Automatic Spherical Continuously Variable Transmission (CVT)

[icks]

I normally lurk and this is probably an obivous and of-course-it-doesn't-work-which-is-why-no-one-has-said-it solution but:

Re: Friction

Can you double up the wheels where the yellow half bushes are and use 2 further balls inbetween the lime green angle connectors? So basically, instead of sandwiching the ball between wheels you sandwich the wheels between balls.

Edited August 19, 2013 by icks

[Technyk32231]

I normally lurk and this is probably an obivous and of-course-it-doesn't-work-which-is-why-no-one-has-said-it solution but:

Re: Friction

Can you double up the wheels where the yellow half bushes are and use 2 further balls inbetween the lime green angle connectors? So basically, instead of sandwiching the ball between wheels you sandwich the wheels between balls.

Then I don't think it would even transfer torque.

[Hrafn]

I normally lurk and this is probably an obivous and of-course-it-doesn't-work-which-is-why-no-one-has-said-it solution but:

Re: Friction

Can you double up the wheels where the yellow half bushes are and use 2 further balls inbetween the lime green angle connectors? So basically, instead of sandwiching the ball between wheels you sandwich the wheels between balls.

Hm, I'm not sure I understand what you mean. The way spherical CVTs work is that a wheel transfers motion to a sphere, which transfers it to another wheel. Since the axes of the wheels are fixed, they can easily be used as inputs and outputs. You seem to be suggesting we use the spheres themselves as inputs and outputs? Is that right? The difficulty there is that the axes of the spheres move, so you need a mechanism to capture the spheres' rotational motion no matter what their orientation is. I've played a bit with using u-joints, constant-velocity joints, and gears to use the sphere as the output (and to get rid of the output wheel) but I haven't worked on it much yet.

[Doc_Brown]

I came up with something, but flickr is broken atm, and cant show ya.

I used old wheels and the ball part fits very nicely! But I cant see any difference in speed when I turn the ball.

[EyesOnly]

Do the spheres have to have much friction to them? There has been a lot of talk of using plastic beads as a replacement for lego balls in GBC modules. Maybe a few beads of some size could help here. The have holes in them that wont fit pins but could if drilled out. But being hard plastic they wont have much grip but if sandwiched between tires it doesn't matter as much as when using a wedge belt.

[Doc_Brown]

This is what I came up with.

But I can't seem to change speed, can you see what I have done wrong?

"LIKE A GLOVE"

Edited August 19, 2013 by Doc_Brown

[icks]

Hmm, i too am apparently a poor writer of technical detail. So i present to you, similairly poor graphical skill!

You were mentioning that the ball was pushing the tyre outwards? So my idea is that the outer idler balls with fixed axes would help alleviate this and maintain contact and friction bwteen the moving axis ball and the inner wheels?

Edited August 19, 2013 by icks

[aol000xw]

I wonder if a sphere without an axle keeping the alignmet can make the automatic CVT function or if it won't tilt correctly. My guess is than given an apropiate mount for the sphere, it will spin like crazy losing some torque but still work. If that is true glass marbles can have amazing traction on rubber (good for torque) and be very slippery on ABS (low friction on mount).

[Hrafn]

This is what I came up with.

But I can't seem to change speed, can you see what I have done wrong?

"LIKE A GLOVE"

It's hard to tell, but my guess is that issue is that the tires are contacting the ball at the wrong point. If they are contacting the ball exactly (or close to) 180 degrees apart from each other, then no matter how the ball is tilted, the radius of rotation about the sphere's axis is the same for both tires. I think that if you move the balls in or out a bit it may work. What holds the sphere in position in your version?

Another possibility is that the tires are contacting the ball over too large an area. In theory, you'd want the tires to contact the ball at an infinitesimally small point; in practice the contact happens over a small area. Different parts of this area want to move at different speeds, so you get "contact spin" and friction. The bigger the area, the worse this problem becomes. The more the tires are pressed against the sphere, the more they're deformed and the larger the contact area gets. The wikipedia article on NuVinci transmissions says:

This is because, for any given contact patch, parts of the ball are going in a slightly different direction and at slightly different speeds than the disc (this phenomenon of traction-type CVTs is referred to as "contact spin"). "The spin velocity (or drill speed) is defined as the difference in the rotational speed of the driving and driven rollers in a direction perpendicular to the contact patch plane. It is caused by the relative difference in surface speeds of both elements across the contact patch and is a major source of power loss in traction drive CVT’s."

Hmm, i too am apparently a poor writer of technical detail. So i present to you, similairly poor graphical skill!

You were mentioning that the ball was pushing the tyre outwards? So my idea is that the outer idler balls with fixed axes would help alleviate this and maintain contact and friction bwteen the moving axis ball and the inner wheels?

:P

That does make sense! Unfortunately I only have 3 wedge belt wheel tires at the moment, so I can't fully test it myself - but maybe I can test one side and see if it makes a difference on that side. Probably not until tomorrow, though - today's busy :(

I wonder if a sphere without an axle keeping the alignmet can make the automatic CVT function or if it won't tilt correctly. My guess is than given an apropiate mount for the sphere, it will spin like crazy losing some torque but still work. If that is true glass marbles can have amazing traction on rubber (good for torque) and be very slippery on ABS (low friction on mount).

Yes, that might work with the right mounting, especially if it includes a central idler wheel that encourages the spheres to spin properly and not just move about crazily. Getting the right amount of contact friction is important, too - my limited experiments with using 6 Zamor spheres held by two large balloon tires (and using an idler wheel) suggests that with low friction, the spheres just act as ball bearings and don't transfer any significant torque.

I'm trying to stay "purist" and use only unmodified Lego pieces, but you're right, glass on rubber has a really high friction coefficient - that's a great idea, and very tempting! Maybe I'll pick up some marbles at the dollar store.

Edited August 19, 2013 by Hrafn

[Zerobricks]

Ball neeeds to spin in more directions, you are limited it to only one plain of movement?

[Hrafn]

Ball neeeds to spin in more directions, you are limited it to only one plain of movement?

Zblj,

I'm not sure I understand what you're asking. Some "wobble" is OK, though not ideal - I think it indicates motion that's not being smoothly transmitted from the input to the output wheel.

You were mentioning that the ball was pushing the tyre outwards? So my idea is that the outer idler balls with fixed axes would help alleviate this and maintain contact and friction bwteen the moving axis ball and the inner wheels?

Icks,

Your basic idea of adding bearings to stabilize the wheel is a good one. In this particular configuration, adding technic ball joints as bearings doesn't work because it introduces too much friction. Each wedge belt wheel is slightly more than 1/2 stud wide, and each ball joint is about 1.4 studs wide, so they jam up against each other too much and there is too much friction for the wheel to spin smoothly. With different wheels, or a different overall CVT geometry, it may well work. There are other parts that can also be used as roller bearings - non-friction pins, pin joiners, small wheels with or without tires, etc. This is definitely a good avenue to explore.

[icks]

I am relieved my idea is sound :p I hope you find the right config and parts to make it work. Look forward to it.

[Doc_Brown]

I am relieved my idea is sound :p I hope you find the right config and parts to make it work. Look forward to it.

Thanks for your diagram! I understand now what I did wrong. Won't work with those old wheels, same cos they were a great fit!

[Lipko]

Well, in princible (as the name also suggests: "Toroidal or roller-based CVT") the disks have to have toroidal grooves in order to work. Simple flat disks won't work as "variable".

Read Wiki

Edited August 21, 2013 by Lipko

[aol000xw]

Well, in princible (as the name also suggests: "Toroidal or roller-based CVT") the disks have to have toroidal grooves in order to work. Simple flat disks won't work as "variable".

Read Wiki

As the main principle behind all this is the relation between angular velocity and linear velocity determined by radius it is posible to make something similar with disks, albeit less eficient as forces are perpendicular.

Instead of tilting a sphere between "wheels" a disk moves transversal to those "wheels". Given a fixed rpm to the input shaft, If the disk moves away from the "input wheel" and closer to the "output wheel" the disk angular speed lowers reducing lineal speed on output and increasing torque. When the disk is centered in relation to both "wheels" the ratio is 1:1