Falcon Cycloidal

Questionably legal.

Dubiously viable.

Poorly rendered.

Optimized for machining? Definitely not.

It’s what you’ve all been waiting for…

The Falcon Cycloidal!

I was planning on fully machining and assembling it before making this post, but for various reasons I doubt that will happen anytime soon so I’m just posting now.

This is a 25:1 cycloidal reduction that replaces both the output shaft and output plate of a Falcon 500. It ends up about 10mm longer than a stock falcon and the entire assembly (according to the cad) weighs 1.5 lbs.

All the bearings are 67XX bearings sourced directly from somewhere in China.

CAD and video of how it works available here or here (first is a .zip, second is a link to a GDrive folder with the same contents)

48 Likes

23 Likes

18 Likes

genuine question why should something like this not be made?

I’ve been thinking about this exact thing for awhile and it seems like it would be a great option

1 Like

Make sure the cycloidal discs are lubricated with something that reduces the aluminum on aluminum adhesion.

I think that it’s a good concept but it has somewhat limited use on a robot since the reduction is basically built into the motor. For all the time, resources, and ultra-fine tolerances, the effort of chaining two of them together for a higher-load application probably isn’t worth it when you can buy/make a multi-input motor gearbox.

Otherwise, for something that’s in a really tight spot or only needs a singular motor with a higher reduction, this is neat.

1 Like

I also have had cycloidal on the brain. Cycloidals have many aplications such as arms and wenches. Any time a high reduction in a small space is needed. We are currently prototyping a neo cycloidal arm. It’s mostly 3d printed. I know everyone says plastic can’t take the loads. I intend to try my best to prove them wrong.

3 Likes

@Ryan_Dognaux Since a 1720 student is developing this, should we plan to see this on the shelves of thrifty bot soon? :grin:

10 Likes

The nice thing about falcons is that you rarely need more than 1. For something like a large arm, you can use 2 of these and have them face each other. For the weight and space, I think it would be at least 3 falcons worth of gearbox before this concept loses out.

Cycloids don’t require ridiculous tolerances for acceptable operation, and have large inside radii, making them ideal for machining on standard CNCs.

1 Like

Wrenches. Or winches?

Wenches are fine without cycloidal gears.

12 Likes

This is cool, don’t get me wrong, but I have a question: What’s the point of a ~25:1 compact cycloidal?

We already have pretty compact (and not too expensive) ~25:1 planetary gearboxes (Vex VPs, Rev UPs, and AM Sports being the toughest of the bunch). The market is is well served for intakes, elevators, conveyors and other lowish torque applications.

I can see a use for a compact reducer in an arm joint, but arms require gear ratios at least 100:1 to over 400:1. I don’t see anyone working on a cycloidal with that much reduction. I don’t think it can be made in 3DP plastic.

6 Likes

I can think of some applications.

2 Likes

The idea behind making this, as opposed to just using one of the planetary gearboxes that you mentioned, is to save weight and space while being stronger.

The Rev UP, while a great little gearbox, is too small and weak to really overlap with the use cases I was imagining for this. I doubt anyone is going to try to put a falcon into a UP to use a climbing gearbox.

The Vex VP and AM Sports, while great gearboxes, can get really long. The gearbox, likely multiple stages to get a large reduction plus the input adapter plus the motor itself add up quickly, and you can end up with 6 inches or more of gearbox and motor cantilevered off of whatever mounting you’re using. Additionally, depending on what ratios you choose for a VP, they can be weaker than the stall torque of larger motors.

I chose this reduction because it is About Right™ for a climber. If more reduction is needed (for an arm or similar), the typical method of gearbox → large sprocket and chain is probably better than designing a 100:1 cycloidal.

Luckily, this isn’t! I’m currently machining it entirely out of aluminum, and may make some parts (probably the cam shaft) out of some steel alloy if it fails.

21 Likes

I had intended to make one in that range out of aluminum in the offseason in 2020 but with the lockdowns, I lost access to the CNC and the students and haven’t had the opportunity to get back to it.

Link to the design on GrabCAD can be found here. More links to the research materials I used are available on that site.

The geometry can be updated to get to ratios down to 1024:1 or more, if I remember correctly that was the smallest I could make the teeth in this package and maintain the ability to machine it with a 1/8" endmill. These ratios are based on the formula found in this document here. And yes I am aware that this thing will vibrate, I never got around to designing in a counterbalance.

1 Like

It would replace our biscuit… if we ever had another application for the biscuit.

1 Like

If you get this puppy up and going, I’d like to know if it’s back drivable.

Might be a tall order, but an absolute encoder on the output would be nice.

Mark, if you want something similar to play around with, Harbor Freight used to sell a lug nut remover that worked on the same principle.

2 Likes

I think this is not too hard - the output plate could be reconfigured it to have a locating feature for a Lamprey magnet. Sadly, the bolt circle on the gearbox is 2.25" instead of the 2" on the encoder board, so we would need another mounting scheme (or a total redesign of the box).

As for design notes: my one possible thing I would consider adding is a plate on the motor side to seal the whole thing off, so you can fill the gearbox up with grease without also filling the motor with grease.

Even for cycloids, backdrive is directly correlated with efficiency. A 49% efficient one won’t backdrive, but a 51% one will (assuming certain things about friction). This one should be backdrive fairly easily.

What about a 50%?

1 Like