Since I have been exiled to Coronacation, I have been trying to develop a strain wave gearbox into something that could used in an FRC use case. @Trevor apparently had the same idea here, with a great writeup on what harmonic or strain wave gearboxes are here. Although both of our designs employ the same mechanics, mine operates with a planetary gearset inside a flex spline, inspired heavily by @LoboCNC’s excellent Thingiverse here.
My goal in this was to have the design be as fully 3d printed as possible, as I have a 3d printer with me but not much else. The black is PETG, the blue is TPU, and the yellow is PLA+. I did a material switch when I printed the flexspline, so that there is a ring of PLA+ on the bottom for rigidity, lots of TPU for flexibility, then back to PLA+ on the top for rigidity again. I chose PLA+ because I can print it at the same temperature as the TPU and it gets great layer adhesion. The bearing uses 6mm airsoft BBs.
This design is about 48.2:1, although I picked that ratio because I am hoping to use this for another project; there is a very large range of ratios this design can accommodate easily
While it spins, you can see that there is nothing attached to to output. I tried to print some knobs on the end of the flex spline, but my printer has been having issues getting them to look right, and I don’t think that it will ever be strong enough. My next idea was to use dowel pins to transfer the torque, but I was hoping to share this with Chief Delphi to see if anyone had any advice for what to do.
I couldn’t find where you developed the reaction force until I looked at the thingiverse link. Each of the planets also engages with the housing. In that case, do you need the flexible ring gear? Could you not just have a stepped planet with a smaller ring gear as the output and be the same thing?
If you’re having trouble taking torque off of nubs on the face of the output ring, could you extend the ring and adding a spline to the OD?
I hadn’t thought of generating a reduction like that, do I have the right idea here?
I took a Theory of Machines course that briefly touched on planetary gearboxes, but only with pure numbers The reason I wanted to work on a harmonic gearbox was I had the great opportunity to work with Motoman industrial robots at an internship, and all of their axis’s were driven off of harmonic gearbox. It was my understanding that harmonic gearboxes are advantageous because they are very low backlash and cannot be back driven, although apparently the back drive part is under some confusion and its 3d printed so will likely not have optimal backlash. I can say that I can easily turn the sun gear, but cannot backdrive it at all from the flex gear without any motor connected to it. A minimal backlash / non-backdrivable motor would be a useful tool for articulation or winching needs imo
Yep, that’s it. I think your original design is working more like “variation III” of a planetary gearbox from the link below. Epicyclic Gears - Roy Mech.
It’s not really a strain wave gearbox because the deforming ring doesn’t engage and disengage with a reaction gear as it precesses around. Instead, you’ve just made a smaller ring that has to deform to fit the larger planets. It’s still a very compact little gearbox, which could be useful.
But making it with rigid ring gears would be easier I think.
If this were a simple planetary gearbox, the outer ring would go somewhere between 1/3 and 1/4 the speed of the sun gear, but it’s obviously a much greater reduction. This is a planetary reduction whose ring gear is ALSO the deforming gear to a strain wave gearbox.
@TheIsaacMurrin, I haven’t done any calcs, but my intuition tells me that at this level of reduction, you’re going to need some pretty serious dowels (significantly thicker than your ring gears) to transfer torque to your output . shaft. As a lead-in to that, it might be useful to print a “webbing” onto the outermost (pale green) ring to provide a bit of stiffness to support those dowels. This webbing might potentially hide the blue ring/strain gear.
I’m certain that the fact that the deformation of the ring gear has no real effect. It just lets you you use planets with the same number of teeth on both sides, even though they’re connected to ring gears of two different sizes. But that’s far more complicated than just using a compound planet and a rigid ring gear IMO.
One other note on the design - you are definitely going to need a planet carrier. As soon as you apply any torque to the output ring, the loose planets will try to twist out of plane. A carrier on bearings will keep everything square and meshing smoothly.
@TheIsaacMurrin, would you please let us know which of these (or perhaps something else) is going on here? I can’t decide from your renders. (In particular, your second render in post 1 is more confusing than illuminating.)
In any case, the roughly 50:1 reduction means to me that plastic dowels to drive the output will need to attach to something wider (and maybe thicker) than the annuli as printed so far, which is why I suggested the wider “webbing”.
I concur with @nuclearnerd that this is not actually a strain wave gear box, but accidentally a planetary gearbox. While the ‘flex gear’ is flexible, it only needs to be in this design because I am forcing it essentially into a space smaller than it should go.
If I did not have the blue flex gear this would be normal planetary, with the ring gear in green, the sun gear in yellow, and (one of) the planets in orange. When I added the blue, I indeed got a reduction, but not through a harmonic. The pink gear would be the output, attached to the blue flex gear.
I hope that clears everything up, sorry I wasn’t as clear as I should have been when explaining the mechanics. My next question would be what are the advantages or differences between a proper strain wave gearbox and a compound planetary gearbox? Apparently you can get similar reductions in both in very small spaces, would backlash and back driving be the only other differences?
The reason I wanted to work on a harmonic gearbox was I had the great opportunity to work with Motoman industrial robots at an internship, and all of their axis’s were driven off of harmonic gearbox. It was my understanding that harmonic gearboxes are advantageous because they are very low backlash and cannot be back driven, although apparently the back drive part is under some confusion and its 3d printed so will likely not have optimal backlash. I can say that I can easily turn the sun gear, but cannot backdrive it at all from the flex gear without any motor connected to it. A minimal backlash / non-backdrivable motor would be a useful tool for articulation or winching needs imo
