pic: CGX-108 front
 

Some dimensions are slightly different to allow for more clearance with the versaplanetary and take up error from machining, but it still fits right on and the external dimensions should match that of a VP.
Weighs 0.17lbs, about the same as or a little bit less than a VP gear stage.

Re: pic: CGX-108 front
 

This Looks Very nice. I know I and my team would be interested in a 20-1 VP stage. Is this something that VexPro or AndyMark may make in the future?

Re: pic: CGX-108 front
 

Thank you. Probably not, but you never know!
Download CAD here:
https://workbench.grabcad.com/workbe...folder/1465067

Re: pic: CGX-108 front
 

I bet if enough poeple asked then they might consider it.

Correct me if I am wrong but wouldn't the center gear be wobbling?

Re: pic: CGX-108 front
 

If the Cycloidal stage is connected to another VP stage or the output stage, the output gear would be stabilized.

The stage by itself would be useless.

Re: pic: CGX-108 front
 

Would love to seem them make this and see what it can do. Prototype it at least.

Re: pic: CGX-108 front
 

The two wobble plates are wobbling, but they wobble opposite each other around the axis (input/output shaft) so there is little vibration.
The input shaft has a shaft coaxial with the axle of the VP, and two offet pieces that hold the wobble plates. The shaft is 0.25" in diameter, and the shaft the wobble plates are on are 8mm, and the offset is 0.030". This allow the wobble plates to be put in, then the spline pressed onto the 0.25" shaft.

It should be able to connect to the output stage or input stage of a VP like any regular stage, I'm not sure what you mean. The output of the VP has 2 bearings that constrain it, and none of the ring gear stages have bearings on them.

Re: pic: CGX-108 front
 

How would the camshaft for something like this be produced? Is live tooling required?

Re: pic: CGX-108 front
 

Probably. It's possible to do this with a 4-jaw chuck, but the time cost would be high enough to merit just using a 4 or 5-axis lathe.
It's also possible to do the input shaft on a mill, or make it in 3 pieces to avoid doing external turning on an offset.

Re: pic: CGX-108 front
 

For the non-engineering crowd, why would you want something like this? What's the benefit of this over a standard planetary?

Re: pic: CGX-108 front
 

I think Anand should find a machining grant for the team to take on the risk of proof/prototyping. There's a lot of work in this CAD and I think the team would be very inspired to go through the process of proving a technically challenging design - especially one that's so tangible with respect to an existing product.

Re: pic: CGX-108 front
 

Large reductions in a small footprint.
http://www.chiefdelphi.com/forums/sh...idal+gearbo x

Re: pic: CGX-108 front
 

Cool! It would be fun to see it built and what kinds of problems are encountered. The cam itself is kind of floating, there's not much to keep it co-axially aligned with the housing other than the cycloid wobble plates, but it looks like a good application.

It should potentially be stronger than a geared stage of a versaplanetary, given it's relying on a lot of pinned connections instead of a few meshing gear sets. If you stack to many stages of a versaplanetary the output stage can fail, since it sees a lot of torque. If you have a stronger stage, you can handle more load and get higher reductions before breaking things (not that you really need to... the versaplanetary system is already pretty solid).

Re: pic: CGX-108 front
 

I'm looking into getting this manufactured. The wobble plates cannot really be machined manually, but I might be able to get the housing and output done on the mill. After that it's a matter of just getting the camshaft and wobble plates done on a CNC.
The input spline I might be able to do on a lathe by boring out a VP stage, but I'm not sure how to do the output. That would have to be wire EDM'd or something similar. I should send an email to Vex asking how they do it.

Re: pic: CGX-108 front
 

For the quantity they run it would not suprise me if they had a custom rotary brouch.

Re: pic: CGX-108 front
 

Always amuses me every time :)

Can this go in both directions, or can it only go in one direction? I remember something about that a while back, but my memory is failing me there.

What practical application would this have? I would imagine this having a slower speed than most gearboxes, but a great amount of power/torque for game elements that would require such.

Re: pic: CGX-108 front
 

It can go in both directions, but has backdrive resistance. Using pins should make it totally anti-backdrive, but I'm thinking about switching to rollers for just backdrive resistance instead.

The 20:1 reduction in the same space as a 10:1 reduction has a few uses. For example, intakes with high-speed motors like BAGs or AM-9015 can easily require more than a 10:1 ratio, and often times a 20:1 ratio could work. This year for 115's elevator we used a 21:1 reduction 2-stage versaplanetary on two RS-775, as well as a disc brake to hold the totes. We could have simultaneously saved a little space and weight as well as removed our disc brake by using a 20:1 cycloidal stage.
Any situation where you require lifting of the robot or something similar that requires full anti-backdrive I would use a pinned cycloidal versaplanetary stage. But if it doesn't need to be that strong and you want better efficiency, a roller cycloidal stage would work better.

Re: pic: CGX-108 front
 

What's the difference between a pinned and a roller cycloidal stage?
Thanks!

Re: pic: CGX-108 front
 

The difference is whether the coupling between the two layers is made of solid pins, or if rollers (long thin bearings) are installed in that location. With rollers, there will be less friction, so that backdrive can work. There are 8 pins in the render at the beginning of this thread.

Re: pic: CGX-108 front
 

<Updates coming in tomorrow>
;)

Re: pic: CGX-108 front
 

You know the drill: video, or it's a bench grinder...

Re: pic: CGX-108 front
 

I briefly considered this, then decided I'm not evil. :D

A video of it working (10:1 planetary, followed by a 33:1 cycloidal stage):
https://www.youtube.com/watch?v=ArUGM51HFVs

A video of a basic anti-backdrive test:
https://www.youtube.com/watch?v=tG_l...ature=youtu.be

I changed the design to use 1/16" pins around the periphery, and found that let me get 33:1 into a single package. I made this one all custom, but the next revision just requires a carrier plate from a 7:1 instead of the modified carrier I had to make (thanks to 649 MSET-Fish for providing 7:1 carriers!)
The noise level is high because the way I machined the housings, I neglected to centerdrill the pin holes before final drilling, resulting in holes that are all very off. The lasercut wobble gears I am using are now too large because of this, and so run very noisy. I will remake the housings next to fix the issue.
Right now I would estimate the maximum non-backdrive torque before the gears slip is around 25 foot-pounds, which is quite low, and the failure mode is that the gears go skew on the shaft as the plastic bores get larger than the shaft over time. Replacing the plastic or using some extra shims to prevent bending on the shafts should fix the issue, but we'll see.
The wobble gears will be replaced with aluminum for some more realistic torque tests soon, and then replaced with steel for the final cuts. For the immediate future, I can get slightly smaller gears lasercut to run it quieter.

As a neat sidenote, I made the offset shafts using a boring head with a backwards boring bar in the mill. At this point I can start from a bar of 3/8" round and turn it into a shaft in around 45 minutes.

Re: pic: CGX-108 front
 

Awesome to see a working prototype! Got any pictures of the inside?

For the wobble plates, I don't expect the plastic to hold up very long at all, but if you have the ability to cut plastic ones easily you may be able to iterate with them until the fit is just right, so you know what dimensions to use for the aluminum/steel versions. Also, will the metal versions be laser cut? I'd be curious if there is any kerf angle as a result of lasercutting that might affect the interface to the rollers. With such tiny pins, the tolerances on all of the parts will probably have a big affect on the performance.

I'm really interested to see how the final one performs! Maybe you could set up a simple bench test apparatus (turn a pulley to wind up a weight, for instance) to test it under different loads. Data is useful!

Re: pic: CGX-108 front
 

Pics of the inside have been uploaded to CD. If you want more specific pics I can get them.
The plastic ones are already wearing out a ton, but I do need to use them for iteration. The lasercutting seems to be accurate to +/-0.001" with almost 0 kerf angle, so I get realistic predictions of what metal ones will do from it.
The metal versions will be CNC'd for now, and wire EDM'd hopefully for the final version.

Re: pic: CGX-108 front
 

This is really cool. How are the gearboxes holding up? Any interesting failure modes observed so far?

Re: pic: CGX-108 front
 

This is a really cool project, and I'm anxious to see the final result commercially available. Although I would be even more excited if you could make your own final stage as well. The Vex final stage has really insufficient mounting strength at high ratios. Two #10 bolts isn't enough to hold the reaction torque (above 100:1 on a bag motor) in our experience.

Re: pic: CGX-108 front
 

Nice! Ever since you posted the CAD a few months ago, I was eager to see it come to life.

Good luck on the final iteration! Hopefully this will be something VEXPro will look into in the future, it really is a nice way to get a compact and high reduction stage for Versa Planetaries.

Re: pic: CGX-108 front
 

While I'm still of the opinion that cycloidal gearboxs are magic, the thought of a 20:1 or higher reduction with limited or zero backdrive in a VersaPlanetary form factor is VERY appealing to me. It could also be a way more flexible and compact solution for applications that many teams currently use worm gearboxes for.

Re: pic: CGX-108 front
 

Thanks! The "good" failure mode is just a ratcheting due to the plastic teeth skipping, and the gearbox still runs afterwards. The more common "bad" failure mode is when the plastic gears go skew on the shaft, locking up the gearbox. Using metal hubs or bearings should fix these issues. I would not use the gearboxes on a robot right now, as the tolerances are all over the place. I'll be fixing that next week.

Thank you, I hope to be able to make this a commercial thing eventually.
We have used the side mounting holes in the past to hold our VPs, so doing that could help. You could also cut 2 more mounting holes such that it holds the VP together; you would use the 2 corner holes that currently house #8 bolts that hold the VP together.

My thoughts exactly! Using the plastic gears allows for ratcheting, so the gearbox doesn't get destroyed by over-torquing the output (like in a collision with another robot), but metal gears if properly constructed would be able to shear the 1/2" hex output before the gearbox fails.

Re: pic: CGX-108 front
 

This is a bit of a resurrection, but has there been any progress on this design? I think this could be a great option for teams wanting to power arms or winches or a host of other subsystems.

Re: pic: CGX-108 front
 

Why not just use a stock versaplanetary?

Re: pic: CGX-108 front
 

Why not use a single 33:1 stage instead of a 3:1 stage and a 10:1 stage? Also, I'm a fan of non-backdriving gearboxes, and this is a much more robust option over the window motors.

Re: pic: CGX-108 front
 

One has thousands and thousands of hours of FRC in competition testing, the other does not.

Easy enough for me.

Re: pic: CGX-108 front
 

Every new design has to start somewhere. When Versaplanetaries first came out they didn't have nearly the testing hours they do now.

Re: pic: CGX-108 front
 

They may not have had the amount of testing they do now. But I'm certain that a company as large as vex would have done some pretty in depth testing before putting it in the hands of first teams. That being said I do agree that this would still be a cool thing to see produced and used on an frc bot.

Re: pic: CGX-108 front
 

The only reason this wouldn't back drive would be inefficiency. Cycloidal drives DO backdrive. :)

Single stage cycloid backdriving!

Re: pic: CGX-108 front
 

I agree that non-backdrivable cycloids are more inefficient. The efficiency is related to the friction the angle of contact between the the teeth, and I think the offset as well, and the static coefficient of friction between the ring teeth and wobble gear teeth is what causes the ability to non-backdrive while lowering the efficiency. IIRC cycloidal drives with static teeth are extremely unlikely to backdrive at the cost of reduced efficiency, while ball-bearing ring teeth allow it to have just enough efficiency to be backdrivable, albeit not without lots of friction.

As a side note, that is a beautiful cycloid box! What kind of efficiency did you get out of it in the end?

Re: pic: CGX-108 front
 

Didn't get to measure it unfortunately. It's in the field and operational though.

Re: pic: CGX-108 front
 

Also bumping this thread because I still love this design.

There are ways to work around backdriving, but backlash is a much nastier problem to extinguish from high-reduction precision systems. This is one area where cycloidal components excite me. Granted, if you put any additional mechanical reductions after the output shaft, you're re-introducing the possibility of backlash. Thus right now we'd be limited by what the Versa output shaft can support.

Even still, this would be one of the few products I'd push to become an "early adopter" of (even if just on practice/demo machines).