What does that mean? The end of term is just two weeks away, and for me, this is the last one. It’s Serious Business Tiem, and I need to push the usual load of grutchy and bitchy things a senior needs to do out the door before it all goes to hell. It also means this is the last summer I have to build things before it’s not cool any more*. Well, so building things is always cool; but I feel like the impact is dramatically lessened once I graduate, since then it means I’m expected to do things like that and it would neither surprise nor entertain anyone. In other words, the real world sucks, but that was an expected outcome. Therefore, I’m going to take this summer to build everything to my satisfaction before people start seriously asking me if that’s my research or something. I have purposefully made no official plans so I can have maximum project time.
That means no job, by the way.
I think I did it wrong. So…. if you need something built, let me know. I’m open to take on consultancy or contract work in the interest of getting curbstomped by real life for experience purposes, but the premise of the decision was that I’m not stuck doing something for a fixed number of hours per day.
Anyways, the three major endeavors this summer will be:
Ah, Chuckranoplan. I’m essentially starting anew in another building hobby by being interested in ekranoplani: aeromodelling. I think through the past semester of being obsessed with fly-sailing vehicles, I’ve picked up a crusty fringe of Course XVI knowledge. Constructing and testing more models would only make that stronger, and besides, I would finally get to use all those model airplane parts that I keep putting in EVs and robots for their intended purpose. Below is the design of the 1.25 meter Chuckranoplan 0004 (distinct from the more shoddily built 0004-FML) seen previously, showing my idea of what materials will be used in construction: a combination of 3d printed shells, hot-wire cut foam, and balsa panels. Oh, and a healthy dose of 70mm ducted fans.
The reason why it’s not already flying is because I’d have to build the hot wire apparatus. It can be as simple as a chunk of wood with a length of Nichrome wire on it, but I just haven’t gotten around to it. I’ll probably start with templates glued to the side of foam blocks, since I don’t immediately foresee myself building a CNC hot wire cutter. Yet.
Sadly, Chuckranoplan will remain a R/C model this summer. I wrote in a proposal to the Eloranta Summer Research Fellowship to see if I can convince them it’s worth giving me money to build a small rideable one. The answer turned out to be no. Hey, didn’t those same people reject Super LOLrioKart last summer too? Yup, they did.
Seriously, guys, what’s with turning down ambitious engineering projects? If you take a look through the past winners, a majority is arts and humanities related. Someone got 6 grand to study Walt Whitman for three months? I can only find solace in the fact that perhaps I went too overboard with their self-ascribed WOW! requirement such that it overflowed and wrapped around, thereby becoming fail. In other words, a case of not taking what they dish out. That, or they have 1. no engineers or sciences staff on the committee, or at least 2. nobody that knows who I am.
Or perhaps I was too straightforward in describing the project and had no flowery language or grandiose visions of societal reform and progress like that which is in vogue today. Additionally, my attempts to get feedback for my proposal have fallen upon silent inboxes – it would seem that I can’t even get a data point out of them to possibly correct anything I might be doing wrong when writing a grant request.
Disappointed Sigh. Well, at least Hobbyking parts are cheap.
Chuckranoplan 0004 isn’t the only model under consideration, since as long as I have parts and Home Depot has funky blue foam boards, there will be more. I’ll probably build several different wing geometries just to try them out. They’ll all be tested around campus and probably on the river once or twice.
2. 3D Printing
I want more horsepower.
In this case, horsepower means more build volume, higher axis speeds, better precision and finish, and no warping. Make-A-Bot was a fun and useful experiment in just how fast I can go from “I want one” to “I made one!”, and much of the design was lifted from existing kit machines (such as its phonetic namesake). The problem is that the current bed-style design has alot of shortcomings.
- It’s open air and impossible to enclose, and the drastic temperature changes from molten to solid room temperature ABS causes uneven thermal stresses to build up in the part. This leads to warping and cracking of large prints.
- The axis inertia is both high and unbalanced due to one axis traveling upon another.
- It’s built way stronger than needed. I used leftover aluminum plates from the robot construction effort, and aluminum is really too strong of a material for use in a process with no tool forces. The extra density also contributes even more to the axis inertia.
- It’s single material with no way to add a support material nozzle.
So, it’s time to upgrade. I’ve been thinking about it ever since MaB was finished and running, but the summary of changes to be made include:
- Full closed cabinet with internal heating. The average internal temperature of a FDM printer seems to be 70+ celsius; in other words, pretty much an oven. This seems to be essential to getting large prints to not fail due to warping and splitting.
- Gantry-type head. I haven’t yet decided between mounting the extruder on the gantry itself (ala Stratsys printers) or mounting it outside the cabinet and feeding through a Bowden cable (such as the Ultimaker, which incidentally features a unique parallel X-Y gantry). Erik von Brujin, the principal designer of the Ultimaker, actually stopped by MITERS while MaB was under construction and I got to see it firsthand, which spurred my desire to make a better one.
- Dual nozzles. Now that Makerbot has come out with a custom cartridge heater for their hot ends, I can totally put together a dual head that is compact enough to not sacrifice build space (the large power resistor heaters on the current MK5 head I’m running are just huge.
Another feature I plan on integrating into the design for future exploration is a Z-axis that is modular and replaceable by a powder bed. I want to see if I can build SLS/SLM/DMLS machine that produces ready-to-use metal parts. This implies a GIANT LAZER will be involved at some point in the future – possibly not this summer, but it is arrangeable. It would also mean the cabinet needs to be constantly flooded with inert gas (argon TIG welding gas is probably enough), and really should not be made out of transparent plastic – granted that acrylic plastic absorbs carbon dioxide laser wavelengths, which is why it laser-cuts so nicely. I didn’t include it in the list of features up there since I’m pretty sure this will end up being a completely separate machine for doing metal, but for sintering plastics, building off the revised MaB platform will be enough.
I don’t have a rendering of this machine yet, since it exists only in my mind and as scratches on a piece of paper, but just imagine MaB but with a big wooden box in place of the aluminum machine bed. And possibly some kind of laser sticking out the side.
3. That Picture I Can’t Explain
So, who’s heard of the Honda U3-X?
Yup. In late 2009, this thing took the Internets by storm. It’s essentially a balancing unicycle but with the ability to balance in the other direction. From the perspective of non-engineers or the technologically stoic, it seemed almost magical. I investigated it only a little bit, writing the whole thing off as being based off some kind of differential (based on their cute CG animations that don’t show you anything) but probably too complicated for me to make since those edge rollers move in the completely wrong direction to be a normal diff, or even a LOLrioKart diff.
Flash forward a year and a half to some time last week, when the latent curiosity was aroused once more by some friends who must have missed it the first go-around. Well, with Shane‘s help, the patents for the U3-X were tracked down, which exposed the inner workings of the mechanism.
Conclusion: Holy crap, it IS too complicated for me to make.
But I was right. It was a differential-type arrangement. The U3-X had two motors, each driving one side of the large wheel disc, akin to driving the output gears of an automotive differential. The interesting part was the use of angled 45 degree rollers to support the center “side-wheel” ring, which is the equivalent of the “spider gears” of a differential. When you have two symmetrically angled dependent degrees of freedom, you can generate a motion orthogonal to the direction the dependent DOFs are mounted on. This is the principle behind mecanum wheels:
Yeah. If you reflected that wheel vertically such that a mirror image of it existed like so:
A net sideways force can be generated by spinning the wheels in opposite directions. Spinning them both the same way would result in the 45 degree forces summing to a 90 degree one, in this case facing left or right. Now imagine if the ground were curled around into a wide toroid (donut), and the donut were wrapped around in the space between those two wheels. Now, spinning the wheels the same way would rotate the entire donut around its central axis, generating linear motion. Adding a difference between the wheel speeds causes the donut to “roll” in or outwards. This causes an orthogonal motion along the ground. That donut, when broken up into discrete units, is the U3-X’s side facing wheels.
Honda’s patents show a very creative chord-contact roller-on-a-roller Inception-like arrangement that looks like it took quite a few hours of 5-axis CNC work to accomplish. I give them credit for thinking this up, because that took a level of design skill and insight I can’t quite yet fathom. The referenced patents show the mechanism in various stages of development.
However, I don’t like that mechanism.
WHAAAAT? Stop judging… it’s not like YOU can build one!
I’ll admit my lack of experience operating CNC machines with more than 2.5 axes. My strong point, however, is turning complex engineered and machined components into junky imitations which can be waterjetted or laser-cut for convenience purposes. And that’s what I’ve went ahead and done in the Picture I Can’t Explain:
Alright, so what is going on here? All I have done is taken the 45 degree rollers and rotated their contact points 45 more degrees about the tangential (alpha) axis of the toroid that the black rollers approximate. They are still contacting the black rollers at an angle. If you have the spatial skills, you should be able to see how rotation of the silver roller will cause the black roller it contacts to also spin. If you cant, I swear I’ll produce a cute CG animation one day.
The silver rollers have all been laid out on a flat plane as a result.
This means I can now WATERJET IT! Here’s a closeup of the side roller disc with the top surface removed.
The construction will be a pretty typical layered omniwheel with the rollers set in captured axles. The middle layer is the same thickness as the axle diameter, so once the top and bottom surfaces are bolted on, the rollers are fully captured. These rollers as-designed are just plain ball bearings on a dowel pin shaft. I wanted something simple that I didn’t have to machine 144 of.
The center wheel set is constructed in the same way… and is pretty much just an omniwheel. The polyurethane rollers (which I will either machine from rod or cast) have embedded flanged ball bearings to take both radial and axial loading.
And here’s the whole thing from the side. Note that this is both a first-iteration wheel design and that I have not yet designed anything else yet. The number of rollers, diameter, etc. can all still change depending on what kind of drive system I think up. I have also not decided what this wheel is going to be mounted to.
But there you go: Honda U3-X
on a stick on a boat on a plate!
In terms of controls, I’ll probably stick with a
programmable microcontroller Arduino-based solution using four sensors – two accelerometers, two gyros, facing orthogonal directions. I’d just throw two Segfault controllers on it (and that might actually work), but for this vehicle, I want it to be able to stay in one place – an extra four degrees of freedom in this case since I have to control position in two directions. Using a programmable digital controller is more expedient in that case for tweaks and changes.
I’ve tentatively christened this project the ME2!!!X because…well, I want one too. Plus, seeing as how Mechanical Engineering is Course 2 around here, and this thing is almost stupidly mechanical in functionality, it seemed fitting.
*Okay, so I lied – I will technically be continuing in the Department of Mechanical Engineering for graduate
studies research ass-kicking this fall. So it’s not really my last term, and I’m starting to get the feeling that I’m one of those lifers that somehow find a way to stick around MIT no matter the circumstance or no matter how it pisses me off. But it means people are even more likely to ask me with sincerity, “is that your thesis?”