Just peaking at this thread for the first time. 

I've good luck with these designs and getting them very close to 100% efficiency in measuring power in to coils and power out via flyback.  I'm using analog current meters and simple DC battery voltage measurement, so can't say the measurements are superb but it's around 90% recovery if the two battery banks are in parallel and not series like Bedini. 

Couple things I've found:
1) High speed is better (+10,000 cycles per min)
2) MOSFET's are more efficient than Transistors (IRF840 or IRF460)
3) Higher Voltage is more efficient than 12V (I've tried 24, 36, 48 volt)
4) Air Core Coils so no iron drag on rotor
5) Bigger Wire Diameter (AWG 20 or less)
6) Swap the batteries every few minutes so their respective voltages stay similar
7) Make sure the magnetic field of magnet extends to the end of coil.  That should help you determine how deep your coil is.

Just sharing some things I found to help..

I have so many changes to this design planned. New shaft size and an overall better build. But I do think this embedded magnet coil concept may have some merit for some kind of use, but I believe I will shelf this project for a bit and return after I have done another design. I do not have a good motor design to compare the performance to. After working on my simple axial flux generator to attach to my projects, I decided what I have wanted to make for a long time is an air core 3 phase dual axial flux motor design. I am certain I can make a nice compact design that will give good speed as well as good torque. I think it will be a good learning experience and great comparison to do efficiency tests against to help me determine if i want to keep going down this road or not. As of now, I have gotten to 39% efficiency with my current design. Before I move forward, I just want to see how good or bad that actually is compared to others. I will revisit this design soon after I have some numbers from another project.