I have tested John Bedini's Magnetic Gate in my spare time this summer. To reduce the effect of friction I put cylindrical magnets on a rotating holder like this:
But this quickly showed that you need a force to push it into the gate, when it has reached a certain point it pops through the gate. The energy to push it into the gate is higer then the outpush, so when the next cylinrical magnet comes it just repulses at the gate. No overunity here.