wheel-idea

[prajna]

It may help you to think about this to consider one ball on a track.  Next to the ball is a box.  When the track is horizontal the force of the ball is entirely downwards to the track.  When the track is at 45 degrees half the force from the ball is acting on the track and half of it is acting on the box.  When the track is vertical all the force from the ball is acting on the box.  Hence, when a ball is running along the top or bottom of the track (in your cam wheel example) it exerts most of its force on the track and very little on the spokes of the wheel.

We really have to consider this model in five parts:  The balls that are to the left of and above the center of the main circle (top left 'TL'); the balls that are to the left of and below the centre of the main circle (bottom left 'BL'); the balls that are above and right of the centre of the main circle (top right 'TR'); the balls that are to the right and below the centre of the main circle (bottom right 'BR'); the ball on the extreme right of the track.  TL balls are quickly decreasing their backwards force on the spokes as they travel clockwise.  TR balls have a mostly small forwards force on the  spokes that changes little until they reach the extreme right.  BR balls have a mostly small forwards force on the spokes that changes very little.  BL balls are quickly increasing their backwards force.

The key ball is the one on the extreme right, which has a very brief moment when it is exerting a strong forwards force on the spoke at a greater lever arm than the ball on the extreme left but this force is brief and drops off quickly as it rounds the corner.  Then the ball at the extreme left exerts more counter-rotation force because its weight is acting nearly directly down on the spoke, whereas the weight of the ball on the right is acting mostly down on the track.

Eventually the system will find a point where the great force from the balls on the left exactly balances the small forces (albeit at a greater lever arm) on the right.

I hope that is clear now.

[hartiberlin]

What, if you modify the spokes from 12 to 2 o?clock , so that all the upper
balls can directly run to the spoke at 3 o?clock and then get the next spoke
back, when it comes to 3 o?clock ?
This way always a few balls will exert at 3 o?clock a big force against the
left balls.

[hartiberlin]

You coukd do it this way, that all spokes from 12 to 2 o?clock are pulled into the direction
of the axis and at 3 o?clock you push them out again between the balls.
Could be done with short pulses on an electromagnet relay-type linear motor.

[Gregory]

Thanks for the sim and for the explanation, prajna! Good.

I've tested a number of this kind of wheels, and I found that they aren't work. Doesn't matter how do you change the track, I think. It can be traced back to one thing:
Weights by themselfs always roll down on a slope. You can't rise the weights to an upper position (on the left side) with the use of the wheel and gravity only. With other words, it's impossible to create any operating cam track, where the weights change position by gravity only. They can't roll up, only down. So, in this case they always stay in a lower position on the right side, this means game over.

Maybe, if the cam track is not a closed loop, and the weights are "free" from somewhere, where hartiberlin say.
I personally don't hope about that too much. But good idea.

Of course, all these problems don't mean that every gravity device is impossible. Just needed to think on new ways, not on the old ones over and over again...
I think external devices are necessary for the real success. A clever method to rise the weights for the less force, and another method to make them to exert a greater force on the wheel when they are going down.

Keep up the good work!

[prajna]

You mean something like the attached Stefan?  Still won't work.