CVT Belts

[Ben Mies]

First off, this is my first real post I think, I've lurked for a long time, and have chosen to post this here because I've found tons of great info on the board in the past. So, I'm hoping you all can give me some advice.

For those of you that have done any research into CVT design, operation, and tuning. I'd really appreciate any input you might have on belt choice and what is commercially available.


The post is long, and I know CVTs get mixed reviews around here, but I hope you find it interesting and it spurs some discussion.

I'm starting a second semester of research on belt driven CVT design. I've been lurking on the site for a while now and read every CVT, transmission, single cylinder thread I've found. We've read the papers, Olaav's book, had lengthy conversations with engineers at CVT companies, basically as much info as we could find last semester, and finally made a math model of CVT operation (neglecting the majority of belt dynamics and some other, hard to place frictional losses). We've built a dyno to test efficiency (waterbrake on output to final reduction / radial torque transducer on input off of engine).

Well, no matter what kind of info you can get through all of that, CVT design always comes down to the belt. It's the constraining factor leading towards all of the other coupled mechanisms (a real bitch in themselves). Proper belt choice shows huge efficiency gains. One paper noted 15% gain by changing belts over the manufacturers recommendation. The problem is, as it has been for tires before the consortium, that there just isn't adequate info out there on belts. You can find width, angle, length, cross section dimensions, groove depth, the basics, for some of them, but there does not seem to be any source for info available on specific belts when it comes to dynamic considerations. Similar to the combustion pulse, drivetrain inertia, torque pulsing issue with tractive effort at the tire, the same is true for belts on the sheave faces of the CVT. We experienced this on our efficiency dyno. Spikes in strain at the torque transducer are over 6 times the known torque output of the engine. And, with a single cylinder at low rpm, there is certainly an issue in regard to belt against sheave dynamics. Also, the heavy weight of the belt and low tension leads to large losses as the belt resonates in between the two sheaves. This slap is of course excess motion resolved at the sheaves and leads to efficiency loss.

Have any of you heard of any teams doing detailed research into belts? Any info would certainly be appreciated?

I've got a contact at Gates that I'll be looking into shortly, but, for the most part, I've found its best to have looked into other available resources before trying a company.

The CVT I am building this year will be for our Baja vehicle. Last year was our first car and we spent all of our time on everything but the CVT (a common transmission choice in BajaSAE cars). The problem with the available CVTs for that crank diameter is that either they are designed for sleds, have very little design emphasis on efficiency, have limited range, or are not designed for the operating conditions (rapidly changing loads). Those with adequate designs, sleds primarily, often use belts designed for over ten times the power. Of course, when low power is transmitted through the transmission, low efficiency is a result. Double cogged belts are the answer to some of the losses, but are still not sized appropriately for our application.

I'm working with next years FSAE team here, and am using this BajaSAE CVT design as springboard for our use of a CVT next year.

[Ben Mies]

First off, this is my first real post I think, I've lurked for a long time, and have chosen to post this here because I've found tons of great info on the board in the past. So, I'm hoping you all can give me some advice.

For those of you that have done any research into CVT design, operation, and tuning. I'd really appreciate any input you might have on belt choice and what is commercially available.


The post is long, and I know CVTs get mixed reviews around here, but I hope you find it interesting and it spurs some discussion.

I'm starting a second semester of research on belt driven CVT design. I've been lurking on the site for a while now and read every CVT, transmission, single cylinder thread I've found. We've read the papers, Olaav's book, had lengthy conversations with engineers at CVT companies, basically as much info as we could find last semester, and finally made a math model of CVT operation (neglecting the majority of belt dynamics and some other, hard to place frictional losses). We've built a dyno to test efficiency (waterbrake on output to final reduction / radial torque transducer on input off of engine).

Well, no matter what kind of info you can get through all of that, CVT design always comes down to the belt. It's the constraining factor leading towards all of the other coupled mechanisms (a real bitch in themselves). Proper belt choice shows huge efficiency gains. One paper noted 15% gain by changing belts over the manufacturers recommendation. The problem is, as it has been for tires before the consortium, that there just isn't adequate info out there on belts. You can find width, angle, length, cross section dimensions, groove depth, the basics, for some of them, but there does not seem to be any source for info available on specific belts when it comes to dynamic considerations. Similar to the combustion pulse, drivetrain inertia, torque pulsing issue with tractive effort at the tire, the same is true for belts on the sheave faces of the CVT. We experienced this on our efficiency dyno. Spikes in strain at the torque transducer are over 6 times the known torque output of the engine. And, with a single cylinder at low rpm, there is certainly an issue in regard to belt against sheave dynamics. Also, the heavy weight of the belt and low tension leads to large losses as the belt resonates in between the two sheaves. This slap is of course excess motion resolved at the sheaves and leads to efficiency loss.

Have any of you heard of any teams doing detailed research into belts? Any info would certainly be appreciated?

I've got a contact at Gates that I'll be looking into shortly, but, for the most part, I've found its best to have looked into other available resources before trying a company.

The CVT I am building this year will be for our Baja vehicle. Last year was our first car and we spent all of our time on everything but the CVT (a common transmission choice in BajaSAE cars). The problem with the available CVTs for that crank diameter is that either they are designed for sleds, have very little design emphasis on efficiency, have limited range, or are not designed for the operating conditions (rapidly changing loads). Those with adequate designs, sleds primarily, often use belts designed for over ten times the power. Of course, when low power is transmitted through the transmission, low efficiency is a result. Double cogged belts are the answer to some of the losses, but are still not sized appropriately for our application.

I'm working with next years FSAE team here, and am using this BajaSAE CVT design as springboard for our use of a CVT next year.

[BMH]

I would like to hear the conclusions to your research and testing! I am a first year student on the Baja team at my University and we use a CVT set-up (like most uni's do). It would be cool to make CVTs a little more competitive to the slightly increasing use of gearboxes.

[Ben Mies]

hey bryan, TTU was impressive last year. you guys have a hell of a good thing going with baja down there. I remember visiting tech when i was looking at schools, I'm from chattanooga, tn and going to WPI up in Mass now, and being very impressed w/ the baja group.

like i said, we put all of our efforts last year into the chassis, and suspension (all heat treated, tons of fea, damper math models and loading case models, etc). It was a very, very well built car. But, we didnt put more than a couple days into tuning the CVT before competition (Polaris P90). And, that resulted in producing less than 5hp at the rpm we were shifting at (after looking at the dyno curves this year). We got 31st last year, and were up to 13th in endurance before a briggs throttle linkage part broke, unfortunate. we did get rookie of the year award though. Haha, watch out for us this year...lower roll center in front (much, much better roll axis), 30+ pounds weight savings, carbon rear subframe, and tons of attention to driveline efficiency.

anyway, there is certainly alot of improvement to be made over the off the shelf parts. we got a CVTech CVT this year from Canada and will be using that as a back up for the custom one. the stock setup yielded efficiency numbers between 55 and 70% depending on the loading (transient, full, upshift, downshift, etc). We didnt get a chance to test different combos on it though as the high dollar radial torque transducer had to be returned to its rightfull owner. one note was though that the CVTech belt was significantly stiffer in bending than the polaris one. Also, the CVTech CVT is extremely simple in design, old technology really. Weight up ramp, compared to the Polaris cam profile swinging weight design. Using available parts though, we were not able to tune the polaris unit to the proper rpm. We were actually looking at manufacturing tungsten weights for it to bring the shift rpm down. Hope you can benefit from any of that immediate info. I know you guys do tons of testing, any good/ bad experiences w/ different CVT designs?

[rjwoods77]

Ben,

I am a former baja guy and use a briggs v twin with a gaged engineering cvt. You have done good by looking into all that you have. Dearborn has used a comet 790 for the last three years and have done well. I feel the 790 is a piece of shit compared to the cvt's that use the more modern kidney weight or in our case a hex style artic cat clutch(gaged) but that is based off of ideas that still need to be proven so take that as you will. Making a custom belt isnt wise because it costs money and time and you need to concentrate on having a ready supply of good belts to use. You dont need to worry about overall range anymore. No more sled pull to contend with. A cvt with a 1:1 high and a 3:1 low will do you nice. To go into overdrive means you lose efficiency and means you have to gear the final drive really high. A .5 to .75 overdrive on a 8000rpm engine with a top speed of 60mph makes for huuuugggeeeee gearing (approx 9:1) in a single stage because if you use two stages with a cvt in formula you should be shot. Makes sense on a 3800rpm baja engine but not formula. Rapidly changing loads are exactly what a drag and snowmobile clutch do. The good ones have rollerized caged helixes on the secondarys (read: no shitty plastic buttons) which coupled with proper primary setups can net you a rediculously quick back shift. Effiency is based off of friction losses and that is what the old comets and cvtechs have in droves. There happens to be a cvt that fits the bill perfect for formula and that is the gaged engineering dominator gx.( http://gagedengineering.com/ ) The use a dual cog with 9 c-to-c (not as small as a 8.16 for a 790 but whatever) that is super stuff and the primarie is 5 to 5.5 inches in diameter and the secondary is 7 to 7.5 inches in diameter. We are super clse to running the car and have already figured out how to get the cvt to enage at off idle speeds by using tungsten weights. Long story short is look at their website and see what they have. Bill at Gaged is a real nice guy and offers good customer service. Tell him I sent you to get his stuff. Just to note the fastest jr drag car out ther runs their stuff and does 330 feet in 4.23sec a 450lbs with driver. The fastest formula cars are 4 seconds at 265 feet.
Rob Woods

[drivetrainUW-Platt]

Not all cvts use belts, some use balls in a case...

What engine are you considering that you want a cvt. If it has a gearbox on it, it would be going the wrong direction to chop it off and put a big bulky belt/pullies IMO.

[Ben Mies]

For next year, we are seriously considering using a Jawa single cylinder found in flat track motorcycles. Other singles/doubles without a tranny are also being considered.

Yeah, the torroidal CVT (two curved conical shapes w/ doughnuts in between) is a great option as are the ball type CVTs which you speak of. The control methods really get to me though. I tend to favor the belt CVT when considering it for these applications for the relative simplicity of the control mechanism.

I absolutely respect your opinion when it comes to the gearbox vs. cvt debate. Both certainly have their pros and cons. In the order of time and development, sticking with an engine with a built in gearbox would certainly come out on top in most arguments. The downsides to CVTs that I see are the loss of mechanical efficiency when comparing transmissions alone, packaging, and weight (as the CVT mechanism is no longer a stressed member of the engine case itself). But...I'm not convinced that we would be worse off by choosing a CVT over a standard gearbox given the competition operating conditions, relative lack of driver training in our teams, its ability to stay at the peak of the power band throughout the course, and proper attention to tuning/designing for an efficient mechanism. That is just my opinion. I've had a fair amount of seat time in a few FSAE cars, been around them for a some years, and done some basic analysis of vehicle speed on course and such. But, I've never built/competed in FSAE.

[Ben Mies]

Rob -

Thanks for the input. Those gaged cvts look great, I actually hadn't seen that yet. I will definitely be in contact with Bill.

We have no misconceptions about manufacturing our own belt. Its a fruitless idea and something that we'll never be able to do here given our manufacturing limitations, lack of resources, money, and time of course. These companies spend years designing/testing belts.

We've strayed from Comet because of the "elderly" design and extreme overdrive in some of them. We went w/ polaris because of its kidney weight design and the success of people's ability to tune it. And, finally we picked up the CVTech one because teams are succesful with it, its easy to reverse engineer, and with the baja sponsorship, its wicked cheap. Although, its design, again, is archaic.

In the past, I've heard bad things about the drag clutches with regard to backshifting/downshifting namely because of the steep helix angles imposing high mechanical impedence as the mechanism tries to drive the roller back up the ramp. There are other factors to backshift / downshift performance of course. A steeper ramp is harder to push something up, makes sense. One thought here is to change the radius of the helix instead of the angle. Think a cone instead of a cylinder. You get the same breakdown of the torque reaction forces, but due to a radius change instead of a steeper angle. Of course, any friction, in any part (aside from the belt) kills efficiency and system reaction as the friction damps the reaction abilities of the system. We'll be using roller bearings and low friction materials for bushings everywhere possible over those damn plastic buttons.

Thanks for the info.

The belt is my biggest concern at the moment. But it looks like Bill at Gaged might have something up my alley or at least some ideas. I'll be in contact with him soon. Thanks.

[rjwoods77]

I worked with Bill to have him make the caged helix that he now sells as standard. I have a video of University @ Buffalo's baja car making horrible sounds. You might be familiar with this. If you are in Ohio and doing the Dirty Dozen or any set of close whoops if you stay in the gas when the rear wheels are in the air and then contact the ground, the helix will separate from the buttons due the clutch sensing an overrevving(overdrive) and as soon as it feels the contact the upper helix slams into the lower helix and destroys the buttons. A caged roller helix solves this problems because it controls this seperation. Snowmobiles started doing this years ago when they were breaking/wearing things out from sudden torque reversals. I showed Bill this video and told him that he should make one and he did. I turns out that he has had problems in the past with secondaries when jr drag cars "pedaled" to not run over the time index. In doing so they were loading and unloading the clutch severely. It was possible with the right chain of events to get the belt to jam in the sheeves of the secondary when the secondary was in it smallest diameter which would lock up the rear tires of the drag rail at 65mph which caused a couple to flip or so my memeory serves correctly. So he now sells the secondary with a caged secondary. He offers about 6 different helix angles so and I am sure if you beg/pay him hard enough that you could get a custom made. To avoid fulcrum arm separation from the ramps from "pedaling" he has helper springs on the arms to keep them in contact with the ramps in on-off throttle situations. They are not standard but he has them. Some baja team in south florida is using there setup and doing well with it I heard. You will have to make some weights out of tungsten to get the thing to engage low enough which we just accomplished. Car should be on the road in 2-3 weeks hopefully and we can start tuning the engine/clutch combo. I would be willing to help you out with things on your end if you would be willing to show me some of your research. If you end up using it I would love to get our mutual teams talking with each other. Bill makes a great product and I can't think of a better way to pay back his sponsorship then with getting him some sales and breaking successfully into a new market for him. I think his cvt is the best one for formula and baja purposes and with great success with his products in our series would make more people gravitate toward cvts since the black magic of cvts wouldnt be so black magic anymore. Feel free to give me a call on my cell after I get back home from my work trip on Feb 2nd.

Rob Woods
585-615-8474

[Ben Mies]

I know very well what you are talking about on the slamming action over whoops. We did not experience broken buttons like you guys (I've talked to another Buffalo baja member about this actually over on minibuggy.com), but the buttons and faces of the helix did exhibit signs of wear. The noise is awful. And, we've certainly put the car through its paces in the woops.

Sidenote: did some really cool suspension analysis on our car with linear potentiometers (the string kind), accelerometers, and some measured, even 2 foot whoops at varying speeds. The damper dynamics over whoops like that is pure evil. Especially when these baja cars just don't have the power to "skate" over them.

Like I said, I will be calling Bill tomorrow. Sounds like a great product and an awesome source.

I will also be in contact with you as well and would be more than willing to share info as we get it.

One note on his CVTs... I have noticed that the inner diameter of the drive pulley is under the outer diameter of our crankshaft for Baja. Now, with FSAE, I'd certainly be able to turn the crank diameter down to accommodate, but with Baja, have you all opened up the drive pulley inner diameter to work on that crank?

[rjwoods77]

I will send you the video if you want. You must be talking about Justin S. We had Bill make a post for a 1" shaft for the briggs twin. The secondary we are using has a 3/4" shaft but could easilt go to 1" for a stiffer jackshaft. Another customer service bonus from Gaged Engineering. What engine are you htinking of running in FSAE? I would highly suggest looking into the briggs twin we are using. There arent too many engines with a pto shaft and that is the best one to use in my opinion.

[Ben Mies]

I'd love to get ahold of the video. My email is bmies@wpi.edu. Thanks. We plan to do some on board video, up close, of the CVT this year during operation in varying scenarios. Hopefully that will yield some interesting results.

I spoke with Bill briefly today. Unfortunately my phone crapped out. Seems like a great guy and a good contact. Thanks.

On the shaft sizing. Thats exactly what I was looking for thanks.

We're thinking about a Jawa single cylinder designed for flat track motorcycles. Namely for it being air cooled and without a transmission. Still being in the rough planning stages, I think we're open to engine suggestions.

On the PTO shaft. I am accustomed to a PTO shaft being referred to when talking about older 4x4 Transfer Cases and farm implements. What are you referring to as a PTO shaft? Does the briggs have a primary reduction pre-output shaft?? That would certainly make life easier, haha.

[rjwoods77]

Pto=crank snout. I really dig the jawa engine but there a bunch of unkowns that need answering. If you look at their manuals they have remarkably short service intervals which in a racing sense doesnt matter but if you are designing for the comp then it would pose a possible serious problem have a short life span engine. That and it is designed for methanol, doesnt have a starter, enough oil capacity and needs a fan and shrouds for cooling since it is buried in the back of the car. Briggs twin solves those issues however it is heavier, bigger,etc. Ill email it to you when i get back home from freezing ass Kapuskasing.

[Brian Barnhill]

Ben,

I don't know if you remember, but I actually used to race flat-track (as well as moto/arena and supercross) so I've had some experience with these engines (and the reliability issues.)

I've actually been thinking about some of it a bit, and have few thoughts. I'll be in our shop area all next week if you want to stop by and chat about it.

~Brian

[rjwoods77]

Brian,

I have been in contact with Jawa in CZ for a while now bouncing emails back and forth. With your thoughts and the info and drawings I have garnered from Jawa I would be interested in helping out conceptually since I have devoted a significant amount of thought into using the engine if you and Be are interested.

[Ben Mies]

Brian and Rob-

Brian, I spoke with you a bit today, but didn't get a chance to talk about engines. I'll be over that way again sometime this week I'm sure.

Rob, good insight on the Jawa engine. Of course, looking at it from the initial standpoint, it appears great for our application. Light, powerful, and w/o a transmission. But, those are serious concerns. Its damn hard to locate any engine that fits the bill i.e. 4 stroke, under 610cc, air/oil cooled, compact as possible, and without a transmission. If there were only a database of all engines out there with these kinds of parameters. Haha. Pipedreams.

I'm still cranking on the CVT stuff and seem to have made some headway in the use of a .780" wide belt for this Baja CVT. I just need some good data on the belt dynamics to size the control mechanism. Sheave sizes and shift range are looking good.

[rjwoods77]

The only engines that are air cooled that have a pto are speedway engines (Jawa,Gm,etc) and industrial engines(briggs,honda,onan,etc). That is what years of searching netted me.