Variable Speed Gearboxes?

My grandpa has been exploring this concept since the '70s and now he’s got me thinking about it, so I was wondering if anyone else has been exploring this idea. So essentially you have two gears; one long, stationary gear and one thinner gear that is free to travel along the length of the other gear. As the thinner gear travels along the length of the stationary gear, the gear ratio changes, giving you a massive amount of different gear ratios to choose from between the starting ratio and the end ratio (example: if the gear ratio at one end is 1:1 and the gear ratio at the other is 18:1, you can have virtually any gear ratio you want that falls between 1:1 and 18:1). I could see this idea being potentially useful in FRC and I was wondering if anyone else had thought about it. I do have a few ideas of my own (including one that will work for low-torque applications), but I haven’t seen anyone else working on this concept.

This concept is an actual thing, however it does not typically involve gears or the exact mechanic you proposed. Some hybrid cars make use of what’s known as continusouly variable transmission (CVT) which utilizes belts and conical pulleys to change the effective ratio (r1 /r2). Belts are more resilient in this application than gears. Check out variomatics if you are curious on how to apply it on a smaller scale. I could see some FRC use but the required pulley sizes for the HTD profile might not be feasible in a compact “gearbox.”

There is also a cone-type CVT (you can read about it well down on the Wikipedia page for CVT) which sounds a lot more like what you are describing, though it uses rollers rather than gears. Using gears with a CVT sounds like a recipe for broken teeth, or at least rapid wear.

Would it be possible to use a modified version of the cone CVT that instead used tapered worm gears?

Two tapered worm screws would not really engage the teeth, but act more as rollers.

If you mean a tapered worm screw engaging a worm wheel, then the teeth would engage, but the angle of the tooth on the worm screw would become shallower as the screw gets larger, meaning that in order for it to engage the worm wheel, the axis of the worm screw would have to be rotated as it was translated along to engage a larger or smaller part of the screw to the teeth on the wheel. At least; I would not be surprised if there were more issues once you licked that one.

This definitely looks like it belongs on a Steampunked robot. Who is brave enough to try it? :rolleyes:

I actually spent about a month of my life obsessing over this exact idea. And aside from the CAD practice I got from it, it was mostly a waste of time. You run into a whole host of issues as your cone gear transitions from one tooth count to the next, you have all sorts of issues with the tooth spacing and the gear mesh. You are extremely liable to both bind the gearset or just rip teeth off. Also, the transitions between speeds is very rough.

If you want to PM me, I can tell you more about what I tried and how I eventually decided it was a futile project. I might even still have my CAD files, but I’m not sure about that. If you want to look into CVTs for FRC, I’d suggest you look at CVTs for 4-wheelers or Gators. I’ve been meaning to design a CVT of that style for FRC purposes but just haven’t gotten around to it.

I’m also still hopeful that there is a way to make a geared CVT, a CVT that doesn’t rely on friction to operate. And I’ve got a few ideas on that subject, but I’m pretty convinced that a cone gear or a spiral gear isn’t the way to do it.

Belt CVTs as found in a gator or 4-wheeler are all incredibly oversized for FRC use (10-15hp, vs our 0.5-1hp). In that size category, Gaged Engineering makes a popular aftermarket belt CVT for race applications. Look up SAE Baja for some innovative CVT designs at a college level. UofA and ETS have both run manually adjusted CVTs in the past.

About ten years back I’m pretty sure either 148 or 217 was running a CVT, but found it weighed ~5x a 2-speed system and was not giving a sufficient advantage to justify the weight. Fun project though, can anyone can find pictures?

EDIT: Search is your friend, check out this thread from 2008…

EDIT2: And here’s the whitepaper from 217’s 2002 championship run with a CCT, slightly different concept. They’re not the only ones who have done it, go do your own searches if you’re interested in learning more :wink:
https://www.chiefdelphi.com/media/papers/1361

On that note, CVTech is another company that makes belt CVTs that are popular with Baja SAE. But CVTech’s are even bulkier than Gage’s, and I think the design on Gage’s would be better suited for scaling down to FRC level.

I’ve been personally pondering on this concept for the past year, it started when I found this video: https://www.youtube.com/watch?v=F6zE__J0YIU
I’d like to build a D-Drive for FRC so badly but I really don’t know how to go about it

Instead of a stepped gear system, I could envision two conical splines that mesh around the thick, lubricated belt. Perhaps a dual sided belt would work, or a very heavy duty smooth belt. I think the solution below is actually fairly feasible, but I would be very concerned about the belt’s wear over time.

http://imgur.com/a/C4LC7

In respect to a CVT transmission, too much cost, weight and complexity for an FRC robot. Plus electric motors produce their peak torque at 0 rpm which is a big reason for a multi-speed transmission in the first place.

Zero turn drive systems ,riding lawnmowers/Bobcats, use a hydrostatic transmission on each side and a single engine powers both sides of the drivetrain. It would be really cool to build a single 6 CIM transmission that powers two hydrostatic drive units. Except for the fact that it wouldn’t be FRC legal.

Argo units and other off-road atv’s use a mechanical version of a zero turn system. There are trade offs in the ability to vary the turning speed of the inside and outside tires. Read more below.
http://www.argome.com/recreational/whatsnew_detaila3d1.html

In the end, an FRC dual transmission tank drive train is the lightest, least complicated option.

Lightweight CVTs for FRC are at the moment nonexistent, and it will likely stay that way due to their nature. All of them are friction-based or hydraulics-based, which means it’s either illegal (by 2016 rules) or difficult to engineer. I tried my hand at making CVTs a couple years ago and found that it was extraordinarily difficult.
If you want something very close to what you envision, a 2-stage ball shifter may be your best option. You can design a ball shifter shaft capable of switching between 4 or 10 ratios very easily, as long as you can find a cylinder or linear actuator capable of shifting between them.

Many clever mechanical devices lose their appeal when autonomy* is required.


  • That annoying burden of pulling one’s own weight while carrying one’s own energy supply.

Variable pulley? Tensioning requires a system with a spring or multiple springs. Mr. oscillation can come to the party very easily. We have a CVT swerve module built and tested as a single unit. Do we need it for 2017. Do we dare go with it ? Questions to be pondered next Saturday and Sunday. CVT the Holly Grail of power transmission.

I have a big passion for CVTs. I did a preliminary (concept design) of one as a Senior Project in college. It was similar to many of the cone style CVTs that are out there. A great reference for these is:

Traction Drives, Selection and Application:

Most run into a scaling and materials issue that causes the user to either need some really impressive material, incredibly high rotational speeds, or power tends to be limited around 10 kW. The belt drives have found a nice way to get to around the 100 Kw levels.

Within the family of CVTs, there is a specific type called IVTs (infinitely variable transmissions). An IVT has its effective gear ratio go to infinity as described as input/output as the output speed can go to “0”. In theory, torque would approach infinity, but at the same time, efficiency goes to “0” as well, so the system just becomes a heat generator which can often blow itself up. The most common IVTs I know of are typically using hydraulic (some times called hydrostatic) transmissions.

One of the neatest hydraulic IVTs out there is the Dana Rexroth HVT 3R.
http://danarexroth.com/function.html
It uses planetary sets as a combining function for three different ranges of CVT, with a seamless transition.

The concept you are talking about of a “geared CVT” has trouble because gears work in integers which means you cannot have a continuous tooth changing geometry and count.

Within the world of “Geared” CVTs, there are a couple that come close. Ker-Train a company out of Canada does a lot of cool Transmission research. They have a few really neat transmissions that use actual gears.

The first, they refer to as a “binary” transmission that has fixed gear sets with fixed ratios, and clutches, but the layout doubles the amount of ratios per gearset. Thus 1 set has 2 ratios, 2 sets have 4 output ratios, 3 have 8 output ratios, 4 have 16 ratios, and 5 sets have 32 ratios. While they are 32 unique ratios, you are starting to get really close to a continuously variable transmission.

The second, they often refer to as “variator” transmission:
http://www.google.com/patents/US6849023
This is one of the most interesting gearboxes I have ever seen. It uses “non-circular” gears to create a speed waveform when given a smooth input. That waveform is then combined with a second waveform via a 3 element differential (think planetary set, or Kertrain has a system called Co-planar loop). There is another element which changes the phases between the two waveform generators. By shifting the waves, the net difference varies with a smooth output. This is really the only true “Geared CVT” I have ever seen.

There are also a type of CVTs called Ratchet drives which are pretty neat.

The most likely thought for an FRC application would probably be something akin to the PRIUS transmission.

As others have stated though, It is really overkill for FRC. Electric motors do a pretty good job of acting like a CVT as it is, so there is a lot less advantage.
In theory you could gain more time with the motor operating at or near peak power, but in practice for what we do, the time commitment to controls and fabrication would outweigh any advantages.

CVTs do have a place though. They are sometimes used in variable speed drives for machinist tools (like band saws and lathes).

Someone more or less posted something like this last year.

Moderator note: corrected dead link from old forum software

1 Like