This is the Reuleaux triangle bearing, a mechanism
which takes advantage of the unique properties found in objects of constant width which allow
them to roll like a ball. But it took a lot of experimentation to get
this thing to work – let's get started. Hows it going guys angus here from maker's
muse. Last year I became obscessed with solids of
constant width – these curious looking forms which, when rolled under a flat plane, act
like spheres or balls – Their unique geometry maintains a constant distance from bottom
to top no matter what orientation their in, which made me ponder if a bearing of some
kind could be possible using them. My first attempt was a basic roller bearing
with 2D forms packed around it, and I concluded that although these objects do maintain a
constant width, their center point actually changes throughout the rotation which resulted
in the shapes either falling out due to insufficient support or colliding with each other, resulting
in more of a sliding motion than true rotational movement you see in a normal bearing. That video went on to get loads of hits and
many of you had really neat ideas on how this could be resolved to make a true functioning
reuleaux triangle bearing – so I got to work. The first and by far most popular suggestion
was to implement what's known as a cage – a supporting structure which spaces the balls
in a regular ball bearing, allowing them to rotate freely while keeping them separated
and not rubbing up against one another. The crazy thing about shapes of constant width
though is they do not trace a circular path as they rotate, but instead, in the case of
the realeux triangle, trace a square. It's this property which sees them used in
special square hole drilling jigs. So therfore, the cage has to look like this! This cage looks visually stunning, and I had
really high hopes... but sadly, like before, the shapes tended to slide rather than roll. Their center of mass means the force of getting
past each sharp angle was much harder than just simply sliding along one of the curved
faces, and they also had to revolve up and against the bearing cage. I had a go at extending the faces to mate
with the 45degree bearing overhangs in an attempt to create more contact and thus more
friction to cause them to roll... but nope, no dice! Another suggestion was the idea of constraining
each shape with an additional ring, however because the centerpoint changes during rotation
it can't just be fixed, it needs to in fact be a slot. So in this version the centerpoint of the
objects are losely fixed to the ring with an M6 screw tapped into the plastic. This allows them to smoothly rotate as well
as move up and down, but still be constrained at 120degrees apart from eachother. Theoretically if every surface mated perfectly
and we didn't have this annoying thing called tolerance to deal with, I see no reason why
this shouldn't work... but alas, more sliding! But it ALMOST works. So I thought to myself, what If I could just
give these shapes a tiny bit more friction to work with? I grabbed rubber bands and popped them around
the shapes – and they rolled! The rubber bands take up any small inperfections
in the 3D Prints and causes them to turn instead of slide, although I did notice just how much
more difficult it was to move the shapes past the peaks compared to the gentle curved part
of the triangle. Encouraged however, I went ahead and designed
a modified version which uses a semiflex insert, printed on my modified Cocoon create with
the flexion extruder – the rest of these parts by the way were printed on the Prusa
mk3, mostly fillamentum vertigo grey and gold happens PLA. Also if you're wondering how I achieved this
surface effect, I simply turned top or bottom layers off in Slic3r prusa edition and used
cubic infill. Looks pretty swanky hey. But by this point I was growing frustrated
with how hard it was to rotate the shapes past each curve, so I caved and went about
modifying the realeaux triangle ever so slightly to give it rounded corners. Turns out creating a rounded releaux triangle
isn't a super trivial exercise however! You need to pay really close attention to
the tangential relations of each part to ensure it's truly a shape of constant width and not
'almost but not quite'. And this is its final form – slightly rounded
with a thin semiflex insert to provide just a little bit more traction to encourage rotation. You have no idea how overjoyed I was when
I tested this out and it worked – it works SO well, exactly how I envisioned it when
I first attempted this rediculous concept all those months ago. It's satisfying to turn and doesn't get caught
up, although interestingly the rubber inserts tend to migrate slowly around the shape – I'll
have to investigate that later. But for now, it works... so what to do with
it? Well, how about hooking it up to a motor. So there you have it guys, a working relaux
triangle bearing turned into a kind of kinetic sculpture, and while the use of solids of
constant width has pretty much no advantage over circles or spheres in a bearing that
I can see, it definitely looks incredible while in motion. I'd like to say this is the end of my mad
descent into these forms but honestly I don't think I'm quite done yet, I've got just a
few more wacky ideas to test out in future. And it is my aim to empower your creativity
through technology and if you would like to join me I'd love for you to subscribe. Also a huge thanks to my patreon supporters
for making all of this possible – there was some pretty crazy late nights involved
in this project including this clever idea, yeah I have no idea what I was thinking. And your generous support keeps the channel
going, you can expect to have the files for the final version and instructions posted
up shortly. Thanks for watching guys. Bye!
