FEMM simulation showing COP 3 and 7

[TinselKoala]

I would think a crankshaft would be best for both the magnets and the bars. Time them using elliptical gearing for the best overlap in operation. Something like a stirling engine but with magnets!
This way all parts are locked together for both push and pull operation and can recover all forces of every component.

The problem of course is that there isn't any source of energy in the magnet-only design. Not like in an alpha Stirling.... in this case the energy source is the difference in temperature between the top and bottom endplates of the displacer cylinder. That is, the Stirling displacer is located in an area of energy flow. No energy flows from permanent magnets, sorry.
http://www.youtube.com/watch?v=aYovJzmCLdw

[Omnibus]

In fact there is no need for an external source of energy because the motor is producing excess energy "out of nothing" from within. From @broli's data it is seen that, the output energy is about 1.15 times the input energy.

Now, that's a small OU but it's a matter of principle as far as violation of CoE is concerned. Science isn't interested in how small or how big is the OU. Even if OU were 1.0000000001 this motor violates CoE and produces excess energy "out of nothing".

That's the pivotal conclusion which adds the motor in question to a number of other motors violating CoE.

A completely separate question is whether or not one can use the produced excess energy for he purposes of a perpetuum mobile. That's an entirely engineering question having nothing at all to do with the fundamental problem of violating CoE. In this specific case the excess energy produced is relatively low and exceptional engineering skills and infrastructure are needed, beyond what almost anyone here can afford, to make a working self-sustaining device.  Therefore, conditions or constructions with greater excess energy are needed to be found out, which will make the manufacturing of a working self-sustaining device easier.

[lumen]

It is easy to see how the motion of the alpha stirling engine is almost what is needed and in fact may be close enough if timed out correctly. By using two cranks linked together with two elliptical gears, the motion could be made to be nearly square. Like the ends of the link arms following around a square block instead of the circular crank.

@Omnibus "the output energy is about 1.15 times the input energy" What is the input energy?

[Omnibus]

@lumen,

The input energy is the energy ewuivalent to the work needed to separate the magnets (observe the device in Butch's sense) plus the energy equivalent to the work needed to pull the cores apart. The enery obtained is equivalent to the work done when the magnets move towards each other spontaneously, attracting each other (the cores being apart) plus the energy equivalent to the work done when the cores close up.

[spinn_MP]

I would not trust FEMM or any other sim program, showing OU.

One of the main problems with Steorn's 2007 Kinetica fiasco (beside the lighting and doubters) was a FEMM program showing OU....