After looking at all the FEMM simulations, Mark Olsen's test runs, and doing a lot of thinking, I feel this is what is needed to get the LPSR to self run.
1. The rotors need to be small in width (leading edge to trailing edge).
The reason for this is that the free floating rotor that is accelerating through the magnet spiral to unlock the stuck rotor never gets into the area of the spiral that has high flux density if the stuck rotor is very wide (leading edge to trailing edge). It hits the trailing edge of the stuck rotor before it can get to this high flux density area.
2. I feel the rotor arms need to have enough mass to have a fly wheel effect though.
3. The magnets need to be located on top of the magnet spiral and positioned next to each other (in near contact) perpendicular to the length of the spiral and a iron (best is low loss Ferro-magnetic material) extension that is solid from the end of the spiral to the start. This will keep the rotor from cogging to the magnet poles. The flux path needs to be through the rotor face from one side to the other, perpendicular to the length of the spiral.
4.The speed of the free floating rotor should be high enough to over come the back drag the trailing end will experience the instant it makes pseudo solid contact with the stuck rotor.
5. You see, it was being pulled from the front and back in different directions before it made contact, but after contact, pulled only backwards.
Thanks,
Butch