The material is called “Alumide”. Effectively, it is aluminum mixed with their another popular 3D-printing material: PA2200 (White Strong & Flexible). The aluminum gives the printed parts a nice sparkling grey texture. The materials feels more brittle to me, and possibly not as tough. This is partly due to the material properties, but it also may be due to the fact that I shaved off a lot of material and shrunk the gear (and the teeth) down a fair bit from the last print. The gear housing and motor mounts were a little too flimsy (1mm thick walls). They held together alright, but I had to be careful when mounting the motors and working with the gears so as to avoid damaging the housing. The surface roughness of Alumide seems about the same as the PA2200 to me. Although I had significantly more trouble working the gears to get them to run smoothly, I attribute that to the smaller gear size and the slightly misaligned (on purpose) worm gear.

One of the improvements I made was to reduce the number of gears types down to three: two wormgears (same size), two large gears (one merged with the outermost limb), and 8 compound gears. I arranged them so that they could all (not counting the worm-gears) share the same six axles. I was concerned that the surface roughness would cause problems with gears rubbing against each other, but I didn’t have a single issue with that, even with the gears packed in so tightly. I even made the housing sandwich together such that I could give them more room if necessary, but I didn’t end up needing it.

As a bonus addition to this project, I found a wiimote at my local thrift shop. I used cwiid (a wiimote library) to take in the arrow commands from the wiimote. The program then passed simple commands to an Arduino, which then drove the two small motors using two H-Bridges and a cellphone charger wall-wart. It is a rather inelegant and jury-rigged control system, but it works for the purposes of testing.

This draft showed itself to be a decent success. It is fairly responsive and fast. It is a little jerky, but that could have been avoided if I had utilized PWM. Control options with the wiimote were limited due to the program having no knowledge of what the relative segment orientations were. In future revisions I will try to incorporate some sort of feedback loop (so the program knows the orientation of the limb). Another degree of freedom would also be nice. Touch sensors will also eventually be essential to a useful robotic limb. Also a rocket launcher would be cool.

The lens (top right), surrounded by kneaded eraser

It was difficult focusing on a subject at this “zoom” level. Even the slight movement of pressing the “take picture” button was enough to ruin most of the shots. Lighting was also a problem. Since there was typically only a very small gap of space between the subject and the cellphone, the light had trouble reaching the region of interest. With a little patience, however, I was able to take the following:

The letter "C" on my laptop's LCD panel. Note the subpixels.

Lincoln's Eye on a US penny.

Part of the text "LIBERTY"

Mintmark on an old penny (note the grime).

Mintmark from a newer penny.

The Lincoln statue in the Lincoln memorial on the back of a US penny. It came out fuzzy, but I had to try.

Surface mount resistors. I am not sure what the standard sizes are, but these are roughly 0.5mm wide and 1.5mm long.

The subjects of the above pictures. Note that the mintmark is below the date, in the bottom right of the penny. Also, the resistor picture is from that cluster of resistors in the top middle of the chip (part of the DVD drive).

I’ve always had an interest in robotics, but I always got stuck when it came to the mechanical fabrication. What was I to do? I don’t have access to a machine shop, and I am not that good at making anything of quality with just my current set of tools. There are robotic kits out there, but I feel like it would defeat the entire excitement of design to buy a robot kit where all you do (mechanically) is bolt the pieces together. I want to design the entire thing from the ground up. Particularly, I want to design the mechanical assembly from scratch. Well, I finally have a method: 3D printing.

3D printing is an amazing branch of fabrication. The workings of a 3D printer aren’t very different than the workings of your home inkjet printer. Just imagine, if you will, that instead of printing a thin layer of ink onto your paper, that you printer printed a thin layer of some solid material instead. Well, if you put the piece of paper back into your printer, and printed the same thing again (and did this over and over several times) the thin layers of material would build ontop of each other, and a 3D solid would form. This is effectively how 3D-printers work. SLS (Selective Laser Sintering), for example, works by depositing a thin layer of powdered material onto the work surface. A computer guides a laser over the poweder, melting and solidifying selective parts of the powder. When it is done, another layer of powder is deposited, and the laser melts parts of this next layer, which will also merge with the melted parts of the layer below, building a sold. When this process finishes hundreds of thin layers, the solid part can be removed from the powder.

This process can get quite costly. The cheapest commercial-grade printers available start at around $15,000. Most businesses that offer 3D-printing are targeting companies that don’t mind spending several hundred dollars on a single test-print to troubleshoot a design. Definitely not an option for an individual like myself. There are alternatives for the hobbiest, like makerbot, which is a $500 printer that prints low-resolution ABS, but even that entry cost is outside my current budget. Thankfully, I discovered shapeways.com. Shapeways is a 3D-printing company in Europe that targets the hobbiest/individual/home-business. They offer a variety of materials (including such options as  stainless steel and colored sandstone). Their prices are reasonable, and shipping is included (which will be a big plus when I move back to Alaska).

Getting something printed by Shapeways is relatively easy. Just give them a 3D model and they get the model to you in 10 or 14 days. I used Solidworks to design the 3D model of the robotic limb (and a bunch of test parts). After uploading and ordering, the following came in a nice little box:

3D printing test pieces

These parts were printed as clearance tests. The two pieces on the far left are to test slot widths. The piece in the middle is meant to see how close two pieces could be printed before they fuse. The part on the right is to test the fit of different diameter pegs in different holes.

My horribly failed attempt at designing a printable Iris Diaphragm

This is my first attempt at a printable Iris Diaphragm. The design failed due to overgenerous clearances on my part. Oh well, something to change for another draft.

3D Printed robot arm

And here is my pride and joy. It isn’t really a robot “limb”, but it has some basic elements (rotation, linear actuation, gear reduction, etc) that would be needed in a limb. All of the parts were printed separately, but designed to snap together. The above picture shows it assembled, except for the worm gears, which need to be joined to the motors before mounting. The gear assembly on the right will allow the entire “limb” to be rotated about that large gear on the bottom. The small gear on the left pushes the linear gear which actuates the pointed tip on the far left. You can see the two slots where the motors and worm gears are to be installed.

I obtained some really small motors from Seeedstudio.com to drive this device. The motors are 15mm long and about 5mm in diameter. I was worried that they wouldn’t be strong enough to overcome the static friction. As it turns out, they are quite sufficient:

I count it as a success. In the second part of the video, you can see me unjamming the gear to get the tip to actuate again. The other gear assembly also had problems, but I was able to keep the problem gear from popping out of its bearing by placing a little kneaded eraser.

Things to focus on for next time:

• Tighten up the tolerances (my peg-and-hole test-piece will come in handy).
• Figure out a surface hardening treatment. The plastic is cheap, and I would like to keep using it, but if it is going to be a bearing surface, it will have to be able to handle the wear.
• Figure out some surface treatment to help smooth the surface.
• Refine the linear gear to prevent jamming (although tightening up tolerances should fix this).
• Figure out an analog position sensor for feedback into control electronics. My goal here would be to make my own servo control, but with less expensive ($1.50) motors and integrated (smaller–well not really) gearboxes.

Let me know what you guys think in the comments!

my first de-RCed car

During one of my many visits to the nearby goodwill, I spotted a rather nice RC racecar. However, like most RC cars at the goodwill, it lacked a radio controller. I thought that this would be a great opportunity to attempt RC car autonomy using the Arduino chip that I had recently been tinkering with. Of course, I failed at figuring out an elegant way to connect the arduino chip to the existing RC circuit, so I went back to the goodwill and purchased another RC car. I got lucky on this one, since the circuitry was very straightforward (and all 0-5v logic).

I successfully bypassed the RC car’s onboard control chip. My arduino chip is in power. I rigged up an IR distance sensor onto a little servo. In my first test run, I programmed the car to maintain a minimum and a maximum distance from whatever was in front of it. The distance sensor is pointed straight ahead, and the car only goes forward/backward. Check out the cool video: