Some CVT ideas

Dedicated to the CVT...

Moderators: Dan J, Diesel Dave, Crazymanneil, Stuart

Post Reply
tappy
I luv the smell of Diesel...
Posts: 227
Joined: Sun Jun 14, 2009 11:48 am
Location: Bristol, UK

Some CVT ideas

Post by tappy » Tue Oct 09, 2012 6:49 pm

A couple of people have asked for some info on a CVT system I was designing a few months back.
I'm a bit run off mi feets at work at the mo, and not entirely sure what people want to know, so I'll start with some stuff and we can see where it goes.
I didn't see the design through because after I'd done a fair bit of it, an enfield gearbox came up on fleabay for £50 and I decided to use that. I'd be very interested in making a CVT and seeing how well it works tho'.
The general aim I had was to create a CVT system that:
- Didn't have a huge clutch mechanism hanging off the end of the crank, or an enormous driven sheave, making the bike look odd
- Had a good spread of drive ratios to allow decent acceleration, decent top speed and efficient cruising
- Could be held in a particular ratio - for engine braking or for slippery conditions
- Had good efficiency and belt wear

Design limitations would be:
- Minimum belt working diameter
- Belt / sheave coefficient of friction

It seemed that the main causes of rubber belt wear and failure is insufficient friction at some conditions allowing the belt to slip, so it becomes "burned", leading to further slip & wear. This also reduces efficiency, but more significant would be the loss of drive - rather like a slipping clutch.
Another cause is too much gripping pressure on the belt causing cycling crushing / unloading with every revolution - this also reduces drive efficiency.

In modern car applications the belts are metal plate belts which don't tend to deform elastically so much, so efficiency is improved. These systems can transmit far more power than rubber belt systems, but are very expensive and use very high grip forces. All in all the metal belt systems didn't seem like the right place to start experiments with CVT design.

Conventional rubber belt systems use two systems to both load the belt and adjust the ratio. A set of centrifugal weights in the driving sheave forces the driving sheaves together as engine speed rises, tending in increase drive ratio. An helical cam on the driven sheaves tries to reduce the ratio when more torque is taken off the driven sheaves. This also results in higher pinching / gripping forces on the belt, which helps transmit the torque.
All in all the torque-cam system on the driven sheave isn't bad, but the centrifugal element on the driving sheave leaves a bit to be desired.
At low rpm it doesn't apply enough force to grip the belt properly, resulting in belt slip. At high rpms it applies too much grip, resulting in high belt crushing forces & deformation, reducing efficiency.
When used with a high revving petrol engined ski-doo or similar this belt slip is advantageous - rather like slipping the clutch to pull away on a 2-stroke, but used on a diesel application it could be rather frustrating.

It seemed to me that achieving a more optimal belt grip force at various conditions would improve belt life, and possibly reduce the diameters required to transmit the torque, so making things a bit more compact. Getting rid of the centrifugal mechanism would also reduce some of the bulk.

tappy
I luv the smell of Diesel...
Posts: 227
Joined: Sun Jun 14, 2009 11:48 am
Location: Bristol, UK

Re: Some CVT ideas

Post by tappy » Tue Oct 09, 2012 7:03 pm

For a given belt geometry, the amount of clamping force required to grip the belt depends on 4 things:
- the torque being transmitted by that sheave pair
- the radius at which the belt is running (i.e. the gear ratio)
- the belt "wrap angle"
- the coefficient of friction between the belt and sheaves.

When in a low ratio the belt radius and wrap angle at the driving sheaves are low, needing higher clamping force for a given engine torque.
Due to the low ratio, the driven sheave pair is transmitting a much higher torque, but at a proportionally larger radius so in theory its clamp force needs to be the same. In practise however the larger belt wrap angle means that the sheave force required is actually a fair bit less.

tappy
I luv the smell of Diesel...
Posts: 227
Joined: Sun Jun 14, 2009 11:48 am
Location: Bristol, UK

Re: Some CVT ideas

Post by tappy » Tue Oct 09, 2012 7:09 pm

As the drive ratio increases, the driving sheaves are still transmitting the same engine torque, but now with a larger belt radius. The belt tension is therefore less, so the belt clamp force can be less. In addition, the increasing belt wrap-angle allows the clamp force to be further reduced.
At the back end the torque on the driven sheaves is lower, but the belt radius is proportionately lower, so the belt tension is similar, and because of the reduced belt wrap angle the sheave clamp force required increases a fair bit.

The CVT system *should* spend more time in the mid to high ratios than it does in the low ratios, suggesting that the belt clamping forces at the mid to high ratios should be as close as possible to what is actually needed. This will reduce belt slippage and belt pinching, reducing belt wear and drive losses.

Conversely the CVT system shouldn't spend too much time in low ratios, and whilst excessive slippage is a bad thing, a little bit is probably useful.

tappy
I luv the smell of Diesel...
Posts: 227
Joined: Sun Jun 14, 2009 11:48 am
Location: Bristol, UK

Re: Some CVT ideas

Post by tappy » Tue Oct 09, 2012 7:25 pm

The system I arrived at used a front sheave pair whose separation was fixed by the rider. For "neutral" it was separated so far that the belt would no longer be gripped at all.
The rear sheaves used a set of torque ramps to apply the correct belt clamp force according to the torque load.

So, as the rider closed the front sheaves together the belt would drag slightly and start to apply a small torque to the rear sheave. This small torque would act on the torque ramps, bringing the rear sheaves closer together and starting to both clamp and tension the belt. This increases the grip and the front sheaves, applying more torque, and so it goes on until belt slippage stops and the torque is fully transmitted.
What I don't know is how smoothly this system would "pick-up", but I don't think it should be any more unruly than a conventional centrifugal system that would tend to oscillate.

This system would vary the sheave clamp force according to the belt grip required to transmit the torque at the rear sheaves. It would apply a smaller clamp force for less applied torque, so reducing belt crushing wear & transmission losses.
The actual applied clamp force would have to include a "margin of safety" to account for the larger clamp force required at the front sheaves in low ratios. The ratio would NOT try and change when torque or engine speed was increased or reduced - the ratio would be entirely controlled by the separation of the front sheaves, which is set by the rider.

The torque ramps on the rear sheaves would have both leading and trailing ramps such that when throttling off, the reversal of the appied torque at the driven sheave would still generate a belt clamp load.

The torque ramps would also have a variable angle. As the rear sheaves move closer together, the belt wrap angle increases, so the clamp force (for a given torque) at the rear sheaves can reduce. This requires a steeper ramp angle with less "wedging" effect. As the rear sheaves move further apart and the belt radius and wrap angle decrease, the clamp force required for a given torque increases. This requires a shallow ramp angle with more "wedging" effect.

The blend of applied torque to belt clamp force is a combination of the ramp angles and the diameter at which they act. If they are placed at a larger radius then the use less force to react a given torque so the belt clamp is reduced.

tappy
I luv the smell of Diesel...
Posts: 227
Joined: Sun Jun 14, 2009 11:48 am
Location: Bristol, UK

Re: Some CVT ideas

Post by tappy » Tue Oct 09, 2012 7:40 pm

For some reason I can't find either the spreadsheets or the CAD data from this project. I'll keep looking.
What I can say is that at the time I stopped working on it I had two problems:

The first is how best to manufacture the variable angle torque ramps on the driven sheaves. This isn't so much a problem as just something that needs exploring - one of my colleagues at work has a far better mind than me for making machined features so an hour or two of his time would probably sort it.

The second problem was how to actuate the front sheaves. The clamp forces required can be quite high.
A simple design would use a thrust bearing to push on the outer sheave. One side of the bearing would be spinning at engine speed, and the other side stationary. Given the forces and speeds involved I was concerned that this might generate a fair bit of heat, but I hadn't got as far as checking bearing data properly to estimate that. The larger problem is that if the control mechanism was pushing only on the outer sheave, then the inner sheave would be pushing on the crank end, and the clamp forces required would exceed the allowed crank end load stated in my engine spec (Hatz 2G40).

A better design then would push on both sheaves, such that no net force would be applied along the crank. Indeed the sheave pair could float on the crank for belt alignment (probably do that at the rear tho'). Now though, we'd have TWO bearings, each with one side spinning at engine speed, one side stationary, and each applying the same force, so generating twice as much heat.

A still better system would use a threaded system between the two sheaves, spinning at engine speed, and only interacting with anything stationary when being adjusted (not very much of the time). I was working on arrangements to try and keep this robust and compact at the time I found an enfield gearbox on fleabay.

Another alternative might be to use fluid pressure. I wouldn't want to use gas as this is compressible to tends to apply force, rather than displacement. This CVT system is intended to control the position of the front sheaves, not the force applied, so this would be no use. Hydraulic oil would be OK but the system would need a reasonably high pressure seal spinning at engine speed. I'm sure this is do-able, but again, I haven't looked at it yet.

tappy
I luv the smell of Diesel...
Posts: 227
Joined: Sun Jun 14, 2009 11:48 am
Location: Bristol, UK

Re: Some CVT ideas

Post by tappy » Tue Oct 09, 2012 7:42 pm

This is probably all quite hard to follow with no pictures or numbers, so I'll keep looking for the spreadsheet and CAD work I did so I can post some of that up.

In the mean time, comments are welcomed. Statically the system seems like it should work, but the dynamic loads and behaviours are a little harder to predict. I suspect just making it and trying it would be the best way to find out, but many of you have experience of CVT systems so may have a few ideas :)

tappy
I luv the smell of Diesel...
Posts: 227
Joined: Sun Jun 14, 2009 11:48 am
Location: Bristol, UK

Re: Some CVT ideas

Post by tappy » Tue Oct 09, 2012 8:15 pm

Right, I've found some of the spreadsheet data & notes.
I'd estimated a belt top width based on engine power, speed & torque in comparison to various existing applications.
Gates blet PL30069 has a top width of 28,58mm and an outer circumference of 848mm. Cross referencing, it looked like Dayco P/N HP2009 was interchangeable and had a 26 degree belt side-wall angle.

Using a centre distance between the driving and driven sheaves of 196,85mm, and minimum & maximum front belt diameters of 83mm and 175mm respectively, this gives driven belt diameters of 196 & 111mm respectively. So that's about 7" diameter front sheaves and 8" diameter rear sheaves, which shouldn't be too fugly.
The minimum belt radius is consistent with numbers estimated from data for various Comet drives. If anyone has more data on minimum belt working radii I'd be glad to hear from them.

This gives ratios of 2.36:1 and 0.63:1, which is a total variation of 3.75. Most motorcycle gearboxes seem to have a total variation of around 3, so this looked like a nice healthy spread of ratios to suit our diesels' limited rev ranges.
The intent was that the 0.63:1 ratio would give cruising speed of about 68mph at 2300 rpm. For the Hatz 2G40 that looked about on the limit in terms of engine capability so an incline or headwind would need a down shift, but this could give very economical cruising.
The engine would never pull this ratio at max revs as it would equate to about 110mph! Instead the ratio at 90mph would be about 0.75:1.
This still all requires a lot of final drive reduction though - front & rear sprockets would be somewhere around 13:52 for a 130/80x17 rear tyre.

tappy
I luv the smell of Diesel...
Posts: 227
Joined: Sun Jun 14, 2009 11:48 am
Location: Bristol, UK

Re: Some CVT ideas

Post by tappy » Tue Oct 09, 2012 8:31 pm

Based on the above numbers, and assuming 55Nm torque at the crank, (maximum torque of a Hatz 2G40), the clamping force on the front sheaves would need to be 2870N in the lowest ratio, falling to about 1360N in the highest ratio. The clamping force on the rear sheaves would need to be 6000N in low ratio, falling to about 2459 in the highest ratio.

It looks from the calculations like I assumed that part of the rear sheave clamping was by torque cam, and part of it by spring.
Using torque ramps set at 150mm diameter, combined with a spring of 5N/mm (not much!) required cam angles of around 10 degrees. For the amount of sheave travel required this needed ramps about of about 85 degrees arc, so you could just about fit 2 pairs of leading and trailing torque ramps into the 360 degrees of a full circle.

These values looked to give about 1.3 to 1.5 x the belt clamp required to prevent slip, using a friction coefficient of 0.17 (about 1/6).
i.e. friction coefficient could fall to 0.13 before belt slip occured.

tappy
I luv the smell of Diesel...
Posts: 227
Joined: Sun Jun 14, 2009 11:48 am
Location: Bristol, UK

Re: Some CVT ideas

Post by tappy » Tue Oct 09, 2012 8:35 pm

In all of this the engine torque is assumed to be smooth, whereas there will be some pulsing.
A heavy flywheel will tend to damp out this pulsing - as would a nice heavy, steel set of front pulley sheaves.

One of the problems of a conventional transmission with a diesel engine is that each time you change gear you have to slow that big, heavy flywheel back down, and then put a lot of your engine's power into accelerating the flywheel as well as accelerating the motorbike itself.
A CVT system that allows the engine to rise to max power rpm quickly, and then hold it there whilst slowly changing the ratio avoids this problem.

BertTrack
I luv the smell of Diesel...
Posts: 326
Joined: Tue Jun 05, 2012 3:53 am
Location: Netherlands

Re: Some CVT ideas

Post by BertTrack » Fri Oct 12, 2012 4:43 pm

This is very nice work you're doing!
Image

User avatar
coachgeo
I luv the smell of Diesel...
Posts: 2002
Joined: Sun Feb 22, 2009 6:00 am
Location: USA Ohio, Above Cincinnati, Close to Dayton

Re: Some CVT ideas

Post by coachgeo » Fri Oct 12, 2012 5:11 pm

Mucho thanks. IMHO you should reallly explore this further and PATENT it. Sell it to CVTech or start your own company!

If you would please....... for us more novice engineers. Please picture and/or explaina "torque ramp". Do not know about others........ but for me that is pretty much the only place I got lost since I have no mental image of this.

User avatar
coachgeo
I luv the smell of Diesel...
Posts: 2002
Joined: Sun Feb 22, 2009 6:00 am
Location: USA Ohio, Above Cincinnati, Close to Dayton

Re: Some CVT ideas

Post by coachgeo » Fri Oct 12, 2012 5:16 pm

hmmmmmmmm.... while searching for "torque ramp" to enlighten myself found this patent. Sounds similar to your thoughts but using an Electromagnet to move the sheeves

http://www.google.com/patents/US7300370

BertTrack
I luv the smell of Diesel...
Posts: 326
Joined: Tue Jun 05, 2012 3:53 am
Location: Netherlands

Re: Some CVT ideas

Post by BertTrack » Fri Oct 12, 2012 8:18 pm

When i get the new belt for my Track i'm taking the CVT off it has torque ramps. Shall i make a vid of that explain the function? (CVtech)
Image

User avatar
coachgeo
I luv the smell of Diesel...
Posts: 2002
Joined: Sun Feb 22, 2009 6:00 am
Location: USA Ohio, Above Cincinnati, Close to Dayton

Re: Some CVT ideas

Post by coachgeo » Fri Oct 12, 2012 10:05 pm

BertTrack wrote:When i get the new belt for my Track i'm taking the CVT off it has torque ramps. Shall i make a vid of that explain the function? (CVtech)
. Sure would love to see it. I understand how a CVT works..... just not privy to knowledge that labeled a part or an action inside a CVT as "torque ramp" is all.

Rhynri
I luv the smell of Diesel...
Posts: 165
Joined: Sat Jul 14, 2012 11:16 pm
Location: MN, USA

Re: Some CVT ideas

Post by Rhynri » Mon Nov 05, 2012 5:08 pm

All the fans in the cheering section are wishing you the best of luck with this endevour, Tappy! :mrgreen:

gilburton
I luv the smell of Diesel...
Posts: 755
Joined: Sat Apr 21, 2007 11:35 am
Location: UK northants

Re: Some CVT ideas

Post by gilburton » Tue Nov 06, 2012 11:04 am

I seem to remember a vintage/veteran bike that had a manually adjustable front pulley. It had a shaft running from the pulley up to a hand crank on the tank which you turned to vary the pulley width??
Suzuki Burgman 650 scooters have a tiptronic type push button gear selection as well as normal cvt. It's a bit complicated though.
Surely something between the crude and sophisticated would be ideal??
I think cvt is good if using an industrial engine as it keeps the revs constant which is what these engines were designed to do.
The downside is they are revving away at lower speeds and you can't just cruise along at low revs, saving fuel, as you can on a manual box.
Possibly a foot operated, positive stop,fork arrangement using thrust bearings to vary the front pulley size? This would also have the advantage of reducing front pulley width as the bob weights etc. might not be required?
The only problem with this is the belt acts as the clutch so some sort of clutch or progressive engagement would also have to be incorporated if the weights were removed.

Rhynri
I luv the smell of Diesel...
Posts: 165
Joined: Sat Jul 14, 2012 11:16 pm
Location: MN, USA

Re: Some CVT ideas

Post by Rhynri » Tue Nov 06, 2012 3:06 pm

You could also just have a centrifugal clutch as well with a low locking rpm.

User avatar
coachgeo
I luv the smell of Diesel...
Posts: 2002
Joined: Sun Feb 22, 2009 6:00 am
Location: USA Ohio, Above Cincinnati, Close to Dayton

Re: Some CVT ideas

Post by coachgeo » Tue Nov 06, 2012 4:38 pm

as shown in the auto CVT thread in here..... you could adjust Drive/primary pulley width by air or hydrulic. Also electric. By the way both of these are common and with the right know how can be scavanged out of todays automobiles with CVT tranny... again see Auto Tranny thread with video.

BertTrack
I luv the smell of Diesel...
Posts: 326
Joined: Tue Jun 05, 2012 3:53 am
Location: Netherlands

Re: Some CVT ideas

Post by BertTrack » Tue Nov 06, 2012 7:55 pm

The video i was going to make has to wait. I put a new belt on my track and didn't have to take my secondary apart.

But where a normal cvt Just goes to it's highest ratio based only on rpm.

The torque feedback type (through the secondary) actually increases the secondary spring force and cause it to shift back without loosing rpm.

The two sheaves of the secondary can rotate on the radius. One side is directly connected to the output shaft and the other to a helix which increases or decreases the spring force of the secondary.

When the pull on the sheaves increases the spring will tighten up and push the belt up. Causing the ratio to rise. And the rpm of the engine to increase with the same rpm on the output shaft.
Image

User avatar
coachgeo
I luv the smell of Diesel...
Posts: 2002
Joined: Sun Feb 22, 2009 6:00 am
Location: USA Ohio, Above Cincinnati, Close to Dayton

Re: Some CVT ideas

Post by coachgeo » Wed Nov 07, 2012 8:31 am

BertTrack wrote:...Shall i make a vid of that explain the function? (CVtech)
Well guess Im not understanding how CVT works as well as I thought, In particular the Driven/Secondary pulley.
driven.JPG
http://www.umaine.edu/MechEng/Peterson/ ... report.pdf

So the Torque ramps are the slope's /Ramps that are cut into a cylinder that is part of one of the sheaves of the Driven/Secondary pulley. They play a part that push the sheves apart by RPM's centrifical forces causing (in this case) rollers to move up the ramp (against spirng forces) which spreads the sheaves? Did I get it?
BertTrack wrote:...The two sheaves of the secondary can rotate on the radius. One side is directly connected to the output shaft and the other to a helix which increases or decreases the spring force of the secondary.When the pull (of the belt?) on the sheaves increases the spring will tighten up and push the belt up. Causing the ratio to rise. And the rpm of the engine to increase with the same rpm on the output shaft.
...

helix is the rising circular path of the ramps around the cylinder cut into it of one of the movable sheave's. Like threads in a bolt.

Think Im getting closer to understanding this.

http://en.wikipedia.org/wiki/Helix

Note: scooter CVT's secondary/driven pulley seem to work a little different. Have not gotten this tied down yet either

BertTrack
I luv the smell of Diesel...
Posts: 326
Joined: Tue Jun 05, 2012 3:53 am
Location: Netherlands

Re: Some CVT ideas

Post by BertTrack » Wed Nov 07, 2012 6:18 pm

Ah! If you look at your picture.

The rolls are connected to one of the plates. And the helix material to the other.

The power take off will be directly connected to one of the plates.

The two plates can rotate independently from one another until of course the rolls hit the end of the helix part.

As they rotate it will cause the spring on the secondary cvt to tighten or loosen.

A scooter cvt will not vary the secondary spring force. Which will allow the ratio to shift only based upon the rpm of the primary.


Where as with this secondary system the increase in torque needed (headwind) will allow the cvt to downshift and the rpm of the engine to increase while the output rpm doesn't change much.
Image

Post Reply