You can tell a lot about a person from their resume. Just today I had the chance to go back in time and look at a number of my old resumes, and I thought it would be fun to show them off. I’ll do this in reverse chronological order, so you can see the progression as it happened.

Freshman Year
The one that started it all… Not bad on design, considering it was my freshman year. Pretty light on experience/information though.

Sophomore Year
HA, I have to laugh at myself after looking at this resume. Loving the cheesy dropshadow on my name. Finally got some real experience to put on the resume, despite it only being my second year at TAMU. The internship I got after my freshman year really paid off, and the CAD skills I refined would help me through the rest of my career.

Junior Year, just after my NI internship
Not much has changed since Sophomore year, other than another internship at National Instruments. After that experience, my objective statement changed to: “An internship at a company that provides both an outlet for my talents in computer engineering as well as opportunities to move into leadership and project management responsibilities, even as an intern.” I had a much better idea of where I wanted to go, but I was still not sure how I was going to get there.

2007
This was the first resume I created after my GPA had dropped significantly. Despite this, I had another two internships under my belt: one with Hewlett Packard, and a second internship with National Instruments. Previous to this, my GPA had been my big selling point; after it fell, I felt it would be better to focus on different aspects of my experience instead. This was the first time that I really put some personality into my resume. I hadn’t found what I was looking for yet, but I had started to find myself.

2009
My favorite resume so far. After a year and a half in a research lab, my experience had grown significantly. I wanted a way to better visualize the time I had spent while I was in school, so the split column layout seemed appropriate.

My work on the happyfuntouch project has evolved into a paper which was recently accepted to the 4th conference on Tangible, Embedded, and Embodied Interaction, held at the MIT Media lab in Janurary next year.

Here’s the paper:
Scanning FTIR: Unobtrusive Optoelectronic Multi-Touch Sensing through Waveguide Transmissivity Imaging.

I’m super excited about the conference, along with meeting a bunch of my heroes in the field. They’re also hosting several workshops on day two, and I will be attending the workshop on Making Textile Sensors from Scratch.

And here we are with the most current iteration of the PCB design. Previously, I had assumed that making all the modules the same was the best solution for this design. In reality, this was only advantageous in terms of manufacturing ease. Separating the phototransistors and LEDs and placing them on different boards allowed a level of miniaturization and integration that I couldn’t achieve before.

In addition, photodiodes were used in place of phototransistors to improve the response of the touch screen and to meet our 100 fps refresh rate target.

Finally here is a picture showing each iteration side-by-side so you can get a feel for how much smaller these boards are.

Feeling much more confident in my own soldering skills, I took the design of the previous evolution to the next logical stage. Featuring 8 phototransistors and 7 LEDs, these boards represent the refinement of the previous design. Feeling good about my SMD soldering abilities, I went with TSSOP ICs and 1206 and 0805 components otherwise. Ribbon cable connectors were still used, though this would be the last iteration to do so.

Unfortunately, an unseen problem with the previous module’s LED driver configuration made its way to this prototype as well. The ULN2003A current-sink driver I was using had it’s ground connected to the load resistor for the LEDs to save space on the board layout. Unfortunately, this meant that when any LED was turned on the ground for the ULN2003 was lifted, and all of the other LEDs would glow very dimly. This wasn’t noticed in the previous iteration which was very sensitive to ambient lighting issues, but the precision of this prototype made the problem clear as day.

Ponoko was again used to manufacture a laser cut enclosure seen in the pictures below. This enclosure design would prove to be the biggest design leap in this iteration.

After the depressing reality of my broken $300 PCBs, I decided to take a new tack. I wanted to try my hand at assembling these myself, so I build a slightly simplified version that had more through hole components. I still went with SMD ICs, though I chose SOIC packages, as they seemed like the easiest SMD package to solder at the time.

These boards are glued to laser cut stands from ponoko, and can be seen fully assembled on the main happyfuntouch page.

In addition to this, the boards are designed to be daisy chained together for ease of connection to a microcontroller. Daisy chained shift registers enable individual LED switching and control the enable bit of each multiplexer. 4 IO lines select the channel on the active multiplexer and are shared among all boards. Each board contains 16 multiplexed phototransistors and 6 IR LEDs.

The complete design can be seen both on the happyfuntouch page and in the picture below. It’s not easily seen, but there is a 3mm acrylic waveguide sitting on the laser cut stands. (all the optoelectronic elements are optically coupled to it)

I wanted to spice things up a little bit here and show the past/present/future design iterations of the happyfuntouch system. We’ll start out with the first PCB I ever manufactured.

I spent about $300 total on 6 of these PCBs, and had them produced and assembled in malaysia. It cost $40 shipping alone just to send the parts there, but I messed up and only sent them enough parts to make 3 complete boards. On top of all that, I used the wrong footprint for a shift register which rendered the boards useless. It was my first big failure of this project.

The boards consist of 32 multiplexed Phototransistors and 8 current-sink driven, shift-register switchable LEDs. I designed them to test the Scanning FTIR principle on a small scale, so each board has it’s own amplifier stage for the phototransistors and they cannot be daisy chained. This inefficiency was quickly realized and fixed for future iterations.

Turns out once Ponoko receives your materials, it only takes them 5 minutes to finish the job entirely.  Glad to see they are working hard (on Sunday nonetheless) to get their SF hub up to speed.

Coming to terms with supply chain delays and manufacturing turnaround time has been one of the most frustrating parts of this project.  I placed 4 orders on June 11, each one from a different vendor, but all necessary for the 2nd prototype I’m building.

All the parts I ordered from mouser arrived the next day, which isn’t surprising since they’re in Ft Worth and ship everything at 8pm the same day you order.  SparkFun was next, arriving the next week.  Ponoko on the other hand JUST ordered my materials, for an order I placed over 2 weeks ago.

Ponoko just opened their US hub, so they were hit with a huge amount of demand just as I placed my order.  Not being a ponoko prime member didn’t help much with the delay since they are taking care of all those orders first.  At least shipping was cheaper from San Francisco than from New Zealand.

On top of all this, the order I placed with Silvercircuits, the PCB shop that I’ve used for all the prototype boards in this project, is delayed in Alaska right now, probably because of customs regulations.

I’m all for rapid prototyping, but when it takes you 2-3 weeks to get results, I no longer consider it very rapid.

Welcome to WordPress. This is your first post. Edit or delete it, then start blogging!