I need a theory for this. Bedini motor phenomona

[yellowsnow2]

The videos shows my Bedini motor following a strange pattern. The conglomeration at the bottom is a neon that lights at about 60vac, a transistor, and some diodes.  From my understanding the neon is to protect the transistor. I don't have a battery connected to charge and I think mainly that is the purpose of the neon.
But why does the neon not light till the second spin?
When I first built it I couldn't get it to work. But it would squeal from being in resonance. When I got it working with more magnets it wouldn't squeal anymore.
And don't forget that Bedini motors deal with regular electricity, fly back electricity, and back EMF.
So it acts differently on the second spin. Is their any theory as to why?
https://www.youtube.com/watch?feature=player_embedded&v=CuIOaYVo8NQ

[TinselKoala]

That's interesting. So let me see if I understand what you are showing.

The motor is sitting there, disconnected. You hook up the battery, then you give the rotor a clockwise starting spin. The motor runs fine and the neon stays out. (If it's an NE-2, which it looks like, it should fire at about 90 volts, not 60, see my video demonstration for how to check this).
Then you stop the motor by disconnecting the battery. You wait a few seconds, then you re-connect the battery and restart with another clockwise starting spin, and this time the neon flashes, brilliantly at first then a bit less so as the motor runs on, but still flashing nicely. And this alternating behaviour, flashing only on the second startup and run, is consistent and repeatable.

Is that a fair description of what you are showing?

I don't know what could be causing it, but my own Bedini motor and similar Joule Thief circuits using a neon in the same circuit location also sometimes behave erratically, but not precisely like that. The only thing I can think of at this point is that some capacitance effect is happening.

Could you please try the following test:

Set everything up so that it is mechanically stable. Ditch the clipleads and connect everything with soldered wires. Run your test again just as you did in that video, with no pause between starts. Then run another test, but pause a minute between the two starts. Does it still behave the same way?

Another test would be to connect the collector (case) of the transistor to a solid Earth ground, and repeat the runs.

[yellowsnow2]

I messed with this some more. But one correction. I didn't disconnect the power leads in between spins. I just stopped it with my hand.
So now it really isn't acting consistently. 
This time it lit the first spin then wouldn't light again after stopping the first spin. After trying several times. And doing all kinds of things differently.
Something weird about this I think. Maybe it is just the clip leads. I'll have to solder some real wires on it.

[TinselKoala]

Ah. So the thing was stopped for a little while with the power connected. Maybe it is the transistor heating up, then. The high voltage behavior of those 2n3055s can be variable, I have found. They "age", and they heat up, and different ones from the same manufacturer can be very different in the HV performance.
Try warming the case of the transistor with a lighter or soldering iron. It won't hurt it, they can get really really hot and still work fine. But try warming it a bit and see if that changes the HV performance. After you solder up the wires, that is.

[SeaMonkey]

But it would squeal from being in resonance. When I got it working with more magnets it wouldn't squeal anymore.

This may be a significant clue.  These
circuits have been known to oscillate
and otherwise function erratically.

TK makes a good point regarding the
power transistor.  It is possible that it
may have some strangeness in it.

Would it be possible to show us a diagram
of your circuit?