Woooooaaah, new stuff.
Our first lab in 2.750: Medical Device Design was a refresher in mechanical engineering fundamentals - designing, calculating, and validating measurements for a kinematic coupling.
Kinematic couplings are really cool - a simple, deterministic way to prevent things from wobbling or be otherwise imprecise in all 6 DOF. One of the professors for the course is really into kinematic couplings, and so of course we the students needed to share in the joy by making some ourselves!
I have to build more gift-pens anyway, so I thought it would be cool to magnetically connect the two halves via a kinematic coupling using buckyball-magnet-toys.
We'll see how my design performs in real life... in ~3 weeks :P
More things I've been working on but hadn't gotten around to documenting: Ripstik is still under construction!
Last time I had gotten to the test-riding phase of Ripstik build and had realized it was just too tall to be ride-able. So began a long series of modifications to the casters.
The original casters needed to became ~0.5" shorter, so I removed the intermediary sections of the caster. This, however, made the casters no longer clear the chassis when rotating further than 45deg. This was a problem solved by a combination of milling/sanding the aluminum wheel-arms slightly narrower and rounding-over the axle bolts.
Similarly, the chassis also needed to become shorter, but the locations of the holes for the screwblocks force me to either cut very little or cut a lot. I'm still deciding whether I would have enough space for batteries if I shortened the chassis a full inch.
Finally on the mechanical side, because I shortened the casters, I ended up shifting the locations of the wheels farther back than originally planned. This made the sweeping motion awkward, so I had to drill adjusted axle holes.
Finally, for 6.131 Power Electronics Lab, I started working on a prototype for a 3-phase sensored-brushless motor controller. By the end of the class, I had behaving hardware but not-quite-working software.
Above: brief calculations for creating specs and designing the motor controller.
Below: circuit diagram
I built a kayak last spring and took it to California and back!
This photo was taken in Steven's Creek Reservoir, just a little bit west of Cupertino.
Here it is, partially loaded up for a day trip. My kayak seems to do best at a total load of 150-160lbs.
I'm really pleased with how it turned out - after two excursions on the Charles River (Boston), it was successfully packed into a luggage and checked onto a plane to California where it has experienced two freshwater voyages and one saltwater trip. Now it's stowed in its carrying bag (a golf-bag dust cover!) back in Boston!
The prompt for this project was a challenge from MIT's Chancellor to "make something creative that crosses the Charles" by May 7th, 2016. Students and MIT affiliates could apply for funding to make these projects happen, and also would be able to participate in the MIT Moving Day Parade!
This year was the 100th anniversary of MIT moving from Boston to Cambridge, and MIT would pay me to make a fun kayak. Wooooo!
I decided to make a modified version of Tom Yost's skin-on-frame touring kayaks, where mine was going to be slightly shorter (for better portability). It turns out, folks, that at 13' kayak performs so much better than an 11' kayak in terms of speed and efficiency that probably any improvement in convenience is not really worth it. (I did beat out all the 8' kid's kayaks though!)
Below are the design plans I submitted for the challenge:
My kayak used skin-on-frame construction, where the crossbeams were made from HDPE connected with tent-pole style 6061 aluminum tubes. The skin was 18oz PVC fabric (normally used for ultralight planes) and the flooring was made from 6061- L brackets and 3/8" marine plywood. All screws used were stainless steel in order to mitigate saltwater rusting.
Tom Yost has an excellent writeup for how to design and fabricate the components of a skin-on-frame kayak, so I won't go too much into detail here. My modifications slightly widened out the profiles of each crosspiece to compensate for shortening length of the boat.
This was a cool project, because I went in knowing two things:
1. I don't know anything about boats. Time to learn everything about boats.
2. Tolerances are +/- 0.5" I have so much wiggle room.
Frame building started with screwing the end crosspiece segment to a wooden frame called a strongback. The strongback helps maintain alignment and has marks denoting where the other segments should be positioned. It was also useful for determining whether the overall shape was symmetric or needed to be adjusted.
Assembly consisted of fitting tubes to crosspieces to test the snapfit, making adjustments, and using bungee cords to maintain compression while I moved on to the next segment. Eventually things stopped exploding without bungee persuasion, so I called it done.
After removing the frame from the strongback and a panicked night of heatgunning/gluing PVC fabric and zippers, the kayak was functionally complete. I poured water into the interior to see if I still had any holes that required waterproofing,
My kayak's maiden voyage was Crossing the Charles on Moving Day! The procession involved paddling from the MIT sailing pavilion to the Boston Esplanade, properly crossing back to the MIT side for the parade, then putzing around in the river with friends. In total, my time in the water was around 3-4 hours with no issues. I was pleasantly surprised to find out I didn't actually need to bring my bilge pump after all.
(Photo and Video Credit: Kevin Hwang)
As shown in the video, paddling this kayak produces a disproportionately large wake. It's clear that my modifications negatively impacted hydrodynamics of the kayak profile. But, my kayak wasn't designed for speed. It works well enough to keep up with friends traveling at a light pace (3.5-4 mph) but I definitely wouldn't be able to race anyone.
The real test of this kayak was packing it up, carrying it through the airport, and trying it out in California.
By the end of the summer, I had managed to get assembly time down to 15-20min and disassembly down to 10min. Sitting in the dirt trying to put the kayak together with just a hex key and a screwdriver was a valuable learning experience, because I quickly identified a bunch of ways I could've made the design better.
I should have drilled screw holes using a drill press to ensure holes were perpendicular through the HDPE so screw direction wouldn't matter. Currently some holes require screws in a particular orientation, which adds to assembly time.
It would've been nice to use two zippers on the PVC skin instead of just a bow zipper, because pulling the skin on like a sock is a tight fit.
Different-colored masking tapes is not going to be a permanent solution to mark which poles belong where.
Additionally, it turns out even stainless rusts after enough exposure to salt water. I will need to replace all my screws before taking the kayak out again in Boston.
This ended up being a fantastic project that I'm still enjoying long after the construction.
I'll end this post with some more photos from California.
Okay, maybe not that legitimate. But closer on the spectrum.
I started by waterjetting flat patterns from 0.5" 7075 MITERS stock,
originally from JACD and Battlebots 2015!
(they stuffed the plate in the back corner, for reasons I'll get to later)
Why five?
Two are holiday presents, but the quality of a carabiner is dictated by how it performs under load. I will analyze the performance of these carabiners by doing experiments with the three sacrificial biners, and will also use them to figure out how these things get put together.
Below is one of the nice carabiners, currently getting engraved on the newly renovated MITERS CNC mill.
Why 7075? Compared with normal 6061 aluminum, 7075 has significantly more tensile strength while keeping similar density and shear strength. This is important in a carabiner, where the strength/weight ratio means everything. Below is a comparison of the two materials, taken from Makeitfrom.com
selected graphs from MakeItFrom.com
However, 7075 only exhibits these qualities at T6 temper. The plate in the back corner was T0 (un-heat treated) and was therefore fairly useless -- it was so soft you could dent it with your teeth.
My process for taking 7075-T0 to T6 went as follows:
- Heat to 900°F for 2 hours
- Immediately quench in water until room-temperature
- Heat at 250°F for 18 hours
Proper heat treatment of alloys is a somewhat precise operation, one I couldn't achieve during this project. For starters, the kiln and kitchen oven I used for heat treatment had a large degree of error. Second, proper T6 requires parts to be strain-hardened while in the kiln but mine were not. Experiments later will show how this imperfect method compares to the datasheet.
I was aware that heat treatment would distort my parts, but I didn't expect the bubbly surface or the iridescent oxide layers that formed.
I also didn't expect cracks to emerge. Half of the parts had cracks consistently 1/3 thickness from the surface, so I attribute this probably to some defect in the original plate. I attempted to fix this by removing more material from the smaller side when facing the parts (ending up with final thicknesses of 0.375".)
Let's talk a little bit about carabiner design.
I chose an offset-D shape (the most common) for my carabiner patterns. This blog has a good explanation of the different kinds of carabiner shapes. The key parameter for an offset-D is a basket (the bigger curve) around 3x as wide as the narrow curve. This shift transfers forces away from the weak gate and concentrates load on the stronger spine (the straight part.) In addition, the curvature radius on the two sides of the D should be identical and large enough to accept the rope you plan to use with it. All other dimensions were mostly eyeballing.
Another major choice was type of closure mechanism.
simply circus.com
There are two main types: keynose (left in the diagram) and hooknose (right). Keynosed carabiners have a slight advantage by not as easily catching on things, and not failing as catastrophically if they do. Black Diamond has a good article about hook-nosed biners catching on things. So with that in mind, I decided to try my hand at making keynoses for my biners.
Fabrication of the nose and gate began with straight slots and pockets as rough approximations, then I refined the angles by filing down the edges until the gate smoothly fell into place.
Following carabiner construction, I had to figure out its spring mechanism.
Now this part was an absolute pain. I thought at this point I was on the home stretch, but ended up spending two additional days being frustrated at bent metal things.
According to the patent drawing above, solid-gate carabiner hinges consist of a compression spring (#40) housed within the gate, actuated by an angled leaf spring (#101) lodged in a notch. The assembly is wedged within the gate when the hinge pin (#36) is in place. You can see the leaf spring in a real carabiner (upper right) peeking from the gate.
Finding a small compression spring was easy; finding a suitable leaf spring was not. After much trial and error, I discovered that constant-torque flat springs didn't work, and actual spring wire (balanced under a set screw) wasn't helpful. I eventually realized that the leaf spring didn't need to be springy at all; it simply needed to have the correct shape. Whacking and grinding small finish nails into shape successfully completed my artisanal hinge assembly.
Testing springiness of rough carabiner
Finally, I rounded off edges with files and passed paracord through to ensure no fraying ropes.
The gift carabiners were then complete!
Woo! Now for science.
The three remaining carabiners also had their load-bearing edges rounded off and received identical leaf-spring notches, though I did not insert springs into the gates. These got pull-tested until failure on an Instron tensile machine. The experimental setup, shown below, was conducted with a 20kn load cell at constant displacement (~2mm/min)
It turns out I made a stupid mistake when designing my gates -- the gate's arms that grab onto the keynose were made too shallow and bent outwards at just 5kn, allowing the keynose to slip out.
Past 5kn, all my carabiners behaved as if the gates never existed, and failed at the standard 6-7kn for an open-gate 7075 biner.
On the bright side, having my carabiners fail at 6-7kn means that my amateur heat treatment actually worked! Woah.
carabiner slowly peeling apart
all three carabiners broke their gates first (left) and a closeup of the damage (right)
The 2002 MIT experiment cycled commercial carabiners on a tensile machine for several iterations and monitored their load-displacement curve. They have an interesting change in slope when the gate catches against the nose, then another change in slope when the carabiner itself starts deforming.
My load-displacement curve is more disappointing, but follows the same pattern. There is a slight change in slope around the 1.5kn mark where the gate engages, then the slope levels out at plastic deformation. There are also distinct short spikes wherever the gate slips, followed by a final plummet at failure.
The third carabiner started out with a small crack near the hinge pin (I messed up while hammering the rivet) and therefore has a really weird graph. Oops.
Trial 1 (a&b):
The machine suddenly cut the experiment short, so the second attempt was with a likely compromised biner
Trial 2:
First downspike is the gate slipping; second is failure
Trial 3:
This gate failed early, probably due to prior cracks near the shear pin
Carabiners post-experiment from left to right
(I couldn't find one of the pieces after the biner exploded)
Notes for the future if I make any more of these:
The keynose slot can afford to be shallower, so that the gate can be made thicker
I should have made the gate pattern to be thicker than the carabiner pattern in the first place, instead of making them the same size.
My derpy heat treatment method wasn't terrible!
It still probably isn't quite on par with commercial 7075-T6
Normally carabiners peel apart at failure, but mine exploded -- I probably didn't anneal it for long enough (250deg oven) and the metal was probably on the brittle side.
The main fracture planes were, for the most part, right at the upper basket as expected.
Making a keynose was not as difficult as I was originally led to believe.
I should engrave things post-heat treatment, not before - removing the deformed bits while keeping the engravings intact was a pain.
I can expect the two gifted carabiners to hold up to 5kn loads and not fail until 6, making them more resilient than the waterbottle/backpack carabiners at the top of the post. They are definitely nowhere near the strength of true climbing carabiners!
((Update 2/24/2017))
Found a cool video of the manufacturing process for real carabiners!
