MAS 863

Fall 2005

   How to Make (Almost) Anything)
THE

CENTER

 FAB

LAB

 Patrik Künzler                                                             patrik@mit.edu 

Smart Cities Group
FOR BITS AND ATOMS The Media Lab  @ MIT
patrik künzler Originally, I studied medicine in Switzerland and in Jamaica. Jamaica provided a great opportunity to practice low-budget and evidence-based medicine and left me with great friendships, sometimes involuntary reactions to Caribbean music, and a basic knowledge of Patois. During my studies, I spent a lot of time doing research in molecular biology labs, including the US and Japan, and restoring old motorcycles and horse carriages. So on I went, to do post-doctoral research in the lab of Susumu Tonegawa, making and analyzing mice with brain-region specific gene deletions. However fascinating that was, I always had my old cars (evil tongues called them rust buckets) to tinker with, so when I joined the car design studio at the Media Lab, I could revive old passions. The more I discover the Media Lab, the more I realize that this is THE place to combine different kinds of knowledge and thinking the generate something completely novel, beautiful, and useful. Or to discover things that I never thought existed. The naive enthusiasm of a novice...
assignment 1

render/ animate a potential final project, for a brief introduction/ presentation off the class website.

 1) an active seating system
that assists its occupant’s
movements, and protects
the occupant, while being
as comfortable as a piece
of clothing.

2) a 3D assembler for 
microscopic objects.

assignment 2

think about, fabricate GIKs

Some thoughts about GIKs:

1) Structure: Try to push GIKs towards maximized volume or maximized surface in a fractal manner. Determine assembly properties by making GIKs with different friction coefficients at their connection points. This could determine size and shape of the most common GIK complexes in a random (self) assembly process. GIKs that are flexible in one plane allow for circles and feedback loops within GIK structures without requiring the extra energy to put in the last part. Maybe this energy helps stabilize the structure, though.

2) Intelligence/ Information: Combine sets of parts that allow flow and processing of information, using simple or complex electronics, light- or liquid processing properties. The proportional amounts of different types of GIKs, as well as some random factors, determine the behavior of the resulting computer.

3) GIK vs. nature: Can we create chemical reaction equivalents between GIKs? Biology typically happens in some sort of liquid or equivalent thereof. How can we mimick it? Can we make self- assembling GIKS that require no electronics?

Predicting randomly assembled structures: The GIKs in the picture on the left have three types of connector bays: Normal, tight (indicated with one circle), and loose (two circles). The GIK on the left has tight and loose in the same direction, the GIK on the right has them at a 90 degree angle. The bays are designed so that loose+loose will barely be stable, and tight+tight very stable, with the other combinations in between. The ration of parts in a random assembly process will determine the likely final structures.The picture on the right shows circles of different sizes to make bi-curved surfaces.

assignment 3

use waterjet; have detailed plans about final projetc, discuss off website

Previously, I have waterjetted a prototype of a "in-wheel suspension" design out of lexan after designing the pieces in CATIA. Now, I waterjetted aluminum support pieces for a seat project, and flextures to connect the plywood pieces that interact with the occupant. Next, I will make snap-on pieces to connect the fixtures to the plywood pieces instead of bolts.

For my final project in this class, I will make a GIK assembler. At first, it will do a one-dimensional assembly process.

One way of doing it would be to use a toy-robotic arm. It would be more interesting and challenging to make a linear actuator, using a solenoid, a spring, and a permanent magnet at first. This will be followed by more sophisticated pieces. It would be cool to make the assembler out of GIKs. This would be a test of what we can put on GIKS, how precise GIK machines can work, etc

.

assignment 4

 

This week, we learned about pcb "board stuffing" and some basics about microprogramming.

We used a pre-designed pcb board in eagle, which also allows work on schematics. The goal was to learn the basics of organizing, machining, and soldering "stuffing" a board. Adjusting the modela milling machine depth was sometimes tricky. In order to save time the coming week, I decided to make cables to connect the tiny13 chip to a parallel port for programming, and a serial port for connecting the board to run the python program. Making the serial connector was a bit diffucult, and it wouldn't last too long, I decided that a drop pf expoxy might help. The board worked in the end.

assignment 5

So, being experts at board stuffing, we felt ready for a bigger challenge: To design a board that would blink when it received data, and echo the input characters we'd give it. Give it "a" it will spit out "b", and so on. The problem was made easier by the fact that we had a template circuit and python scritp for both, the echo, and the blink function, we just had to combine them. This didn't poof to be entirely trivial, since assembly code has some quirks, but it was even more of a learning experience that way. Cunningly anticipating future challenges, I decided to get everything running from my windows machine. I got some tips from a previous fab class entry, and help from linux- and windows proficient class members. I had the boards working on the linux machine in the FAB area, but it would not give the correct output characters when I ran it from the cygwin console. Instead, I had to switch to the windows command line. Still, it wouldn't always give me the correct output. There might be a problem with the serial port buffer. I could get my oh-so-convenient port replicator to work, except for the parallel port. There are still lots of challenges ahead, but I'm feeling more and more that I could make almost anything... Here is the schematic, board, and movie for the echo:

I found another way of getting the correct result using the echo program without having to alter it: If I put the LED on the connection of pin8 (VCC) and the input from serial, TX (with a resistor, of course) it will light up whenever Tx pulls down pin 8 to give it an input. Also worked;).

For my basic understanding of electronics, the class "Practical Electronics" at the Edgerton Center, has been extremely useful. In order to make microprogramming easier, Jesse and I started an "Assemblicon", where we (try to) explain each line of code.

 
assignment 6

This week, the assignement was to scan and print an object in 3D. That meant egomania time: I scanned and printed my face (50% scale, of course, to save material). It worked amazingly well.

Neil had made a modification to the cam.py program so that it would interpret color intensity as depth in the z-axis. this of course lent itself to a hack: Take pictures of a face with light coming from different angles. The shadows cast would be interpreted as height and therefore translated by cam.py into 3D deformities. We could get nice 3D shapes, but we had problems getting them into correct STL files that do not contain any nonclosed surfaces. Below are the original pics (<=200x200 pixels), and 3D renderings, negatives, positives, and two Janus head pieces.

assignment 7 Now we can use our skills to make a game. My idea was to make a simple keyboard that has three keys, from which the famous five tones from the movie "Close encounters of the Third Kind (1977)." The nature of these notes is such that all five notes can be played with three keys (each key playing only one note) only, since two of the five notes can be generated by combining two of the other five tones due to interference phenomena, so that our brains would hear five distinct tones. This little trick would also protect me from potentially getting sued, since I would not actually perform said melody in public (!), but only three tones from it. I started, but then got stuck and decided to postpone. I found the tones and their corresponding frequencies on the internet. Writing the code for the speaker device to play them turned out to be rather difficult, so I didn't get to finish in time.
assignment 8

Making molds for polymer injection was the task of this week. Injection molding is a science and black art by itself. To keep things simple, we made a 2.5D mold fout of 6061 aluminum on the HAAS mill. The piece was designed as a .dxf file using a template, http://fab.cba.mit.edu/classes/863.05/classes/11_21/template.dxf, and then turned into a program for the mill using the FeatureCamm software. Theoretically, the milling machine can mill to tolerances of 1/10th mill, but for now, we were sticking to simpler to make dimensions. I tried two types of GIKs. They came out pretty well, the polymer we used was somewhat springy and elastic and therefore was quite GIKable. here is the result, GIKs at two scales:

assignment 9

Using Internet Zero i0 to communicate. We decided to make boards for communication by IR. We modified the i0.6. circuits and programs, and we saw packets being sent to our transmitter IR LED on the oscilloscope. But strange things happened, and after a few clicks, all our three circuit boards started to behave differently than expected. This was most likely due to the fact that we did not put an input limiting resistor in series with the LED, we only had a pullup. A pin on the tiny can typically source 10-20mAmps, 50mAmps is max. Neil also suggested we put a resistor in front of the phototransistor in order to regulate its frequency specificity.

I didn't do the project of using a very loud/ big/ visible/ obnoxious device to secretly send data using i0 (secretly because to a bystander or to an observer, the events would seem random due to variation and length of the click window), that's for another time.
assignment 10 Weird materials assignment: Everybody in the class got a bunch or interesting and/ or weird materials to try out. I tried using electrically conducting epoxy in order to efficiently put electronic components on GIK, which might be useful, as shown in the animation for assignment 2. The conductive epoxy did not work well. It was too sand like, not sticky enough, and expensive. I ried mixing it with urethane containing epoxy, in order to add elasticity, stickyness and volume. It would be ideal if the conducting parts where GIK make contact would be slightly elastic. This worked better than the conductive epoxy alone, but still not well enough, even if I ratsered grooves for the electronics on the lasercutter. Using vinylcut copper taope is easier.
final project

My final project for the class is a collaboration with Jason Lapenta, which involves GIK carrying electronics. The idea is for the GIk to communicate as they are being assembled and to transmit this data to a computer, where the assembly of GIK can be followed in real time. We are doing this by giving each GIK at Tiny13, four visible light LEDs, and four phototransistors, one of both for each possible connection.

Routing a board in 2D (well, it's really 4D) was a challenge at first, but turned out to be fun. We designed a simple circuit at first. Sensing is non-directional, while sending out signal, is. This way, the pattern of GIK can be determined. We are not finished with the programming yet, but we have made some prototypes, and I have refined the technique for producing sandwiches of 4D boards. It is still slow and tedious comparing to sticking vinylcut pieces of copper tape to an acrylic GIK, see above.

The board consists of three layers, an outer layer for making electrical connections between GIK, a top layer containing the voltage regulator, the capacitor, two resistors, and the phototransistors, and a bottom layer containing the Tiny13 nad the LEDs. This is convenient, since the LEDs are each connected directly to a pin on the Tiny13. it also helps optically separate the LEDs, which is important, since they are being used in a directionally selective manner.

To cut out the board on the Modela, we used two runs with a 1/32 bit, 0.033 inches deep, and 0.66, respectively. We used the same bit for the holes, which were drilled 0.07 inches deep. A sacrificial layer of acrylic was used.

In the first prototype, I put the boards, consisting of one outer layer and either the top or the bottom layer, together so that the electronics would be on the outside. This would make it easier to troubleshoot. For programming the chip, I soldered wires directly to the pins, wich is much more convenient than using clamps. For the vias (the connections between the boards), I simply used stripped solid core wire. Soldering was easy, since all the relevant connections were on the outside. For space reasons, and for reasons of efficiency, I soldered the optically relevant elements in sideways.

With the second GIK I made, I arranged the boards so that the electronics would be on the inside. This created new challenges: The cables for programming the Tiny13 would have to be eailsy removable and would therefore have to be soldered to an easily accessible location, and, the relevant connections for the vias would be on the inside of the GIK and therefore not accessible with the soldering iron.

For the latter problem, I tries using soldering past at first, but it didn;t work well. Instead, i put solder on the pads around the holes. I also clipped the vias after soldering them to the first board so that they all would be different length. This made it much easier to put the two boards together, since I'd have to deal with only one via at a time, due to their different lengths. A drop of solder on the vias at the desired final height was also useful. After positioning the second board at the correct height, I simply heated the vias from the outside. the transferred heat would make the solder on the vias melt together with the solder on the pads. Alternatively, heating from the outside and adding solder to the inside, also worked.

The next step is to make the GIK communicate with eachother.