Added second coil.
Rotor is going at 330 RPM on roughly the same peak current. Setup was not optimized for duty cycle or reed position too accurately, as this is just a quick measurement to see if everything is going ok.
Note that the net power is showing bigger hills, which means cemf is kicking in.
(First plot is voltage across two coils, in red, and current, in blue)

More about material T38.
I'm attaching here the specs of toroids that I'm using in current version and also of the alloy T38 (which I guess is MnZn Ferrite). Its EPCOS part no. B64290L618X38, dimensions : 25.3X14.8X10 coated with blue epoxy.

Its Mu_i is 10000 and H(sat) is around 100 A/m from the datasheet. So from
H = N*I/lc
where N is number of turns, I is current and lc is flux path length, we can get the saturation current for say, 100 turns of coil:

100 = 100*Isat/lc
lc = pi*( 25.3+14.8 )/2 = 63 mm = 0.063 m
Isat = 100*0.063/100 = 63 mA

Or say you want Isat = 1 Amp, which gives N = 6.3 or 7 turns. So you would expect that just 7 turns at 1 Amp or 100 turns at 63 mA shall saturate the core completely and the rotor will not be attracted to the core at all. Sounds like free lunch.....

But, don't be surprised if the core refuses to follow the text book in presence of a strong neo. In my case the coil is taking max 2.8 Amps@100 turns and the core still shows strong attraction for the magnets ! I have no idea why this is.......... anyone knows ? Isn't the core supposed to "disappear" for magnets when saturated ?

Either the formula I'm using is wrong or something else is happening here. This is an unknown land for me.

Did some more research on this. It turns out that the above calculation will not apply in presence of an external field. The correct equation is:

B = mu(H+M)
or H = B/mu - M
where M is the Magnetization of the core (which behaves as a magnet in presence of a strong neo magnet)

The Hsat taken from BH-curve is meaningless here as it is measured at M=0 (i.e. for non-magnetized core). So my updated view is that the current required to make the core invisible would be sum of currents required to demagnetize the core and re-magnetize it in the direction of current. I'm guessing that this can be huge.

Rather interesting results here.
These readings were taken at low RPMs (around 75 to 150 or so). And by placing a flat pickup coil of ~500 turns very near to the magnets (which lowers the RPM even more).

One would expect the net efficiency to decrease at low RPM, because the power induced in pickup coil decreases. But the interesting thing is, the power wasted in inductive rise and fall in input side decreases and the cemf is nearly 0. So after deducting the heat losses from the Ein, the net energy transferred is very very tiny.

I guess I have totally replicated the Steorn's Eorbo demo now. Learned many things during this built and my sincere thanks goes to steorn team.

There are still many things to resolve, and as I'm using a crude setup and not so accurate instruments, I can't place my 100% confidence on these results at this time.
.

Data for the third plot above is here....
http://www.overunity.com/index.php?action=downloads;sa=view;down=397

This one is most interesting, the input curve shows no slope at all, which means no energy is being transferred to the rotor.

Now the main issue I'm facing, which can spoil all these pretty curves and render them meaningless. Below are some variations of the plots obtained by varying the value of Rin (=the resistance of the 4 coils + sense resistor) by a tiny bit each time. (Note that the actual value remains what it is, it was changed only in Excel intentionally to see the effect it has)

Assuming Rin=1.465 ohms, Ein gains an upper hand and we have CoE as usual.
Increasing the Rin just by 0.01 ohm, makes both slopes equal. Increasing it a bit further make Ein negative (a gain, instead of a loss, which could be very interesting if true). So the results are above the noise floor but measurement needs strict tolerances.

An error of +/- 0.01 ohm can make results invalid. If you see the steorn's plot, it is a lot neater and would need even higher error margin. One wonders how they measured Rin, and whether any error in Rin measurement was responsible for their claim of OU and they don't know this.

More experiments are needed.

Excellent work!  If you could measure the instantaneous value of the resistor during the experiment, with an additional channel, then we would know if the OU is real or not.  This would allow us to use the instantaneous resistance value instead of a fixed value for the resistor.  As we can see with your data, a very small change in the resistance can make the results invalid.  Do you have a 4 channel scope?

Thanks,


GB

Data in text format (CSV) for anyone who is not an Excel2007 user.

http://www.overunity.com/index.php?action=downloads;sa=view;down=398

@Omega_0,

No need to repeat you're doing a great job. I'm trying to convince prestigious labs to try to reproduce these experiments but so far I'm only meeting with complete lack of willingness. They don't want to hear about it. So, I just wanted to let you know as what the status of this is as of now. I'll continue the effort and will keep you posted.

Wonder if you're following the other thread where I was hoping that somehow this discrepancy will show up theoretically. That would've been the only way to raise any interest at all. So far to no avail. In addition, I'm studying the possible sources of experimental error and it very well may be that the 8 bit scope I have is just of not enough sensitivity for such studies, despite the current probe I got. Even Steorn's equipment may not be of enough accuracy for this kind of claim. This is something that has been bothering me from the get go but that's the most I could afford. What do you think about that?

Update at last...
I've gotten hold of a 4-wire kelvin bridge micro-ohmmeter which can measure up to 4 decimal places and is calibrated using a standard resistance box at the factory and the calibration can be traced back to at least national level standards. Another news is that, now I'm a proud owner of Metglas cores, both MAGAMP and MAGNAPERM (its part no. 2510P4AS and 2510V4AF). A dozen of them

So first thing I did is to measure all the resistances involved. It turns out that my earlier estimates do not match well with new (more reliable values), as I doubted. Now plugging in these values makes OU disappear completely. So a new struggle starts now.

Earlier I was using the scope values to get the Rin (= resistance of sense resistor(Rs)+that of coils), assuming Rs = 1. So I always got IV-I^2R = 0 perfectly when a DC was passed into the coils and current and voltages were measured with the scope. Which may mean that the earlier plots are still valid, but there is no surety that system is OU in reality.

With actual Rs and Rin values measured from micro-ohm meter and pure DC applied, I get a residual net energy (Pout<Pin, See the attached plot-1), which means the readings must be compensated to get a 0 net energy (Pout=Pin, see plot-2)

These baseline measurements are very important. I must measure every resistance at actual operating temperature to get an accurate enough reading and then must also compensate the scope values to get an agreement between micro-ohmmeter and the scope.

After that I must apply these corrections to the pulses measured during an actual run of the device. Its really too difficult and delicate affair.