1. "Three phases are about as many as you need to pull this off. More phases just means more rectifiers and interconnection but no more power, since you can put the same amount of copper under the magnets with three."
Correct me if I am wrong, but part of what makes Windstuffnow's design work so well is the little amount of turns per coil. In his last example he made ten turns per coil, and is contemplating wether or not he should have only have made eight turns if I remember correctly... I would venture to guess that by doing so decreases the amount of resistance.
That last sentence is where you go wrong. Yes, fewer turns of thicker wire mean less resistance. But they don't mean less copper losses. What makes Windstuffnow's genny's good is that he used a high total cross-section of copper per coil.
Imagine a hundred-turn coil that produces one volt per turn and one amp through the coil. It will have a certain amount of power lost to resistive heating.
Now split it in the middle, into two fifty-turn coils, and connect them in parallel to get two amps at 50 volts. Same windings (though connected differently), same current in the wire, same losses, same power output (though at a different current/voltage tradeoff).
Now suppose the particular windings in the two 50-turn coils are side-by-side all the way. No voltage between them. So remove the insulation between them and weld them into a single figure-8 wire. No change. Reshape the wire from a figure-8 into a circle with the same cross-section. Still no change. Split the coils again and reconnect for 25 volts and four amps, 20 volts and five amps, or one volt and a hundred amps. No change yet again.
You CAN reduce your copper losses by making the wire thicker and keeping the number of turns the same. That reduces heating losses in the coil. If your limit was how much you can heat the coil you can thus get more power.
But that also makes the coil fatter. You can only go so far before something else has to change - like increasing the magnet spacing - to make room. Increasing magnet spacing reduces power. You can get it back by using stronger magnets or spinning them faster for more voltage. But eventually you goof something up that moves your limit from heating the coils to something else you can't fix (like making your prop too inefficient.)
So let's not fool around with anything in his excelent design except the number of turns and wire thickness. As long as we keep the wire cross-section times the number of turns constant we can play with the turn number to trade voltage for current and not affect power output, heating loss, or (unless we get down to a few turns of bussbar) how the coil fits into the space. Even conduction of heat out of the coil remains unaffected (though finer wire may transfer a little more heat to the air if the ends of the coil are exposed).
And that lets us adjust the designed output voltage/current tradeoff of the generator to our liking, without affecting the total power delivered.
Is my explanation clear?
