycode pattern around the wheel. This is the new feature needed to make a mechanically balanced wheel for a fast turning position encoder.

I started to add STL support a while back, but that's not even close to complete. The primary output is a bitmap ready for printing on a laser printer.

My conceptual process to turn these into actual plastic objects was:
1. Print out object onto tracing paper with a good laser printer (new toner). Inkjets don't work as the inks are UV transparent.
2. Take a piece of thin steel, and spray it with PCB photo-resist. Bake dry for impatient people like me.
3. Place tracing paper pattern on metal. Cover with thing sheet of transparent plastic (not glass which is UV opaque) to hold pattern close to surface. I used the plastic from a CD cover, whatever that is.
4. Blast with UV light
5. Use developer to bring out the pattern on the metal
6. Using electrolysis, "melt" away the unprotected part of the pattern, leaving a thin metal version of the pattern. Alternatively use copper sheet, and use ferric chloride to remove unwanted metal.
7. Press metal pattern into molding plasticine. Pressing it deeper makes a wider part.
8. Create plaster of paris mold from plasticine and let it dry.
9. Melt plastic into mold

My goal was to produce one of the gray code sensor wheels and Forrest's sample involute gear pattern.

I tested steps 1-5 but it seems my photo-resist is a bit too old now. In fact it's quite a few years beyond it's "best by" date, so it's time to order some more. The patterns were not holding well enough for me to carry out the remainder of the steps. I will try again when I get some more photo resist, but all the local suppliers seem to have stopped selling it. I'll try RS.

I tested step 6 using several different electrolytes. Plain salt (NaCl) was the fastest but has the downside that it produces some chlorine gas, so ventilation is a good idea. It doesn't need much salt though. I hand painted a pattern onto a piece of metal, and the pattern was cut out with good fidelity. It takes quite a few hours to cut, so you have to have some patience. It makes the more conventional ferric chloride approach more appealing if I find or make some plain copper sheet. I have lots of ferric chrloride, so I might try that next time. The electrolysis approach produces nice steel templates and just seemed cooler :)

I have previously tested steps 7 to 9 using PLA to produce some gears. It works quite nicely but destroys the plaster mold so it can only be used once. It also takes a bit of scraping to get excess plaster off the gear before it will work nicely.

So all steps tested, but not quite working yet.