pic: CGX-104 cycloidal versaplanetary external



A cycloidal gearbox made using Lil’ Lavery’s cycloidal gear generator equations.
This is made to work like a versaplanetary, in that stages can be added or removed by simply stacking, but it is not compatible with the versaplanetary directly. It is slightly wider and taller (a few hundreths) but is considerably shorter (almost 3/8" for two stages). It also weighs slightly less.
Pictured is a two-stage 10:1 x 10:1 cycloidal versaplanetary. Stages are easily generated by plugging in Lil’ Lavery’s equations; I will put CAD up once I make 4:1 through 12:1 stages.

Currently there are no alignment mechanisms betwen stages (I can add those) and there is no way to reduce the vibrations in each stage. However, the eccentricity of the gear is a mere 0.03", so the vibration should not be terrible. I will see if I can balance the gears better to reduce it further.

Questions and comments are welcome.

1 Like

I love the compact design. Perhaps designing a small coupler that goes from a versa planetary to this would be something people would find useful. Hope to see some CAD files soon.

My equations? :confused:

Thanks for the credit, but I don’t think I ever got around to finding you any equations.

As I understand it, the advantage of cycloidal is the larger ratio of reduction available in a single stage, improving efficiency. Stacking gearboxes that match VP ratios would give up that advantage.

If you made cycloidal gearboxes at ratios of 36:1, 40:1, and 45:1 that could take a VP output as input, then with a single VP and a single “VC” (VersaCycloidal) stage, you could cover the range from 108:1 to 450:1 with no gaps larger than 12.5%.

Here’s the table:


     Cycloidal Ratio
VP    45    40    36
-----------------------
 3   135   120   108
 4   180   160   144
 5   225   200   180
 7   315   280   252
 9   405   360   324
10   450   400   360

Of course, you must put the versaplanetary first because they couldn’t take those large output torques.

You distributed a solidworks model with the cycloidal gear generated from equations on a spreadsheet.
EDIT: That would be s_forbes, not you. Sorry.

The problem is, a 10:1 reduction is pretty much the biggest size you can make using 1/8" pins and a 0.03" eccentricity. If I made the gearbox a little bigger, I could add more reduction.
I figured the low backlash and backdrive resistance would be worth it.

Another advantage to this over a vp to consider is that at the moment, from looking at the renders available, this might be machinable by a team with some cnc access. Whereas a vp is something you must buy. For some teams buying the cots item is easier, for others, getting the machine time is easier.

Could you share the link? I can’t seem to find it on chief.

This is CNC-able, but some parts, such as the output shaft, would require annoying setups. I might try and work on that later.

It’s not Lil’ Lavery, it’s s_forbes, my bad. I can update the description.
This post: http://www.chiefdelphi.com/forums/showpost.php?p=1486447&postcount=7
With the zip file.

Very nice. Glad I see others working on this technology.

FYI…from my own experience building these, the 0.030" eccentricity will be violent and rob efficiency.

It doesn’t seem like much, but spinning at speed it’s wicked.

Thanks for the advice. I’m trying to figure out how to add a counterweight without needlessly using space and weight.

Everyone of these units I’ve ever seen has either had multiple wobble plates out of phase for balancing or a dedicated balance weight.

There’s no free lunch.

It’s no coincidence that you only see these types of drives in major industrial applications where a bit of extra mass and volume is worth some of the cycloid’s benefits.

Here’s an unbalanced unit designed to work with RS-550 and RS-775 motors built for the 2012 FRC season.

You can see that this unit is very polished and was even intended for production…however, its promise didn’t match with its performance.

I was thinking of putting a second gear of the same mass 180* out of phase, but on the same shaft, such that I would basically be getting the face width of both of the cycloidal gears in strength. So essentially having multiple wobble plates, yes.

You must put the second and/or third plate must be on their own eccentricity that is also out of phase.

I have to say, this looks like a fun engineering challenge, and good on you for designing this.

In regards for a team machining this though, I always try and recommend buying as many parts as possible (of course this depends on your team’s budget however), rather then machining them. The reason is, you can put that much more time into either designing, programming, or practicing with your robot.

In the first few years that I was on 772, we machined a lot of our own parts. However, when I started to design the robots we transitioned to buying more COTS parts. We focused on designing more, rather then machining. And we were “done” the robot a week earlier then normal, and had a lot more practice with it. The robot’s level of performance went way up too.

Remember, your robot isn’t going to become a great competitor because you custom built a transmission, but you can get better by designing the whole robot better, having a great strategy, having reliable parts (with easy to find replacements), and having more practice.

While designing this, I made sure to make it such that each could be machined with only access to one side (with the exception of the output shaft, no flipping of parts is required). That being said, I more designed it as a “replacement” for a versaplanetary; hypothetically, if I had access to Vex’s machining resources, how would I be able to use them to make a gearbox? The end result actually requires less machining in some ways, although I’m not 100% certain how they make the output shafts on the versaplanetaries.
I agree with you though, COTS parts and driver practice is magic. A few days with our robot between SVR and Champs doubled our potential score.