@ Jimboot,
Thanks for running these tests, it will help us nail down what is going on a bit better.  But what I am referring to is not just turning off the meter, I mean to disconnect it from the battery all together for a while, while it is running.  Then measure it for a couple seconds and disconnect it again.   Then compare the results to the times you just left it connected the whole time.  That is what I did, and I am seeing a definite pattern.  But interesting enough, it still does not appear to do it with my nicad or my nimh batteries, just the disposable alkaline's so far.  Like I said before though, it is just a theory that the multimeter could be leaching some juice into the battery.  I'm quite sure there are quite a few things going on here we are not seeing yet, that could be throwing our results all over the place making them hard to nail down.

Also, I still do not think I have quite the same voltage coming from the generator side as the battery pulse voltage.  But I definitely see the effect of the bemf on the rotor when I start it.  I can give it a good flick, which I would guess is about a 400-500 rpm spin with my finger, then when I flip the switch to turn on the pulse circuit, the rotor immediately slows down to operating speed.  So there definitely is current moving back to the batteries from the generator side, or I do not think it would slow down when flipped on as quick as it does.  Lens Law is applying the brakes to a certain degree for sure.  Much more tweaking to do.

@CP Can you adjust the orientation of your coils? So all positive sides are closest to the mags?

@Laurent

Thanks for your tests. I think it turns out from them that the high amplitude voltage spikes have little energy content, unfortunately. I say this from the tendency as you measured the increasing voltages in the capacitor for the increasing resistor values, meaning less and less loads. I assume if you try to charge a 100 or 220uF capacitor, the charging speed reduces and so does the final voltage level in the cap.
The 22uF capacitor charged up to 100V contains 0.5*C*V² Joule, i.e. 22uF*10^-6*100*100/2= 0.11Joule. It means that from this capacitor you can take out 0.11W in 1 second and it gets discharged. (I cannot recall what current your setup draws from the 4.5V alkaline, say it is 10mA, then 4.5*10=45mW power that is taken out from the battery in every second. This sounds like OU if you compare 45mW to the 110mW BUT question is how long does it take for the motor to charge the 22uF up to 100V? More than 1 second?
(You wrote 60V went in rapidly and then it went slowly up to 100V, how slowly?  If you consider, say, the 60V went in just during 1 second, the stored energy in the 22uF cap is 22*10^-6*60²/2=0.0396Ws  i.e. 39.6mW in 1 second, now we have underunity wrt the 45mW input in every second.)

rgds, Gyula

PS1  when you write 0 Ohm i.e. zero Ohm, it usually means a dead short, a heavy piece of wire, at least for me, and I do know you meant a no load case i.e. no any resistor at all as a load.    In fact you used an infinitely high resistor in that case, not a 0 Ohm.

PS2 In the meantime Frenky answered how he thinks his no diode setup charges back the battery, it surely involves the spark when created at the switch-off, it is able to conduct via the ionised air gap hence the high amplitude voltage spike (the flyback pulse from his coil) finds its way via the ionised air between the relay contacts back to the battery, a current path is established.  This is Newman motor principle as Frenky also mentioned.  Whether OU or not it remains to be seen....

Edited for typos

@ Gyula
here some result of my latest config
i am now trying  new A with o.3 mm copper   and 9.8 mh and 14 ohm,  and the config is JB config   - + + - - + + -
and so far i understand your questions   i made this test
- I disconnected the plus and minus of the bridge from the circuit
and i put the scope probe  on the plus and minus
basic data 
one 22 micro F 400 volt  across  + and - of the bridge rectifier
 Main battery 4.5 volts nominal alcaline  at now 4.36 volts
1- without cap    that is direct on scope  the spikes can reach 230 volts
2- connect the cap  and it charges very rapidly to up to 60  volts and than slowly  up to 100
3 -i made the different testing  as you suggested      i put resistors acros the cap
and the results
  0    ohms (direct)            resistor     = the cap charges to 100 volts
  10 k ohms   resistor       = the cap charges to 3,5 volts (almost immediatel)y
  20 k ohms  resistor        = the cap charges to 5.95 volts
  47 k ohms  resistor        = the cap charges to about 10 volts
 please let me know what the math says.. ??  is it interesting ??
good night
Laurent

@CP Can you adjust the orientation of your coils? So all positive sides are closest to the mags?

With the setup I have right now I can't without redrilling holes and screwing them back down.  I will go ahead and do that so we have another confirmation of data, but I want to try it with my current setup tonight with a second set of reeds pulsing the entry of the coils also.  But I have been keeping that in mind also and have planned on redoing them anyway.  I will be adding them to movable brackets with nonmagnetic thumb screws for tightening.  That way I can move them around at will and also test that side of your theory.

As far as the alternating polarity of the coils, my honest opinion about that setup is that it is not necessarily out of the ordinary in how it is working.  This is just my theory of course, but it appears to work better because you bring one side of the coil closest to the magnets.  But you must remember, at the same time you are also moving the other side of the coil further away.  In my opinion, there is two generating pulses for every pass of the coil.  One incoming, and one as it leaves.  We are pulsing the coils on only one of them.  The one we are pulsing the coils on we are not seeing as strong effect of the Lenz Law because we are countering it while driving the motor, and collecting part of it on the collapse.  But the other side of the coil is simply getting the full effect of lenz law, because it is just being a generator.  By moving your coils as you have, and changing the polarity they are hooked up, you simply are only making effective use of the one side in which you are pulsing.  The other side that is showing almost all lenz law effect is pulled away from the magnets and now shows nearly none.

I think this is the majority of the reason you are getting a better run for it.  But I also think if we used the bipolar circuit, or at least pulse both when entering and when leaving, it would show the same or better results.  Then we are countering lenz law and capturing in both times per coils, instead of getting dragged down half the time from it.

@ Jimboot,
I'll bet if you changed the coil polarity from -++- -++-  back to the original way -+-+ -+-+ , but you arrange your coils back to normal with all left or all right sides of the coils facing closest, and move the other sides away like you have now.  You will see the same results you are seeing right now.  Having the polarities swapped like you do, could help and hurt either setup due to them canceling each other out.  But they not only cancel the lenz law which is slowing down the setup and a good thing to do, but I think they are also cancelling out the recovery of it, which we do not want.    Setting up the coils the way I'm suggesting, in theory, would keep the polarities from bucking each other and allow the lenz law to be captured on the side we are pulsing, and it would also cut down the lenz law on the side that we are not because that side of the coil will be further from the magnets.  I hope I explained that well enough for understanding.... It's of course only what I THINK is happening.  Any thoughts are welcome...

Hi All,

Here is youtube video I have seen referenced in the Steorn Demo thread. The builder experienced no voltage loss from his batteries during a 24 hour continuous run. He uses magnetic gear for the rotor, and the actual rotor magnets are two cylinder magnets in the middle that interacts with two stator coils. The full circuit consists of only the battery, 2 reed switches, two high speed diodes (they look like as 1N4148 or 1N914 or similar), two coils and one series resistor.

Here is the link: http://www.youtube.com/watch?v=dYgsuJT1zwg 

I do not wish to distract anyone's attention from his own activity of course but I think this video is worth seeing.

rgds, Gyula

I'm not terribly impressed with the Steorn demo.  After several years, millions of dollars and promises of 'unlimited energy', they are producing *way* less than a watt worth of excess energy (21,000 joules over the course of a week = .035 watts).  Enough to give you such a nasty shock if you put it all into a tazer, but it would only run a 50 watt lightbulb for 7 minutes.... then you'd have to wait another week to get another 7 minutes out of that lightbulb.  I think that chasing after other designs will be more productive.

Hi solinear,

I respect your opinion on Steorn but the video I referred to has nothing to do with Steorn!!  Ok?

Thanks,  Gyula

You're right - that's a neat little setup he has built in that video.  I wonder if he could kick-start it with a capacitor with dead batteries and keep it going.

With the setup I have right now I can't without redrilling holes and screwing them back down.  I will go ahead and do that so we have another confirmation of data, but I want to try it with my current setup tonight with a second set of reeds pulsing the entry of the coils also.  But I have been keeping that in mind also and have planned on redoing them anyway.  I will be adding them to movable brackets with nonmagnetic thumb screws for tightening.  That way I can move them around at will and also test that side of your theory.

As far as the alternating polarity of the coils, my honest opinion about that setup is that it is not necessarily out of the ordinary in how it is working.  This is just my theory of course, but it appears to work better because you bring one side of the coil closest to the magnets.  But you must remember, at the same time you are also moving the other side of the coil further away.  In my opinion, there is two generating pulses for every pass of the coil.  One incoming, and one as it leaves.  We are pulsing the coils on only one of them.  The one we are pulsing the coils on we are not seeing as strong effect of the Lenz Law because we are countering it while driving the motor, and collecting part of it on the collapse.  But the other side of the coil is simply getting the full effect of lenz law, because it is just being a generator.  By moving your coils as you have, and changing the polarity they are hooked up, you simply are only making effective use of the one side in which you are pulsing.  The other side that is showing almost all lenz law effect is pulled away from the magnets and now shows nearly none.

I think this is the majority of the reason you are getting a better run for it.  But I also think if we used the bipolar circuit, or at least pulse both when entering and when leaving, it would show the same or better results.  Then we are countering lenz law and capturing in both times per coils, instead of getting dragged down half the time from it.

@ Jimboot,
I'll bet if you changed the coil polarity from -++- -++-  back to the original way -+-+ -+-+ , but you arrange your coils back to normal with all left or all right sides of the coils facing closest, and move the other sides away like you have now.  You will see the same results you are seeing right now.  Having the polarities swapped like you do, could help and hurt either setup due to them canceling each other out.  But they not only cancel the lenz law which is slowing down the setup and a good thing to do, but I think they are also cancelling out the recovery of it, which we do not want.    Setting up the coils the way I'm suggesting, in theory, would keep the polarities from bucking each other and allow the lenz law to be captured on the side we are pulsing, and it would also cut down the lenz law on the side that we are not because that side of the coil will be further from the magnets.  I hope I explained that well enough for understanding.... It's of course only what I THINK is happening.  Any thoughts are welcome...

Thanks CP. That really helps me. I'm starting witth my Orzy tonight. I have some 12mmm Metglas squareloop toroids. I'm hoping the big air coils will saturate them. I'll let you know if I'm dreaming later. I'll hook the coils up the right way this time. Also I'd appreciate anyone's help who can tell me where I should hook up my scope probes to? Thanks