Practical, experienced robotics engineer looking for opportunities to work with engineering teams to solve problems, in whatever aspect of production. Outside of work, it's been nearly two decades that I have been building, iterating, and testing ideas in my free time - with more complexity every time.
Largely, this was established for MIT admissions, then maintained for job applications, and now serves as a slick repository for as much design data as I can digitize. Look around and enjoy!
Want more project details? New ideas? Did I spell something wrong? I'd love your input!
Email: tneedham 'at' alum 'dot' mit 'dot' edu
For about two years I’ve been working with a group at MIT’s CSAIL to develop enclosures and support material for some cutting-edge research they’re trying to deploy. It started off, as described in another section, on enclosures for wireless charging, but then I switched to another group (that demoed for President Obama!). My contributions started small, laser cutting square boxes for travel. I stuck around though and helped them refine the box into a more presentable shape, and ultimately (through four major revisions) into a design we can confidently contract out for mass production.
I’ve worked with professional product designers and contract manufacturers, refined the design and tolerances to make sure it will meet our specs, and gone through dozens of prototypes to visualize designs. It’s been a fantastic experience to see the project grow and really see how consumer products are developed. More to come as we document it better!
Over the summer I worked for a Tier One Honda supplier that was doing extensive testing manually. Many of the tests involved durability cycling of seat components 3,000-5,000 times and would take weeks, or would have to be removed from environmental test chambers and interrupt testing.
My Internship consisted of designing, building, testing, and documenting about a dozen boxes that would automate various aspects of the testing. Some interfaced with seat heaters through the CAN bus to cycle them on and off for days, others directly controlled power seat motors based on limit switches, current sensing, and temperature of the motors being driven. There was also a box to test power lumbar features inside a chamber that would cycle from 21C to 80C. The box was triggered off of the temperature change and would cycle the lumbar, checking the current draw for signs of premature binding as the part wore in.
The box pictured is the motor controller box, which had its own power supply, microcontroller, motor driver board, LCD with serial readouts, four temperature sensors, four limit switches, and two bidirectional 15A DC motor outputs.
The lights on the front would let the operator know if the box was functioning well at a glance, and it had an LCD for more detailed information, errors, and startup procedures. It also has two bypass switches to manually adjust the motors once the box is connected.
My best friend recently bought a car, and with the success of modifications on my own car (also documented here) combined with my “improved” engineering skill we decided to add a few interesting features to his Golf GTI. Improved is in quotes because my ambition grows with skill, so everything we’ve done on his car stumbled just as much as 5 years ago when I did mine, just with fancier tools.
This is the button panel he uses to control the major systems, it was waterjetted from ¾” aluminum stock, then CNC machined to get the relief on the opposite site for light diffusing and electronics. This piece was a sample for which finish he preferred, we went with a hand-sanded satin, dark acrylic, cast epoxy in the rectangular label cutouts, and machined aluminum buttons.
Using the same processes, this is one of the replacements for his pedals (the footrest is pictured). Waterjet carbon fiber, home-cast silicone rubber for the grips, and CNC/waterjet aluminum surround. This picture is a render, I’m still working on manufacturing these pieces in between final projects during the semester.
This is a rear view of the button panel, showing the PCBs I had manufactured in their mounting spots. I designed them in EAGLE, had them made in China with a white soldermask to diffuse the LEDs better, then assembled the SMD parts myself under a binocular microscope. Very surprisingly, it worked the first time I turned it on! Because of that I have 5 sets of these things and no idea what to do with them…
Another component of the button panel – the light gasket. I was very proud of how professional this looked when it came out of the laser cutter. It’s a very simple part, just stops light bleeding over from one button to another, but out of 1/32 rubber foam.
These are the final buttons, that by now are installed and being used in my friend’s car. They have Dykem on them to protect the finish while I inspected, finished the bottoms, and transported them. They were done manually on an engine lathe, since the CNC mill wouldn’t have given a good finish. I had expected to have some variance and was worried that they would bind once installed, but somehow all the purple ones were within +/- 4 tenths (0.0004”) of the spec dimension. I’m still very proud! (and recovering from those hours of careful turning…)
The “Glamor shot” of the business end of the lights. Showing custom PCB, assembly on heatsinks, black machined mounts, busses, and LEDs with thermistors.
All nine machined mounts; holes in the center for screwing into ceiling/loft, PWR rails with crimp connectors to PCB and between rails, and side supports to prevent heatsinks moving either vertically (falling out) or translating (sliding out).
Four of the eleven PCBs printed and assembled, serial #’s 6,7,8, and 9 - so these will be mounted above my desk, sink, entrance, and closet, resp.
All nine units, LEDs fully installed with thermal grease and screws, but PCBs only set on to test screw hole alignment and fit (PCBs were designed and sent out before heatsinks arrived - wasn’t sure if the scaled dimensions from Ebay would port over correctly. Thankfully, they did!)
My previous overhead light power supply. Took in 12v from a brick mounted underneath, and stepped it down to 3.4v to drive smaller and lower power SMD LEDs in a larger grid over my old desk. Worked ok, but not dimmable, had a loud fan, and created about a dozen shadows from each different LED. This regulator setup was made several years ago, to charge the supercap flashlight I made for my dad (also before I really knew how to use LM317s).
These lights make up the most electrically/electronically involved project I’ve ever worked on, having a batch order of PCBs just for them, wireless radio protocols to be implemented, power signal circuitry design, and lots of GUI programming on the controller side. At press time they’re not installed yet, but have passed final testing and are now only waiting on the receiver code to be finished before they can go up (the master code can be more easily modified, and is much more complex than the receivers’ which only have to set PWM levels and report back their data).
The goal of the system is to replace the stock lighting in my dorm room, which has only two points and leaves many areas dark. Additionally, they offer a chance to experiment with mesh networks of data generation and transfer, and truly upgrade the lighting, not just distribute it.
Each receiver (there are nine total) has two 10W LEDs mounted to a heatsink (one 6000k, one 3000k, controlled separately); a control board (Arduino Pro Mini interface, NRF24L01 transceiver, MOSFETs, indication and error LEDs, 4 thermistors, other support circuitry); two 40mm fans, an acrylic cover/diffuser, and a machined mount as a go-between for heatsink and ceiling (it also carries the PWR and GND rails). The power LEDs can be individually and completely dimmed, as dictated by the master controller, to change the color temperature and intensity as the day progresses. In the morning it may be very bright and mostly cool, then change to balanced temperature and full brightness in the afternoon, then dim into evening, and then cut back on blue light (warm the temperature) as it gets into night. Each receiver only has to set the duty cycle for the MOSFETs - the master controller has the RTC and lighting algorithm to decide the setpoints. The nine discrete light sources are to be spread around the room, with each serving the area much better than the current lighting (the closet, for example, doesn’t have any). The fans are also on PWM-able MOSFETs, and so can be throttled as load changes throughout the day. Temperatures of each power diode, the ambient air, and the mean heatsink are all read by the microcontroller and fed back to the master wirelessly, and aid in keeping the fans at the right power setting (generally, the lowest possible setting to keep the LEDs at temperature). Other feedback includes system LEDs for coded errors (overheat, LED failure, etc) and measurement of line errors (SMD LEDs to check line status, 5v and 3.3v rail status, and a measurement circuit for incoming line voltage that also gets sent back to the master controller). Power is supplied by a 360W converter, putting out a line voltage between 10 and 14VDC to each receiver depending on load and line losses (which should be minimal with 10awg wire). The power supply is about 50% overrated for the system, and has a master relay controlled by the base station, as well as a bypass controlled manually (both the shut the receivers down, as those line status LEDs would be on all night otherwise). The base station runs the GUI, which presents graphs of all possible data for the past 24 hours, and logs all data in larger intervals. It also hosts the master receiver, RTC, bypass control board (for manual adjustment of brightness/temperature/fans/power), and voice shield (which is to be the main interface with the system). The base station is the last component to be developed, i.e. still on my to-do list; but it will get done!
The “Glamor shot” of the finished device, the capstone of my Junior year summer in high school. Here you can see the user interface (three buttons), emergency stop (for when it inevitably caught fire), oven with temperature fuse and thermistor, power supply, and refrigerator. Also visible is the conveyor platform, halfway between fridge and oven, showing how you would pick up your cookie.
Opposite view of the first, with top lid removed. Now visible are the fridge internals (temp sensors, laser diodes, selector carriage, and tray holders) as well as the actual refrigeration system (two Peltier plates mounted between fan/heatsink combo on inside, custom water cooler blocks on outside, coolant radiator with fan array, insulation). Much more of the electrics and support systems are also visible (the control board housing on the left, bell to announce cycle completion, power supply and signal wires, mounts).
An early view of the interior of the fridge: dispenser tray is installed and visible, selector arm is visible along its carriage behind the tray assembly, Peliter cold-side heat sinks are visible before the fans were mounted and thermoformed to them, and the shell can be seen before the insulation was applied. The conveyor swing door and track can also be seen, without the lead screw installed.
One of the workbenches as it normally ended up after a days work. Tray assembly, fridge wall components, and a couple test systems can be seen.
Microvendor - which is what this thing is - was a project my friends helped me realize before our senior year of high school. They humored me by helping out almost every day for a month before we realized it was far beyond what we were capable of building. Still, we kept at it, and the self-contained kitchen, the size of a microwave, baked its first cookie to perfection the last weekend of summer.
It was goverened by an Arudino Mega, which monitored and controlled the solid state refrigerator, oven, cooling reservoir, dispenser trays, user input, and power consumption. Owing to budget contraints, everything had to be custom made, giving us excellent practice tolerancing lead screws, calculating thermal and electrical loads, an designing fail-safe systems so users would get cookies and pizzas, not shocks and burns. Ca. 2012
Driver’s view of the steering wheel, with blue-illuminated paddles and buttons.
Cleaner view of wheel, with modifications now visible as sandblasted aluminum paddles/armatures.
The original board layout, done largely during AP Calc, before I knew how to use Eagle. Still, learned a lot from having to do ripups manually, and the board ended up working without ever failing!
An early schematic for mounting the mods to the wheel (have to run cables without impeding wheel turning) and layouts for the paddle hinges (moved in two dimensions, and had to make contact with switches with just the right ‘feel’ to them).
The original pinout, probably also done during Calc, to make sure the d-sub connectors I planned on using would be enough and that the microcontroller could handle it.
I inherited a bright red VW New Beetle from my sister, who had it passed down from my mom. When I got it, it was missing two hubcaps from Ohio State frat guys, had a manual tran I didn’t know how to work, and was pretty small for the hardware runs and mods I thought I was going to be doing with it (I thought I wanted to drive our truck - man I was naive!). Clearly I came to my senses, and realized this was easily the best car in the world.
Once I got past just how good it was to drive, as it really had its own spirit (read as: “it broke a lot”) I found out how much potential it had. There was tons of extra space in the engine bay, and convenient channels under the body… Since we didn’t get heated seats with it, there was extra space in the dashboard cavity, and the steering wheel was practically designed to have paddles mounted on it! Plus the knee guard gave easy access to the engine firewall pass-through and fuse box, plus other warranty-voiding peripherals. All told this car taught me a lot, and ended up the most successful set of projects I’ve done yet. Oh, and the curved rear roof added a lot of extra power to the subwoofers.
Inventory of systems I added: old-fashioned “ooga” style horn in engine bay; RGB LED underlights, with independent strobe controller; 1kW custom sub box with 2 12” woofers, ported and tailored to fit in the trunk. Audio input wired into master sound system, with bypass switch to disable. Tunable through car audio system. Power cables were woven inside the body panels and under rear seats. Whole sub box was press fit by rear seat instead of mounted to eliminate rattle and still allow fast access to jack and spare tire just in case ; steering wheel paddles, wired through control box to coordinate iPhone functionality and car audio/light systems. Paddles handled input volume and song skipping (next/prev) through two-way hinge system, lower buttons had sub power, ooga horn button, Siri/home button extension, and underlight remote power switch; control box took in steering wheel signals, parsed them into the different button combinations for the iPhone headphone controller, and sent them back to my phone. The output was then passed through to the car audio internally, eliminating wires and enabling playing from 3.5mm jack (previously didn’t have one). The lights were run through a SSR and 555 timer as controlled by a pot mounted in the dash, system power was made to look like an OEM switch by the heated seat controls, and accent lighting brightness was controlled by a symmetric pot running an LM317 and outputting to all the blue lights around the buttons. It was a great car, and really too bad that I had to get rid of it - there’s no place or reason for a car in Cambridge, and even though my parents loved the subwoofers, it wasn’t worth it to keep around. Someday, though, I’ll have another bug!
The glamor shot of Mark II, designed, built, and tested the second weekend of the competition. Here it’s shown on a pretty creepy mannequin hand (and all the hardware or whatever).
The first iteration from MakeMIT Phase I, it's clear why we needed to focus on ergonomics for the second version. Still, this was an effecctive proof of concept and performed admirably for the limited time, resources, and engineering put into it.
Our display in the MIT Museum studio off Lobby 10! Complete with a vignette of demo pictures and descriptions playing on a loop. Shows both Mk I and II, and extra hardware from each.
During the final presentations of Phase 2 - what better way to show its ergonomics than to put it on and take it off during the demo! On the right is Mk I, plugged in to show actuation.
The first sketch from the first day of my first idea of how this would work. Showing the top and side views of the integrated bias tensioner, braking mechanism, and cable spool. From here it went into SolidWorks, then off to be laser cut (acrylic & ABS) or 3D printed (nylon-type plastic).
Three of our team this year also did MakeMIT last year (its first year) and walked away with nothing to show for it - we had a great time, and worked ourselves silly, but we didn’t have any expectations for greatness going in to this round because of it. We had a vague idea of what we wanted to do, and really just a group of folks with an extra Saturday to burn through. 18 hours later though, we walked away with Oculus DK2s, prize money, and an invitation to come back the next week - we got 2nd place!
The Force Feel’d is a force feedback glove worn on the forearm and hand that provides tactile feedback to VR environments. Specifically, it provides safer and cheaper feedback than existing tech. Rather than servos that are rigidly connected to the user’s fingers, or cables held tense by linear actuators, we opted for a fail-open system that never transfers force to the user. We used fishing line connected to finger grips as a flexible but taught transfer cable that stops the user’s fingers from extending beyond the object as perceived in VR. There was a constant, slight bias tension placed on the line to keep it taught, but otherwise no other force would be applied to the fingers. Instead, the cable was spooled up on a drum connected to a braking mechanism, preventing the finger from going any further forwards when the brake is actuated. In this way, even if there is a catastrophic failure in the control system, the user is never in a position to be harmed (unlike direct actuation systems, where an errant servo movement could torque a user’s fingers in awful directions).
Finger position was captured by a Leap Motion controller, fed into a custom VR environment, and then rendered into a virtual hand. The mapping was quite accurate, owing largely to the Leap’s precision, and provided enough detail for us to model different types of blocks to play with: some solid and hard, others slightly compliant, and some rigidly attached to the floor - then all these could be made tangible through our system. The full user experience included our glove, a Leap Motion controller, laptop, and Oculus Rift DK2 to give a totally immersive environment to the user. Then, by varying the exact position of the brake, we could change the finger resistance from only the bias tension (quickly ‘learned’ by finger muscles) through viscous damping for a spongey block, to a hard stop for solid ones. It was a very neat demo, and worked surprisingly well for the 30-some hours we put into it total. It also set the stage for next year - we got second place in both rounds of the competition, next year we have to do better!! Hopefully with a team as good as this year’s (that means you guys, Ever, Po, Meghana, and Josh!)
A rotated view of the control pedestal, wired up and ready to go. 6v and 12v bus bars can be seen in the control panel, and the switches up top that moved each motor forward or backwards.
Above is a robotic arm made for my school's Science Olympiad team: 4 DOF, 2 end effectors (grippers and an electromagnet), and a control pedestal - on wheels!
That last bit was especially important, since the thing ran on a car battery, plus two other SLABs. All told, it was easily 100lbs wheeling around Ohio and various host schools. It did quite well in the competitions, considering its operator (me) and was a good opportunity to design linear bearings, controllers/dampers to prevent overshoot (shocks to the system) and balance forces to make the 'bot fit in the start box, then unfold and complete all its tasks. Ca. 2012
Test stand for the most “advanced” engine I worked up to, showing oxygen cylinder, extinguisher for after the run, kerosene tank and pressure line, and needle valves to adjust propellant flows. This stand would be mounted on a test bench, put 50’ away from anything important, and then lit to make an awesome maelstrom of flaming kerosene in my backyard. It was easily seen by neighbors driving by, but never once got the attention of the police (kind of troubling, now that I think about it).
This is the first iteration I put together just after I pieced together my mill and found a better supplier than Amazon for metal stock. It ran on oxygen, propane (to preheat) and isopropanol (to run) and was so much more containable than the previous version that I ran it in the garage without much worry. OD was 1.25”, ID was 0.5”, nozzle major dia was something like 5/32” if I remember. It made a very nice, cute blue flame out the back, but I was never confident enough to run it very long, as I didn’t have thermocouples to keep track of the throat temp (I didn’t want to melt the aluminum).
Not mentioned in the text above - there’s supposed to be the fuel injector right after this assembly, but the last test I ran with it ended in weird and unstable combustion, and very drastic pressure changes. After ramping down and extinguishing the setup, I took the chamber off to find the injector missing - that’s what caused all the malfunctions! I couldn’t find where my gorgeously machined steel nozzle had gone though.. There was only a little blob on the bottom of the chamber with some stainless steel screws sticking out of it - the chamber temp and runtime were enough to melt the injector right off the stem! You’ll have to take my word for it that the injector was nice, though. 8 micro holes and two pieces of steel faced, lapped, and screwed together since I couldn’t find a sealant rated for those temps (I thought the kerosene would help cool it down a little though - sheesh!)
I am far more passionate about rocketry - or rocket components, anyway - than lighting or audio or robotics. Unfortunately, though, suburban Ohio isn't terribly friendly to extremely loud and very large fireballs in backyards - let alone Cambridge! So rocket tests are few and far between, but still loads of fun. These are examples of my low-budget progression, from designing the test system plumbing and electrical(above) all the way to a working setup, on the title card. It was a great showpiece! I only wish I could find one of the videos of it, and that it doesn't get shown to my neighbors back home... Ca. 2011
For a project class this semester, I was the only one with practical electronics experience and so I got to revamp a board leftover from the previous group. It’s a power converter from a small (30W) fuel cell to a stable DC output, with some special features like maximum power-point tracking (MPPT), shorting/purging the fuel cell, and talking to a Marine battery over a “smart battery” bus.
This is a typical setup I used to diagnose the board, which still doesn’t work, to see what needs work first. It never got to the set output voltage, drew far too much current for the output, and drastically overheated a few components.
This is a close up of the overheating components, namely the undersized inductor “LR 4R7” at the top. It was also far too close to the other components, and had an I2C line running underneath it that was undoubtedly picking up noise.
This is a thermal image taken during normal or “well-behaved” operation (since it still wasn’t really working…). 111F is barely even noticeable as hot, so this was acceptable.
In a subsequent test the temperature would spike and come close to reflowing the solder on adjacent components. Here it peaked at 505F, but in other tests it would get to ~560F before I could turn off the power supply.
For some other students in the same CSAIL lab I’ve worked in for a while, I made industrial “accessories” to help with the main project. These first two are forms out of aluminum to hold plastic parts while we were thermoforming them into front panels. The orange plastic was to check that the form allowed for shrinkage, held the pieces well enough, and let us work around it. The paddles on locking handles would eventually be covered with adhesive-backed Teflon film to help release and protect the plastic from marking. It was giant but very fun to make and worked well – until we changed the front panel design again…
The next project was an antenna positioner for another grad student. He needed a surprisingly accurate sweep, with very little play perpendicular to the travel. It was a great project, because his constraints pushed me from just making a simple belt drive to actually finding a cheap linear positioner and build the support materials for it.
The solution was a stepper-driven lead screw style platform. It only came with the mechanicals and a motor, so I had to find and size a driver & microcontroller, install and incorporate limit switches, and write the code so he could seamlessly run tests, without worrying about my positioner. The final setup wasn’t very pretty, but most of the tape was gone and the black plastic top plate worked very well with his antenna mount. The system worked well, and I believe he’s just about to publish his findings.
My roommate from freshman year upgraded his room while I was gone over the summer, including a nice loft with a swinging ladder to keep it out of the way. This piece was an extremely over-engineered, over-built, and over-praised latch to hold the ladder closed. It was 2” square rod cut and drilled to hold the pin, which has a screw tapped into it as a handle. The rod and hole were polished and lightly oiled to slide easily with minimum play, since that play would translate to his ladder shaking when he climbed it. The slot was cut in two milling operations on a manual Bridgeport. Then the mount holes were drilled, the whole thing sandblasted, and the receiving part drilled & sandblasted. To attach to the ladder, I waterjetted some aluminum plate to that it would fit, sandblasted the couple pounds of aluminum, and screwed it to the loft. It works really well, and still seems overdone!
The finished surface of my new dorm room desk - MDF on top of the substructure, sans any fasteners. Instead it’s held down by its own weight onto caulk, which prevents squeaking or shifting, but also holds well and doesn’t build up stress under deformation. This also allows for a perfectly clean top and easy replacement ($25 for an entirely new desk surface - not bad!).
The finished substructure, showing ties between supports to add rigidity and support for the rather floppy MDF.
Just the supports, with ties waiting to be mounted. Each support is drilled into the wall with overrated concrete anchors, so the limiting factor of load rating is the pull out strength of the wood (not the fasteners, all those are fine). Like I tell people - you could probably sit on it, but I’m not about to try it with my computer and everything back on it. After all, you can do anything ONCE…
Here is the whole setup, computer back up, subs from the bug in place, overhead light modulues ready to be programmed - all in all, much improved over the previous setup. Plus, the reat of the room layout benefitted too!
This is a very different desk that I made a few years ago for one of the collaborators on Microvendor - Andy - who ran all our software. He needed a new desk, and I told him I would love to help make something more future-proof (and way cheaper) than anything he could find online. We brainstormed for a bit - originally wanting to make an edge-lit plexi top - and finally settled on a simple black stained top (one of the first times I really finished wood, and it worked surprisingly well) The table can be reconfigured into a ‘L’ that fits in a corner either way (long side vs. short side) or it can be pulled apart and used as two standalone desks with four legs each. The surface is maple (I think?) with black stain and 4 hardcoats sanded matte to provide a good, non-impressionable writing surface. From what I hear, he still likes and uses it today!
Any good project starts out with overthinking, and my latest loft is no exception. Step one is always to make a detailed room schematic with accurate dimensions, so drawings are to scale and it's easier to calculate materials later
A few mintues after making that drawing, and bingo! New loft, all done. Ok, maybe it took a little while longer, but the more iterations I go through, the faster each feels.
Beyond just this one, I’ve also made one for my brother Ben (his was a prototype for mine, using compression fitting to tie into the wall. Non-marring and prevents shifting!), and one last year for me, my roommate, and his girlfriend (though originally it was only designed to handle 2. Safety factor might have been stretched a bit…). That one was recycled to use this year, after refurbing many of the parts. It’s held up very well to this day! I’ve also lent tools, muscle, or experience to many on my floor who were building or designing lofts, and to my friends back home. They’re great challenges and good ways to practice engineering for cyclic loads over huge timeframes. Oh, and you better make sure they don’t fail!
Detail of a few of the 120 or so SMD LEDs I hand soldered before I had any real means for SMD processing. During this time, and largely encouraged by this project, I was working on making my own reflow oven (though that never came to fruition)
The full view of the business end of this light unit; 5 different branches PWM controlled through MOSFETs from an ATTiny84
The controller board for an even higher-power array, with RFID verification to make sure my roommate wouldn't turn it on and blind me in the middle of the night. Shows Arduino Nano socket, dimmers, and LM555 circuit.
My first polished trial of using cheap 10W LEDs from China, this array of 6 runs off 12v and has a fan mounted under the heatsink to keep everything cool. The heatsink was so perfectly shaped for this setup I couldn't help but thermal-glue 6 of them on.
Side view of the same array, showing heatsink, wire routing, fan, and hardware.
“Periodic cupcakes” made with Julia and Marissa for a Science Olympiad team dinner. Laid out on top of the same honeycomb cardboard used to make the canoe! They tasted great, maybe that’s only because we were icing them for so long without getting to eat any though…
Supercapacitor flashlight I made for my dad after realizing I could fix the issue he always had - just needed a quick flashlight that’s bright and durable (not an iPhone) that could actually charge quickly! This is just the schematic, the real one was built and worked, but then I scrapped it a while later to reuse the supercaps… Totally worth it (sorry dad).
One of the earliest projects I have well-documented, this was also my first extensive use of a welder.
Made to hold another gift inside, this was the best thing I've ever lasercut - it all fit together on the first go, and had a sliding top lid that was routed out (top not shown because it has the recepients name)
My favorite teacher, ever, has an ongoing rivalry with his brother competing in their sons’ pinewood derby races with their own cars. Once he knew he had a resident engineer who didn’t mind blowing off homework, he asked me to make a car that could knock the socks off his brother’s in a ‘renegade league’ for the parents. I took the drive motor from an RC helicopter and geared it down slightly onto the rear axel of his car - then pumped 200% more current through it than it was rated for. It collected hair and dust, though, and eventually overdriving took its toll on the motor, such that it didn’t actually beat his brothers EDF powered car… BUT it still won a best in show for the LEDs on top, headlights, and control system that far exceeded any others. It sensed changes in the cars orientation from an accelerometer and throttled the motor based on whether the wheels were slipping or not, and had a pre-calibration cycle for each run (which was a convenient time for a lightshow). After the embarrassment of only beating a standard car by a hair, I’m spurned on to go even faster - and ultimately beat the world record. This may very well be my thesis project someday. More power, more controls, and more data!
The test cell control panel, where all data is collected and processed and all timings, temperatures, and ratios are controlled.
The test engine, now one of two in the same test cell, with gas supplies, thermocouples, and electronics wired up. To the right is the instrument tower, with controls and signal processing equipment.
The top of the modified 4cyl diesel engine, where you can see the topmost cylinder has its diesel injector replaced by a sparkplug and pressure transducer.
The old intake gas heaters, now replaced with a more efficient and less leak-prone system I helped design.
More details to come once I figure out exactly what I can and can’t say! Until then...
Sloan Auto Lab Responsibilities: Assist graduate students and researchers in designing and completing experiments on a diesel engine running one of the cylinders experimentally to produce syngas; repair or improve the engine platform and test cell for reliability, maintenance, and better functionality.
CSAIL Responsibilities: Assemble and give feedback on iterated PCB layouts for different subsystems of the research project; design mechanical cases and component layout for wireless charging receivers; design, assemble, or repair mechanical supports and cases for novel wireless charging controller.
The first high-powered light I made, three zones of SMD LEDs running off and 18v tool battery from the shop. Also had a fan to help keep it cool. Still works to this day! Though some of the LEDs burned out (the ones you can see are up off the board a bit didn’t get proper cooling). FUn project, and great intro to high power lighting. Also served as the test platform when I made a strobe for the first time, which led to its integration to my car modifications.
This torch was made just a few weeks ago when I was last home. My friends were over (Matt, Zach, and Cameron - remember this?) and we needed something to do, and I still had plenty of MIG wire and some oxygen left in my torch’s tank, so we bent some steel rod and welded on a support and bolted it to a cleaned up (then burned, for effect) piece of wood. Dip some shop rags in kerosene and ta-da! A torch Indiana Jones would be proud of. The sound when it’s swung around is fantastic.
This torch was made just a few weeks ago when I was last home. My friends were over (Matt, Zach, and Cameron - remember this?) and we needed something to do, and I still had plenty of MIG wire and some oxygen left in my torch’s tank, so we bent some steel rod and welded on a support and bolted it to a cleaned up (then burned, for effect) piece of wood. Dip some shop rags in kerosene and ta-da! A torch Indiana Jones would be proud of. The sound when it’s swung around is fantastic.
This is the bow of a cardboard canoe I made for a group project in AP physics sophomore year. It’s not done - promise - but when it was it easily sailed across the pool and held all three of us (four? Don’t remember). It lived and died valiantly, ending with a recreation of the Titanic (I was the iceberg) where we ripped the hull in half using this steel cable. It was a fun day, and I hope I can find the video for it again!
One winter I got cabin fever pretty bad, and so decided to save up some money and make a zipline from one tree in our backyard to another (about 200'). It still works today!