CVT Control Method

[CWA]

Hi guys, like many on here, I've been doing a theoretical study on the feasibility and gains involved in running a CVT gearbox with an IC engine.

Currently I feel confident in assessing the various pros and cons, but what I am actually unsure of is how these gearboxes are controlled in real life? Is there a direct linkage between accelerator pedal/lever and engine throttle valve? How are the ratios changed? In my mind, for maximum control, I would like throttle, engine speed and wheel speed to be inputs to some kind of control unit, then electrical output signals sent to a throttle servo and some kind of actuation device for the ratio selector (cones and belts in mind at the moment). But I can't imagine something like, for example, a snowmobile, is this complex. So how are they controlled? I am aware that my description is not legal in the regs, as there must be direct linkage of accelerator pedal to the engine throttle.

I am currently based in the UK, if perhaps I lived somewhere snowy I would just ask to go and have a look at someone's snow-mobile. Unfortunately this isn't an option for me! So I'd appreciate all the info I can get.

Thanks,

Chris

[Canuck Racing]

They primarily use spinning weights, though the more advanced ones are now starting to be electrically actuated. Check out the baja sae forums for more info and plenty of tuning conversation.

[mech5496]

Chris,

you might want to take a look on the Suzuki Burgman SE-CVT. There was a paper on it used on a FSAE 4cyl engine a few years ago, by the Eindhoven team if I can recall correctly. A similar system is used on Hondas' S-Matic CVT.

[tromoly]

If you want to go that route, use an engine that has a PTO output shaft, don't even bother adapting an engine with an internal transmission to it. My school used to adapt I4 bike engines to use a CVT, out of the seven years we did that only one (1!) year did we not have an engine failure of some kind, many times that was due to oiling issues. We switched to a Yamaha Genesis 80FI engine which is designed to use a CVT last year, it was definitely a big improvement over blowing up bike engines.

A buddy of mine who is now a grad student is designing such a system you're talking about, we'll see how it ends up when completed.

[mech5496]

Agree with Troy on the PTO engine; no point in carrying around a built-in gearbox anyway if you go CVT!

[apalrd]

I suggest looking at snowmobile engines and CVT's if you choose to run a CVT. There are several in the 600cc 2cyl range, with varying power outputs, but all generally modern designs, as the snowmobile industry only recently started introducing 4-stroke engines. They usually have a tapered output shaft and a large end bearing to support the CVT side loads, and a matching CVT primary which mounts directly to the engine shaft (it literally can't get any easier for you). Snowmobile CVT's are mechanical devices, tuned with springs and weights. The primary is speed-sensitive and the secondary is torque-sensitive, together they 'find a happy place' in terms of ratio. Below the idle speed the primary will fully let go so you don't need a clutch either. The belt is easily removable by hand with the engine off.

Automotive push-belt CVT's are quite different. They usually contain a torque converter with lock-up clutch (which itself can act as a CVT, at lower efficiency), planetary transmission (to provide reverse, or sometimes 2 forward ratios and reverse), and the variator. Usually it's all controlled with variable-force solenoids. A VFS controls output fluid pressure proportional to VFS current. Usually the control strategy for the planetary transmission is similar to a 'normal' automatic garage shift (engaging the planetary to launch/creep, performing a clutch-to-clutch shift between multiple forward ratios, ...), the torque convertor lock up can be electronically slipped as usual (e.g. during engine bumps such as DFSO or MDS transitions), the variator primary pressure is controlled to be a safety factor above the belt slip force at the given operating torque (minimizing the primary friction losses), and the variator secondary pressure is controlled to hold ratio. The ratio is generally selected based on look up tables.

I have also seen systems where a mechanical-type CVT is pushed by an actuator. Usually the primary is pushed to reduce numerical gear ratio and 'pull down' the engine speed for cruising fuel economy. Not sure how much applicability this has to FSAE, but it could be a neat test.

[dynatune]

One of my 3 final engineering graduate thesis themes I have done in the late 80s, was about a developeing a Control Strategy for a CVT gearbox - which was in those days a hot item. The basic findings of that study was that one needed to use 2 control algorithms which are superimposed in a very specific way. A traditional "fast" PD controller that can react to "events" and a slower "PI" controller that allowed to "oversee" the longer term event. For example if the car is running at a constant speed and the driver wants to accelerate the "initial" adjustment of the PD controller would most likely force the CVT gearbox to kick down going to a higher ratio (lower gear) which would actually make the car slower at that point in time, whereas the slower PI controller would just allow the ratio to be changed in such a manner that acceleration would be the primary objective and slow down the change of gear ratio to such a level that both objectives - increasing ratio and increasing acceleration - were being achieved .. Quite tricky balance to find

Dynatune,
www.dynatune-xl.com

[Dunk Mckay]

To elaborate on the spinning weights method a bit (or put it into different terminology), I've often seen them referred to as centrifugal cams. The profiles and distribution of mass of these are carefully designed to control the exact ratios wanted at specific speeds. The ability to quickly switch out these cams for a different profile/mass distribution, allows you to test different setups, or have different setups for different performance goals (max speed, max efficiency, max acceleration, etc).