I kind of love cycloidal Gearboxes, they’re a similar concept to a harmonic drive but without having to worry about the flexible spline. First read about them here http://lcamtuf.coredump.cx/cycloid/ and now really want to make one.
From what I can tell the main disadvantage for FRC COTS is that they have to be designed for a specific ratio. Unlike a planetary where you can combine a couple lower reduction stages for a variety of high reductions the manufacturer would only be able to offer a few high reduction ratios.
That aside, I really want to make one for a bag motor to steer a swerve, would get rid of the really tall vex planetary problem 696 encountered this year.
To add to pros: they can also transmit lots of torque because 1/3 of the rotor is engaged at all times, and the vibrations can be dealt with for the most part by a second rotor 180 degrees out of phase.
Does a cycloidal gearbox require the use of the cycloidal spline, or can a version of it be done with the involute form like a gear tooth? I’m finding it difficult to generate the spline in SolidWorks.
Cycloid gearboxes are nifty, so I’d use any valid excuse to put one on a robot. Even so, with the number of COTS parts available these days, it’s probably going to be a lot easier and cheaper to build a high reduction mechanism with gearboxes/chain/sprockets.
I spent a little while trying to figure out the math behind the gear profiles for these last year, and learned a bit. As I recall, I wasn’t able to track down the exact profiles they use in the industry. The profile for the funny shaped gear is constructed as shown below, and the ‘ring’ gear is typically made of a set of round bearings (note in the second photo below).
Sifting through some of my old files turned up these; feel free to peruse. This is a two stage version that doesn’t use rollers for the ring gear, so friction is expected to be an issue. Equations are set up in solidworks models, and values calculated with the spreadsheet. The me-from-the-past should have commented on things better, but apparently he didn’t care enough to do so. If someone has experience with designing these things, critique and explanation on correct design approach would be wonderful.
I actually just designed one of these recently, with the ability to stack built in. Most all of the parts are designed to be waterjetted/lasercut/routed or some other form of 2d machining. That way it’s relatively easy to manufacture. There are numerous videos and whitepapers/pdfs on the subject that I used to generate the profile and do all the other math. Here and here and here are the three videos I referenced the most. Thisthisthisthis and this are also very helpful links. Enjoy!
That could be pretty simply solved by facilitating the use of reduction before the input to the cycloidal gearbox. 8mm keyed shaft inputs would allow for direct drive from CIMs/MiniCIMs, but also potentially VersaPlanetary, CIM-Sim, and/or CIM-U-Lator. Allowing for 3/8 or 1/2" round or hex inputs would open up possibilities even further.
These are really cool, and definitely something I’d love to see used on an FRC robot at some point.
However, I’m not seeing any major gap they fill in terms of already available COTS items. Small footprint, high reduction gearboxes are already available in the versaplanetaries, and other COTS options offer similar advantages.
I suppose if you needed a gearbox for a very small area with a large reduction that cannot be backdriven, but I feel that would be a very small number of potential teams/robots.
Where would you see this used? Robot arms/elevators come to mind as the main area, but there are numerous other options.
So, IMO, a cool thing I would’t mind seeing as a COTS option, but not a game-changer.
EDIT: I’ve been informed that I’m wrong. Disregard above.
The lack of backdrive and, equally important, backlash make them ideal for precision manipulator control. I remember multiple teams struggling with one or both of those issues when designing adjustable angle shooters for 2013, for instance. That’s the implementation I envisioned for them, but I’m sure the brainpool in FRC could find plenty of other uses for a high-reduction, ultra-compact gearbox.
Perhaps one of the gearboxes in which the teeth are actually roller bearings. With the simpler ones (pegs in holes), I don’t see how the system could be back-driven; there’s too much friction. [gut feeling]At ratios much over 100:1, you would probably have to intentionally design a gearbox for it to be back-drivable.[/gut feeling] Designing it to NOT be back-drivable gets easier as the ratio increases for any given architecture.
I just drew up and 3d printed a 6.66:1 one for a CIM, and it works like a charm. I can put enough torque on it I am worried the solid PLA will break and cannot backdrive it. With the drawing I was happy with, by swapping out 2 small parts I could easily change it to 5:1, 10.5:1, or 22:1. it is also super compact, being only 2x2x0.9", it is comparable in size to a versa-planetary and if made from alum/steel could easily handle more force*.
Another interesting point about this transmission is while the output is coaxial the input, they are counter-rotating
*Planetary gearboxes generally fail under load when teeth shear off inside. All failure points inside this gearbox are much thicker than the tiny gears we generally see and in theory would handle more force before shearing
Can you post a picture of this? And perhaps a video of it handling decent amounts of torque?
I’m very curious about how this 3d printed version holds up. If that can handle certain loads, there’s no reason a metal equivalent wouldn’t work fantastic.
Did you use the hypocycloidal ecuation to draw the profile of the gear??
I just designed one using tangent semicircles and im planning to build it, but im concerned that this characteristic may not contribute to the non-backdrivable feature.