PhysicsProf Steven E. Jones circuit shows 8x overunity ?

[JouleSeeker]

[snip] But first the data using batteries in the circuit -- schematic above.

10 Oct 2011:
@20h00, Vinput = 1.934V (two AA's in series),  Voutput = 1.031V (on the AA's in parallel, charging).


@22h048, Vinput = 1.943V,  Voutput = 1.0340V.
So both voltages actually went UP a little, per the Keithley DMM.

The next morning, 11/11/11
@8h00, Vinput = 1.9447V,  Voutput = 1.03460V

@10h44, Vinput = 1.9438V,  Voutput = 1.03464V
Vinput is down, Voutput continues up.

@0:15 (this morning 12 Oct), Vinput = 1.9466V,  Voutput = 1.0356V

@6h21 (this morning), Vinput = 1.9441V,  Voutput = 1.0362V

At this point, with not enough sleep (yawn), I accidentally touched the DMM probe wires in such a way as to discharge the input batteries for about a second; this ends the run.


Observations:  Vinput seems to go up and down, but does not vary much.  Clearly the input power is very small...  The output power is also small with Rb = 0.91Mohms, but the voltage on the two rechargeable AA's on the output climbs rather steadily... and slowly.   

The efficiency?  Voutput rose from 1.031V to 1.0362V so the 2 AA's on the output leg did charge as expected given the direction of the diode in the LED, while the Vinput was roughly the same... but I don't think this is particularly conclusive due to battery-relaxation questions.

Conclusion:  this run was about 46 hours.  This is too long for rapid progress -- to change variables such as Rb, or the transistor or the capacitor, and to see results.   I will go back to capacitors and see what I can learn (in the absence of a calorimeter... sigh...)

I should add that the results are more encouraging when one accounts for the power consumed by the LED on the output leg of the circuit... it MAY be ou, but Pin and Poutput are both small and I'm not too happy with working in this range frankly.

The output LED remained visibly lit throughout the run.  The output was pulsed, I observed (again) on the oscilloscope.  Scope shots and photos later.

Steve

[albertouno]

Joule Seeker
I  did not include all the details before because of the message length.  If you want to see my calculation of Pout for the resistor load test,  here it is:
First Test:  Voltage across resistor :  saw tooth like pulse with rounded top at 7 volts peak, and base of 1.5us duration.  Shape factor  of 0.6 used to estimated average voltage over the pulse duration.  Period of pulses was  5.3us.  The equivalent DC voltage over one period  is then 1.19 volts.  The output power is then 1.19^2/1000, which equals  0.0014 watts.  With a measured input power of .012 watts, the gain is 0.113, not including the LED.   As we know, this measurement is somewhat subjective.  Accuracy of this calculation not estimated,  but  I would have to be in error by a factor of 10 to achieve overunity.  Highely unlikely,  unless you see a calculation error.
Power for the LED used similar method.  Voltage pulse peak 11.0.  The  dc equivalent voltage over the period equals 3.2 volts.  Current in output across a 10 ohm resistor was .0014 amps.  The Power for the LED was estimated at .004 watts.   Total power out was .0059 watts.  Gain for circuit was 0.470.
Second test:  Assuming an  ideal capacitor of 10.3 farads is a reasonable estimate of the battery discharge characteristic (see previous attachment), then the usual ½*C*V^2  equation can be used to estimate the original charge on the battery after  1.5 hours of charging.   The voltage was 1.278 volts at the start of the discharge, hence 8.41 joules was the estimated energy of Pout on the charged  battery.   
I called this a “ballpark” estimate because the two discharge shapes are not identical.  Again this would have to be hugely in error to achieve overunity.  Since my Pout estimation method has not been previously verified, I can’t say what the error range is.  So far, I have not seen any better method used for batteries that can be done in a short period of time.    But not in love with this method. 
If we have a device that can generate over  800% gain, shouldn’t  that stand out in the measurements, even if they were as much as 100% in error? 
Ciao,   Alberto

[JouleSeeker]

  @alberto -- you say your build does not appear to be ou, based on your method of calculating Pin and Pout.  That's probably the case, but if you will look at my first several posts on this thread, you will see that I based the ou calculation on power in and power out as measured using a Tektronix 3032 scope that both multiplies V(t)*I(t) and calculates the instantaneous P(t) and plots that, and then calculates the mean of the power in -- same method for power out.

I do not trust a method in which V and I (not instantaneous) are calculated separately and then multiplied (as you do) to estimate P, because I have found serious discrepancies with that crude method of estimating -- compared with the Tek 3032 method.  The latter "power-waveform" method also has potential errors, and that is why I'm seeking -- and using -- other methods of measurement also.  Note that I claimed "evidence for" ou using the Tek 3032 P(t) method described -- not "demonstration of" ou.

Your battery method is probably better than multiplying estimated V and I (not instantaneous) -- will you pls give the initial and ending voltages for your Vinput and Voutput?  also, if you have a chance, you could then compare with the results I presented above? -- where I DO provide the  initial and ending voltages for  Vinput and Voutput and it looks pretty good actually, IMO.

There may be significant differences in our builds also.  Let's try these tests:
1.  what is the inductance of your bifilar-wound toroid, for each separate winding?  I had L values of approx. 390 and 395 uH on my ferrite-toroid, with 20 windings in each one (total of 40 windings, in all).   Let's compare.

2.  My build lights the LED just fine at 51 kOhms -- bright, at 0.9 Mohms quite bright,
and at 4.6 megaohms -- dimmer but clearly lit.  How does your build function at these resistances on the base resistor?

[JouleSeeker]

I attach a photo of my battery-recharging sj1 circuit to go along with the schematic for the circuit, which I post anew so everyone can see both easily.  Again, the Rb base resistor is 0.9 mega-ohms, and the LED is lit visibly.  L ~ 390 uH for each "half" of the bifilar wound toroid, clearly seen in the photograph.

@alberto -- I ask that you do the same, the schematic and photo for your build, so we can make comparisons.

[albertouno]

Hi again JouleSeeker:
I am running  low on spare time right now, due to my forthcoming trip, but will try to provide some of the  data that you requested.  I would certainly like to see this project successful.  I suspect that my build is somehow different than yours, particularly for the coil, and possibly due to some resistive  breadboard connections.   I did just build a second circuit model with short soldered wires, and it does behave differently with the same coil.  Strange!   I will have to explore this further when I get back.  Used  a newly bought  breadboard for my test.  Do you have a hard wired model?
For the battery test, the voltage at the end of 1.5hrs of circuit charging was 1.278 volts.  After 300 seconds of discharge into 22 ohms, it was 0.123 volts, 0.09 volts after 600 seconds, and 0.067 after 900 seconds of discharge.  A depleted NI-CD  AA600 Malibu  battery was used.
Using my equation for the equivalent capacitor values:   C=T/(R*(ln (v1)-ln(v2))), where  v1= the starting voltage for the discharge, and v2 the subsequent voltages for each time, I get 5.8 farads, 10.3 farads and 13.9 farads respectively.  The spread is larger than I would like, but selected the middle value to calculate the battery energy at start.  My curves are attached again for reference.  Note the relative energy obtained would be proportional to the value selected.
The blue “Goldmine” toroid coils may not be  consistent, or of good quality.  The meter I have only measures to 2 places, and I get .04 mh that’s millihenries on each bifilar winding, with 12 turns each.  However,  that coil drew an unusually large DC input current (20 ma) so I went to a different coil material  for my test.  This one uses a rectangular ferrite core with a measured wide frequency response, and had 16 turns.  The inductance was 0.67mh each side.  It only drew about 4 ma DC at the circuit input.  This DC amounts to the dominant input power on my model.  Your toriod is clearly different.
I will have to try the base resistor tests later.  Used  55K for my test.  In the past, when I went to higher values, I think that the output voltage across a resistive load was also lower, but need to repeat this test.
Hope this helps,
Alberto