How To Make (Almost) Anything

Anmol Madan, Human Dynamics Group, anmol@media.mit.edu

Welcome to my web page devoted to MAS 863.04 ! This page started out as a quick hack, but is slowly evolving into a detailed representation of my work in the class. ;-) Almost all pictures were taken using my Nokia 6600 cell phone. 

This page explains all my assignments in this class. Click here to see a storyline illustrating the evolution of my Final Project.

An interest  in InterestNetworks
CAD model of the final project

At this point, all I had for my final project was a name. The idea was to learn to use a 3D modeling environment. and I chose Blender, amongst 3D Studio Max, Rhino, Solid Works, and others.

Blender is open source, and took about half-an-hour of familiarization with the manual, and a great TA session with Manu. Another reason to use Blender is that it integrates with python, for real-time applications.

The relevant image is the Blender design (blueback). It shows a model for a GSR bracelet and leads that extend to the palm, a 3d accelerometer package and a fictitious head-mounted display. 

High-energy laser weapon systems
Laser, Water-jet and Vinyl cutting

Laser-cutting is quite a bit of fun. Great for acrylic, cardboard, and effective at resolutions up to a thousandth of an inch. I made this bracelet with interlocking links, where each link fits tightly with the next. I got the fit right on the first try, and its important to remember that  the laser beam spot is about a thousandth of an each (Thanks John!).

The bracelet is made of multiple such links. Some of the links are embossed with my name, affiliation and phone number ;-).  A lesson learnt was that the links have a high-failure rate, at the female side of the interlock. Next-time, do some back-of-envelope stress analysis before designing the shape.

The bracelet is made of multiple such links. Some of the links are embossed with my name, affiliation and phone number ;-).  A lesson learnt was that the links have a high-failure rate, at the female side of the interlock. Next-time, do some stress analysis before designing the shape.

Open you eyes for the very first time
"Hello World" milled micro-controller board

We made our first micro-controller board, based on Neil's Atmel ATtiny15 design. The board is powered from the serial port of the computer, and sends out "Hello" over the serial line. Milling on the Modella  is an easy way to make a one-layer SMD PCB. (More than one-layer is possible too). We used Avra to assemble, and Uisp over the parallel port to program.

As you can tell, I haven't done SMD size soldering for a bit, and could use some practice. Neil had us use the black programming connector. Although convenient, the spring was an issue (we removed it for future programming). I think it might be worth breaking out an ISP programming header for the AVRs. I also struggled with the serial board connector, which I learnt how to fix  in the next micro-controller assignment (Thanks Vincent!)

The 3D scanner is an amazing piece of hardware! I had this vision of scanning and reproducing my face & head, and printing out miniature 3d key chains - at conferences, interviews, social events, people wouldn't ever forget me ;-)

Scanning and 3-point resolution are explained really well in the GeoMagic manual. Remember to convert the point cloud into polygons and then decimate. The fill holes function sorta works. Hair turned out to be really hard to scan, so I had to redo scanning,  wearing a cap.  Also, the fewer scans you register into a ploygon mesh the better, since you end up with few deformities (notice my deformed ear ;-).  (N-t-m, add picture)

My scans were relatively quite good, especially since I couldn't keep my head on the turntable for autoscan (Unless someone decapitated me ;-), and did this almost entirely by myself. (Try aiming for the bottom of your chin with the laser scanner and pressing mouse buttons at the same time ;-)

When I tried printing out my 3d head,  the Zcorp rapid prototyping system generated not-very-well bonded plaster ;-), which did not last long enough for me to take a photograph. The way to solve this is to scan and import into Blender, convert the surface of my face into a solid, maybe overlay a texture of my face, and then print on the Zcorp. 

The statasys uses ABS plastic, which is much tougher than the Zcorp, although it takes longer to print and is monocolor.  Personally, I enjoyed working with this machine more becuase of the durability of the final printed product. I made a case for a bluetooth GSR module, which I also used in my final project.

For a change, I designed this in SolidWorks instead of Blender. The comparison between intuitive ease of use between SolidWorks and Blender is like MS Paint and Autocad. Blender may deliver more, but for 3d-printing type of applications, SolidWorks is easy, fast, reliable and exports to STL without any hassles.

Sensorless sensing 
Vinyl-cut and Etched Microcontroller proximity sensors      

For this assignment, we had to build a capacitive proximity sensor based on Neil's EFE board design. The sensor is essentially a conducting pad, one terminal connected directly to the AVR and the other through a 1 Mega ohm resistance. Human capacitance and the 1 M resistance form an RC circuit and we measure the charging and discharging curves, As your finger moves closer, capacitance changes. The A/D on the AVR isn't  fast enough, so we undersample the voltage  to extrapolate the curves (ideally for both charging and discharging curves).  The picture tries to capture some of my notes from class. Neil's group has taken this to the extent of identifying objects/substances based on  frequency curves ( e.g. milk from water). The AVR sends the measured value over the serial line to a computer, running python.

To add some spice, Neil wanted us to vinyl-cut the boards. My vinyl board worked but was flaky, so I etched another one (on the right). As seen, my soldering seems to be getting better (Use the microscope!). The conducting pad on the bottom right of both boards is the capacitive sensor. Also, the right way to solder those serial connectors is as follows.. tin and trim wires with an mm of length exposed, place on connecter pin with conductor touching mid-clamp, lock mid and last clamps and put a drop of solder (N-t-m, add pictures). 

An interesting thing for me to try out is to make an accelerometer using the capacitive effect sensor, using a suitable material on the water-jet or Modella.

A  firm warning from John about machines and blood.  We had to make an injection molding mould using the milling machine. As usual, when my creativity wanes, my ego creeps in. I decided to make a business card/ name plate mould. I was out of Media Lab business cards, and could stamp out a few thousand to last me this month ;-)

FeatureCAM turns out to be really cool to look (as it demonstrates the milling pattern), and the milling machine resembles an out-of-place nuclear reactor control station. 

We used this assignment to mill a mold for injection molding. I designed a pair of  chopstix with an "A" on the top, and about 4" in length.  I made the original design in Blender, but then John had to fix it in OMax  Layout. ;-)  

Since I was late with the Injection Molding and needed supervision, I spent a fair amount of time with the Vacuum former. An interesting and easy to use piece of equipment, you basically wait until PET melts (sags by about an inch), and then vacuum suck it onto your mold. The nice part is unlike Injection molding, your mould can be anything that doesn't melt easily, balsa, wood, clay or even plastic. I decided to make packaging for some accelerometers, which someone joked were now sturdy enough to be put on  skateboards.

The top-row left image is the Mark I design, which has the boundary edges sticking out for better bonding. Mark II is more ambitious, and aims for a tighter fit. The third  picture illustrates the differences between them more clearly.

For injection molding, I decided to make personalized (mini) chopsticks. Since I was going home for the holiday season, it would be nice to have something interesting and usable for my friends and family. 

Internet Zero (i0)
Communications

This was our first and only assignment in teams, and the idea was to build click encoding and click decoding protocol based on internet0. The fundamental idea is that it is easily possible to run TCP/IP style packets by simply encoding frame structure in micro-controller firmware, rather than implementing a multi-layer stack and supporting API.  Raffi's papers describing the project can be found here. 

We made two boards - the click encoding and the click decoding board. The click decoding (bigger, up front) was designed and etched by Gerardo (Thanks for the pic!) and Vincent, around an ATmega32. With help from Amon  and Hector, I wrote the async state machine code that read  UDP packets in click, and sent them over the serial lines to a computer. The second image describes the i0 protocol. 

rcv_int.c - ASM in avr-gcc to decode clicks

Microtext
Micro Fabrication

Manu gave us a detailed introduction to the  excimer laser micro-machining system and the confocal microscope. The laser is great for machining and micro fabrication up to a few mils.  At the Fab lab, the excimer system is used to develop on three different kinds of base media, and the combination of these can be used to create floating structures. As it turns out, a major application of excimers is in laser eye surgery.

For this assignment,  I wanted to micro-fabricate some text on a metal base using the spot mask, which I would use in the display assignment. 

Exam Assist projected display
Display Interfacing

Neil gave us an introduction to LCD interfacing and also showed us how he could use an ATtiny13 to generate an NTSC baseband signal. The picture shows a Hello World signal generated from an AT Tiny.- exciting stuff!

I wanted to build on my previous assignment, and create a projector-type pen which would use a the microtext as a slide, and a tightly fitted LED for illumination.  Based on some back of the envelope calculations, by holding a pen at a distance of few inches, I should be able to project a readable size image.