I ran across this while looking for new CVT technologies:
http://www.fallbrooktech.com/NuVinci.asp
They are trying to market this as an improved bicycle transmission, but I think it would be really cool on a FIRST robot. 
I wonder how much these will cost (they are not yet in mass production), but if they are for bicycles, I would think that they should be rather inexpensive? 
I saw that a while back (I keep my eye open for “developments” in CVT technology), and blargh–I’ll wait for user reports. It would be nice to experiment with, though.
Edit: I was thinking of the older shimano nexus stuff (and their more current stuff.
Very different technology, for sure. These things look to be promising, but I doubt we will see any progress made in the bicycle industry using these. They will have the same problems that the nexus hubs have/had. Their size and form factor just will not work well on a bike. Using it as a hub is not appealing to anyone other than the recumbant users who love their nexus drivetrains. Maybe the downhill bikes where the extra weight isnt an issue and the frame can be modified to accept the CVT inside the frame, similar to how the current high end GT downhill bike is made.
It’ll be interesing to see either way.
Uh… bike parts are rather expensive. If you look through bike magazines, gear sets, frames, shocks, etc. are generally upwards of $1000… and since this technology is “new”, especially to the bikes, expect it to be way out of your price range.
i think it could be scaled to fit in a bike fairly easily, but it seems that it would still be more complicated than normal bike shifting.
Also, if it breaks, it looks like it would be a lot harder to fix than bike sprockets.
One benefit i see is that it would probably take up less space, though, because it is planetary, might be difficult to fit on bike, because the lined up sprockets fit so well on a bike, and this cvt would be signifacantly wider.
Very cool, i wouldnt be suprised if it becomes popular in all sorts of applications.
I too have been watching the NuVinci websites for new news and am most interested in the Light Electric Vehicle (LEV) products described here: http://www.atcnuvinci.com/products_lev.php. The LEV setup looks pretty ideal for a robot drive train if the weight is reasonable. The FAQ says the LEV should be available early next year.
But on their website they claim that it will be “inexpensive”…I guess we’ll find out when they actually start selling the technology.
It would be really neat to have an off-the-shelf CVT that we could use on our robots! :yikes:
The video describes it as being filled with liquid. I’m sure it’s sealed, but that may not fly within the rules.
Still, an interesting concept.
It would be terrible for a robot.
CVT has no way of preparing for an impact such as 2 robots crashing into each other.
100% of the time you would lose a pushing match against a geared robot.
I know from experience, my moms SUV has CVT and ive pushed it to its limits out at Laguna Seca Raceway.
I think that saying 100% of the time is a bit bold. The technology is unproven, and it would probably require a different driving strategy as well as some design considerations to accomodate the different demands of the system, but just striking it down at this point seems a bit hasty to me.
No, 100% of the time is right.
It has no way to prepare for for an impact, and once impact is achieved, it has to torque itself back down which can take anywhere from 1-3 seconds.
ALOT can happen and 1-3 seconds, including alot of lost ground.
Go drive a CVT car and you will see what I mean.
Actually, Ed is correct. Making blanket statements about CVT/CVPT systems like “it would be terrible for a robot” and “100% of the time you would lose a pushing match against a geared robot” is a bit presumptuous. There are many, many cases where a transmission such as this may be the absolutely perfect design choice. You cannot presume to understand all possible application areas, design constraints, performance requirements, and efficiency characteristics where such a design may be applied.
In addition, you are presuming that the CVPT being discussed is the same as the CVT in your SUV, with the same capabilities, performance, and issues. It is not. Making a complete condemnation of the technology based on a single experience with just one point in the solution space is erroneous. An implied statement like “I don’t like the CVT on our SUV, therefore all CVTs are bad” indicates imprecision and assumption in the analysis. Whereas “I have seen some issues with CVT on automobiles that MAY also occur with robot CVT transmissions - we should look into this some more” is a more appropriate response when the available information is incomplete and imprecise.
-dave
When did I say I didnt like the CVT on our SUV? Good job putting words in my mouth. In fact, I think quite the opposite. It is incredibly enjoyable to drive around town, and if accelerates only seconds slower than a Honda S2000.
HOWEVER, I wouldnt use any sort of CVT on something that tows or pushes such as a front end loader, simply because the pushing power would be delayed, as well as it would be incredible stress on either the balls or belts.
The same is true to robots. Sudden collisions or load on the drivetrain would reak havok on a CVT tranny.
Ever repaired one? I have, and it wasnt a pleasant experience.
It sounds like you’re assuming a CVT wants to be some sort of constant torque output. That might be true of certain types in certain applications, but the “New CVT technology” at the heart of this thread is merely a transmission with a continuously variable “gear ratio” using spherical transfer instead of toroids and disks.
Did you take a good look at the diagram and description on the web site that Eldarion posted? I’d be very surprised if the CVT you repaired has much in common with the one being discussed here.
It certainly sounds like it would work fine on our robots from what the company is putting out there about them.
They are also making the argument that it will shift better and faster than standard geared transmissions. I believe the problem with the slowness or slack that you felt in the automotive CVT is from not only a different design but also its computer controls, A computer adjusts how it shifts and the ones in those cars are designed for smoothness not for power.
The Yonge-CVT website provides a lot of really good information about CVTs. Including descriptions and critiques of other CVT technologies, such as the NuVinci and Anderson CVTs.
Actually, sudden collisions and rapidly changing dynamic loads are exactly why you would want to use a CVT for an application like a FIRST robot.
Follow the load path for a typical FIRST robot with a traditional gear train providing power to the wheels (or treads or whatever). When the robot slams into a wall, or rapidly goes from forward to reverse at full speed, the shock load transmitted through the gear train will be significant and can be enough to cause failure of the drive train. Examination of a number of case histories has shown that the in the vast majority of cases (better than 95%), the weak point in the system is in one of two places. If the gear train is designed properly, the failure mode will be a loss of traction between the wheels and the ground, and the wheels will just spin in place. Annoying, but not fatal.
But the other frequently seen mode is when the wheels do not break free, and the entire shock load is transmitted through the gear train. All too frequently this results in broken gear teeth and a trashed transmission requiring extensive repair or replacement. This is because the gear teeth tend to be the weakest points along the load path. Unless specific precautions are put in place (ie. designing primary failure points with shear pins, properly calculating tooth face width to increase load tolerance, etc.), the result is that a gear is turned into a disk. This failure mode makes up the vast majority of transmission failures that are seen at FIRST events (other than the problem of having the silly wire clips fall out of the Bosch gearboxes and causing them to drift into neutral that was a big problem a few years ago, but fortunately we don’t have those any more, so the less said about that situation the better).
Alternately, the failure mode in a CVT is a non-fatal event. The load path between the input and output elements is across an intermediate friction drive (based on the design, either balls, rollers, disks, or splines are used as the intermediate elements). When the system is over-loaded, the friction in the drive system is overcome, and the elements slip in relation to each other. In the worse case, there will be surface wear on either the intermediate element or the torque plates which may shorten the overall lifetime of the drive. But it is not a catastrophic failure, and a properly designed CVT will have wear compensation adjustments built in that will enable the CVT to continue operating.
This is exactly the sort of capability that you want to have on a robot that is going to see repeated severe shock loads. In a situation where you have a choice between a traditional gear train and a CVT, and both have been properly designed to withstand the same loads, and you know/suspect you are going to have dynamic shock loading situations that exceed the load ratings of the transmission, then selecting the system with the non-fatal failure mode would be a prudent choice.
Yes, I have repaired one. I have also designed one from the ground up.
-dave
What type of CVT was it?
Also, since you seem to know quite a bit about CVTs, is there a way to make one that is hard-linked (i.e. not friction based)?
Thanks!
CVT’s are a very interesting propostion. I have heard of some planetary kind in passing, but never got a chance to hear/research more on it. Does anyone know what at all I could be talking about? It has something to do with the speed of the outside ring gear…
As far as on a FIRST bot, I think it would depend entirely on the game, and the weight of the module. If I could take 10 pounds off by having a single speed, and instead use the 10 pounds on a scoring method, I would. If the module were light enough, It very well could be worth it in the long run.
The Yonge-CVT is hard linked. It is definitely a unique design and seems to be very well researched and thoroughly documented. You can even read his development journal. It looks a little complicated and big for a FIRST robot, but it would be really cool to prototype it.
The main page of his site has links to how a high speed and regular speed CVT would work.