Engineering
Noribo
2.009! Senior Product Design Capstone! Woah!
Okay, that should convey the standard excitement levels for 2.009 here at MIT. Big course, big ideas, big teams, and a very unique experience. The premise is to get a team of 20 MIT students together, give them $6500, and see what they can design and engineer over the course of a term. Results are always interesting, and the end is a culminating presentation that had over 1000 people present in MIT’s largest auditorium to review and critique your project. Beavitvl.
The theme of the year was “food”, very open and with endless possibilities. Our team went with building a novel way to deliver food in restaurants, in this case a small swarm of robotic plates that would ferry food from the kitchen to you. They had several interactive features, most importantly a dance they would do when a plate was removed from them. The idea actually started as a robotic coaster for delivering beer in bars, seen as a way to alleviate the pain of waiting for a bartender in a packed bar. With time the idea muted to all-purpose food ferrying, so it goes.
Our robot in an exploded view.
The bot was a fun design project. I was in charge of bringing all the disparate parts together into a final package, and given that a low profile and small form factor was key to success there were many challenges in packaging everything so tightly. Not to mention all the design for manufacture we had to include in our build. Still, at the end of the day I brought a lot of innovations to the table when parts didn’t fit and made sure all the layout worked well. Beyond that, when a problem arose, it was often at the intersection of two or three team’s parts, and I found myself being the liaison between designs more often than not.
Our finalized bot, looking something like a UFO
A depiction of our bots in action
The bottom of our robot
Our bot with a very adorable sushi plate on top
Our bots went through quite a few design revisions, mind you. There were basically three stages. During the mock-up, we threw together a bot in an inordinately short amount of time, having a radio-controlled bot with the basic form factor done in some 5 hours one Saturday. Surely a testament to the rapid prototyping skills of a modern caffeine-injected MIT student with a nice machine shop at hand.
The first bot had an acrylic shell that we laser cut and assembled with Charles Guan’s infamous “T-Nuts”. Powered by an Arduino and a random smattering of battle bots components, it served bravely, but was dismantled to make the second iteration. This guy featured a 3dp chassis with an acrylic top plate that was etched to produce interesting different designs. It also featured LEDs for different underglow colors, which the judges loved. Guidance was provided via wall-following with IR sensors, the idea being that it could follow a small lip placed around a bar.
[Image]
[Image]
A really fun project, a lot of good memories, and a experience I wouldn’t pass up. Thanks Prof. Wallace!
-Cd
ExoSense
One of the projects I’m working on currently is to develop a new set of senses for people. The field’s loosely called “exosenses”, and right now we’re flying the idea of developing new senses using tactile feedback and a smartphone, immersing the mass of digital and networked information available through feedback. With time this feedback becomes integrated, the stimulus becoming a new independent sense as real as any other.
For now this is taking the form of an Android phone talking over 3G with Google Maps and interfacing to the body with a set of actuated solenoids on a belt. Drop a waypoint or tag a reference, and Android constantly gives you feedback on the heading of the marker. All without ever looking at a map or thinking, just feeling through your new sense.
The Arduino, compass, and demux for actuating the belt
FSAE Engine
A large image dump while I had our engine block open. What a beauty!
The engine is a CBR600 F4i model, though heavily modified for our use. The crankshaft and flywheel have been lightened by four pounds, the pistons are high compression (14.5:1) from WiseCo, the pistons and head have been coated with a high performance, low thermal conductivity ceramic coating, and first gear has been locked out to make finding neutral easier (first gear is useless as the car is so light that the wheels just stay lit in first).
She’s super sexy. Once everything’s back together, I’ll post pictures of the dyno readouts and let you know what kind of specs she has.
Till then, enjoy the pics,
-Cd
The open crankcase with our beautiful lightened crank all hanging out
A shot showing all the power components in the crankcase
Close-up on the crank with connecting rods and polished journal bearing surfaces prominent
Lower half of the block with journal bearings exposed
Combustion side of the head, showing the spark plugs and valve surfaces
Our ceramic coated pistons and cylinders, with a dirty water jacket surrounding
The head of the engine with intake ports covered in tape to prevent dirt from getting in. Also note the valve buckets on top, which get hit thousands of times a minute by the cam and wear very little
The cam carrier for the head, journal bearings evident
Our homebuilt dynocell, using an eddy brake. Note the soundproof, explosion proof walls, built by your very own
The alternator for engine, showing the three phase windings
The clutch pack, which transmits power from the crank to the transmission via the back gear, through the friction discs, and into the center spindle
The computer to run the dyno cell
Diff Brackets
A crucial part of any racecar is having a powertrain that won’t explode on takeoff. Designing this requires some really careful attention to detail and ensuring that both the models and the simulations aren’t lying to you as an engineer.
These new brackets hold the differential to the car. The diff receives load from the engine via the drive chain and applies it to the two rear wheels, all the while allowing some relative slip between the two wheels so that turning isn’t impossible. I’ve designed these brackets to be made out of 7075 Aluminum, a high strength, low-weight alloy. The design is based off a concept known as “trussed webbing”, which allows for high strength with minimal material by placing most of the material in only the load bearing plane. To prevent buckling from occurring, trusses are implemented to give strength to the thin webbing elements.
I first modeled the bracket in solidworks, generating the beautiful stress profiles seen below.
Solidworks Simulation FEA of stress profile in a loaded configuration
Many, many design iterations later, a successful model was developed and Keith Durand, the grad student responsible for FSAE even happening these days, machined up the marvelous little beauties. Here they are in full glory.
A view showing the finished brackets with trussing
Here’s an image showing the clamping zone up close. This region’s fascinating as the inner circle is actually a clamp that holds the bearing on the differential. The small hole below holds a nut, and a bolt comes from the end, squeezing the circle and pre-loading the bearing properly.
The neat technique used to pre-load the diff bearings
Up close shot showing the webbing. The low material is only .095" thick!
The back side of the bracket, showing the flush face.
3-Axis CNC Mill
The Epic 2.72 Mill
2.72 holds a certain special place in the heart of anyone who’s ever taken the course. Stockholm Syndrome, some would call it, but I prefer “Survivor’s Glory”. The course centers around the design and build of a precision desktop lathe using parts and a budget provided for in the class. Marty walks you through a lot of design steps and labs, and every two to three weeks your team meets with him personally in design reviews.
These events are always exciting. I remember that for the first design review we got an email two days before saying “none of you have been in contact with me about what exactly I want to see in the first design review. Interesting, given that I’m your boss here and I expect certain elements to be complete. Further, I refuse to accept incomplete presentations, and will require you to push back a day. Each day you push back removes 75 points from the 100 possible design review points, and do note that negative scores are possible.”
No one had completed what was necessary to present the first day.
We came in on the second day and showed off our work, which Marty was quite pleased with, but regardless we walked out with 15/100. We all hoped these were “who’s line is it anyways” points.
Our team significantly differed from the standard 2.72 team in that instead of designing a lathe as was typical for the course we opted for the alternative design goal of making a 3-Axis Milling Machine. The specifications weren’t clear, and we were forced to execute a very open design loop in order to decide on what our final capabilities could be, and from there work back to the actual machine design. The experience was quite worthwhile, and more to follow on our actual design decisions, but for now these images will begin to convey the work we put into building this monster of a design project. And remember, we not only made design decisions and modeled each component, but also machined and assembled everything here.
The 18V Dewalt drill motor and belt drive assembly
A good shot of the mill bit (3/8" 2 flute) and work table, with a "sacrificial" piece of Al bar stock on board
A sample light cut (long traverse) and a full width, high speed, high depth-of-cut (jagged death-looking traverse)
The mounts for the two tables, with flexures visible
Our special flexure for the lead screw thrust nut. Works incredibly well, allows around 1/8" flex in any direction, but holds several 100lbs axially
Our vertical bed translation axis, using a 6in main support column and 5/8in drill rod for translation.
Kinematic Coupling joining our spindle to the support neck. Theoretically allows us to change to other head attachments while maintaining endpoint accuracy
1972 Yamaha RD350
This bike was the love of my 2010 summer while staying at MIT doing undergraduate research. I picked her up for a dime when it looked like a mix between a vintage tracker and a kid’s dirtbike. Terrible shape, really, but a diamond in the rough if you know how to hold a wrench right.
I spent quite a few nights laboring long and hard to bring out what I wanted to see in the bike. She’s almost there now, and the work I’ve put in has completely transformed the look.
To date I’ve:
- Replaced the rear shocks with some beautiful Marzocchi Strada Ducati adjustable shocks
- Found a new side cover and oil tank with the retro R5 chrome accent
- Picked up a set of cafe handlebars
- Bought a new cafe seat
- Installed reasonable (not-70′s-enormous) turn signals and brake light
- Completely tore down and rebuilt both carbs
- Added a new four piston Tokico front brake caliper with custom mounting bracket
Have to say I’m pretty proud. Next up? New seat, Aluminum this time, and maybe some head work to boost performance a bit.
Full on of the bike's left side.
Nice four piston Tokico brake caliper and custom mounting plate
The new Strada shocks. So nice
Rapid Protocasting
Today was, among many reasons, an excellent day. After hardly sleeping, I awoke strikingly early to a warm spring morning, and had an early meeting with Hugh Herr, leader of the biomechatronics group over at the MIT Media Lab, where we talked about some work I could do for him that might be of some mutual interest. I then immediately picked up some books for my American Revolution course, skimmed the relevant passages, and felt I knew enough to head to class.
And then promptly skipped the first half hour to do something AMAZING. I’m taking 3.042 this term, a course focused on working an independent project as if you were an internal project in a company. The project I’m working on is using 3D printed concrete to create molds for metal casting. The problem, which you will see after thinking for a few, is that concrete melts at the temperatures of molten steel or other alloys. This makes molding less of a useful thing and more of an exercise in futility.
Nonetheless, I skipped class to do the first step of our experiment for the day; a real live steel pour. The tech my team is developing, see, is a ceramic slurry which can be slip cast onto the concrete mold. As the slurry is poured into the mold, the concrete sucks out the liquid, pulling the ceramic particles in the slurry onto the surface of the concrete. Once a thick enough layer is built, the remainder of the slurry is poured out of the mold, leaving a thin particulate ceramic covering the mold. Firing the baby in a furnace sinters the ceramic, and a beautiful, thin, hard, high temperature coating is left on the mold. I did this for about 6 different test pieces, using different soak times in the slurry to build different layer thickness, leaving several nice looking molds with our most successful ceramic layer to date on the surface.
I went to the rest of my classes and the magical hour of two finally came, when 3.042 starts and we could finish our experiment. Now that we had a few sample molds with high temperature coatings, we did the only thing you can do with such a trite object, which is pour many pounds of 1600 degree Celsius brilliantly bright steel into them and pray the molds don’t explode. The pour went fairly brilliantly, and we should have fairly good baseline data to push our experimental exploration in a solid direction. All things considered, an ace day for an ace project.
Stay tuned, soon we’ll be CADing up all kinds of crazy shit just to make molds and cast it in various high temperature alloys. Flywheels, hands, pretzels, whatever we can think of, we’re going to cast it and see what happens! Whoo, Science! No, not quite, ENGINEERING!
Till then,
Cody Daniel
Molds ready for casting
A microscope image of the ceramic layer we use
Here's a cast impeller using our process
An example of a cast flywheel we did
TV Oscilloscope
A quick update on a random and fun evening. A hall mate and a few friends from around Senior Haus showed up to tinker around with a TV we were disassembling. One cool thing found online to do with a spare CRT is to drive the deflection coils with audio frequencies and make a cool visualizer out of the thing. We did just that and had a grand old time messing around with Audacity and some music files.
The basic premise is pretty straight forward. Take a TV cover off and check out the big tube inside. Be careful not to break the tube (it can implode) or to short the capacitors on the flyback circuit (the one with the transformer and a line running to the back of the tube, where the electron gun lives) or touch the anode running into the tube near the screen, with the big high voltage line and the rubber coating and what not. Now that you’re looking at the tube, find the terminals for the deflecting coils. They’ll look like this:
Close up on the deflection coil terminals (Large posts with green wires attached).
These coils control the horizontal vertical deflection by generating a magnetic field and literally pushing the focused beam around. Now that you have these terminals identified and some leads so you can do something fun with them, grab yourself an amplifier and run the leads into the output posts. Join some speakers into the same posts, and find an input source for your computer. Here we’re using a spare low quality amp and small computer speakers with a USB DAC hooked up via RCA, but an 1/8″ to RCA wire would work just fine as well.
With this, you’re ready to drive the channels with your laptop and some nice free software like Audacity to make some really spectacular things happen. By driving the left and right channels, you can independently drive the horizontal and vertical deflections of the beam. So you can do something simple like have the left and right channels be driven by a track, and you’ll get circles for in phase balanced music, and fun shapes otherwise. Even more interesting is to drive the horizontal axis with a sawtooth wave and the vertical with a signal, say a square wave, and generate an oscilloscope image of what you’re dealing with. Neat! You also still get music output from the speakers so the whole thing makes for a cool system visualizer that is live and a bit different from what you’ll normally see. When you don’t drive it, the beam sits centered on the screen at fairly high intensity, meaning the phosphor will be burned out fairly quickly. In fact, the TV screen isn’t designed for this heavy of a load, so you’ll burn the phosphor fairly fast this way. But CRTs are old and cheap, so no big loss.
Below you’ll see a picture of my hall mates watching the image kick on, the initial stand still beam, and a tracking sine wave. Try it out!
Electron Beam sitting still on screen center.
Electron beam etching a sine wave across the TV Tube.
Take care and enjoy kids!
Formula SAE
Ohh racecar. The Formula SAE team at MIT, otherwise known as racecar, is a second home to me. We reside in building N52, behind the MIT Museum, in a shared machine shop space with several other MIT vehicle teams. I’ve slept in that shop, BBQ’d for the masses, and executed more mad engineering over all nighters than any other place on campus. It’s home to me in a lot of ways.
Here are a few of the pieces I’ve worked on.
Our first render of the new MIT FSAE car
The team teaches what few other projects at MIT can, which is the necessity of quality engineering in any project. Our driver’s life lies in our hands when we finish putting the car together, and we must have complete faith in our engineering and design of the critical components on the car. If we machine a part wrong, don’t check our sums, or don’t use our tools properly, the car can quickly fail once put to the ground, and this is our shared responsibility and liability.
Fortunately the team has a lot of background in executing rigorous design loops, and at the end of the day we haven’t had any major failures to date. One of the greatest learning experiences I’ve had at MIT thus far has been building a part for FSAE, slaving in all the design and machining hours, only to put it on the car and have it fail utterly and completely. Then the fun of analyzing the failure, looking for what variable was missed in the model or what error was made in the assembly. Being in such an environment really brings out the true innovation and meat of an engineer, and in no other fashion do I think you can learn in a more genuine MIT way; really being out there using your mind and hands to assemble a very cool, very fast race car.
Various parts designed can be found throughout the website. If it looks fast, it’s probably for racecar.
-Cd