I noticed that the muller replication is the same sort of principle as bedini, but instead of using one pole on the outside, both the north and south poles get used by turning the magnets sideways and putting coils on the sides instead of the outside.  (guess you could put a set of coils inside so you get outside/inside but anyhow)

the current rotor plate with circular magents reminds me of a SEG.
http://www.youtube.com/watch?v=F_Lk1VFgvUk

Except instead of having the magnets rigid in a plate, they are free to move... There's no way to get torque off of this configuration, but if power generation in a muller is greater than input, then you don't nesseciarly need torque - just apply the excess to a good torque generator like QM motor(?); then this rotor design has 0 parts to wear.

Unlike in the video, I think the coil design should be more like the muller motor being mounted above and below the SEG instead of as a C - I'm not sure, but I'm pretty certain that the middle is like the one direction you can move a magenet vs  acoil and not get a reaction - although because there is a large N/S field opposite to the roller's S/N maybe that causes a flux motion that can be picked up - haven't tested that yet.

Thank you Eric for the ferrite core cutting tip it is a good method to use on this small scale replication.   After this build is made I have had an idea which will simply use magnetic field displacement in a solid state configuration.  This will preclude any moving parts.  My experience designing CNC circuits will be beneficial in sync firing electro magnets, forcing the field(s) to chase the electrons out the conductor(s).