Is joule thief circuit gets overunity?

[poynt99]

Regarding the apparent negative or zero-crossing current trace, sorry Lawrence I do not trust your scope.

I won't believe the current trace crosses the zero-reference line until I see it on a higher-end scope. Why can you not borrow an Instek scope from Mr. Zhou?

[Void]

Try to analyze the strange Board 113.
Board 113 was the strange board withwaveform totally different from other Boards.  Previously, I just treat such Boards as reject.  Now on TK's advice, I keep them and try to analyze them.
Just show the strange waveforms first.  The first comparison of Board and connections appeared correct.
*** It looks like the best thing for me to do is to keep the Board.  Wait until I find the right person to debug.  I do not have the energy, skill and equipment to do a proper job.

Hi Lawrence. Feel free to send this board to me for testing if you like. I can confirm if it is doing anything unusual or not. 

[Void]

I used the data logging feature of my scope to log the data points for the input and output voltage and current waveforms for my 'standard' joule thief circuit, and then used Excel to crunch the numbers to calculate the instantaneous power values and to calculate the overall average power. If anyone sees any errors or problems with the  measurement/calculation method, or with the way I set up the formulas in Excel, let me know. 

I connected the scope probes as shown in the attached schematic. Connecting the probes this way at the input does not invert the input current waveform, to make things simpler. Since I am using a 2 channel scope and just moving the scope probes over between input and output measurements, this probe connection method should be fine. If using a 4 channel scope to measure both input and output waveforms all at once, then you could use the probe connection method that Lawrence has been using, but the input current waveform will be inverted. You would not be able to use my scope probe connection method if connecting all four probes from a four channel scope at the same time to the circuit. 

Vin was set to about 504mV.

Here are the calculated values:
Pin = 2.954mW
Pout = 2.034mW
Efficiency = 68.86%
See the attached .xls files for the actual recorded data points and details of the calculations.

In the attached scope screen shots, the yellow traces are the voltage waveforms, and the blue traces are the current waveforms.

[TinselKoala]

Re 113: more likely to be a "damaged" transistor than wrong toroid connections. There are 4 possible ways to wire the toroid into the circuit, I think two will work and two won't work at all, and of the two that work sometimes one works slightly better but the frequency should be pretty much the same for both working hookups.
I have a special 2n2222 here that is partially failed; it acts somewhat like the 113 board, if I recall correctly. Or the transistor might even be a different type. Did you check the markings?

One thing is certain: two of your resistors are in backwards.

[TinselKoala]

@Void: One problem with the way that you are doing it, with one scope, is that you are comparing a set of input power measurements taken at one time, and for a certain duration, with another set of output power measurements taken at a different time. Hopefully you have at least equated the number of full cycles, which is probably more appropriate than equating number of samples, but I don't really know which would be best.
I've suggested this alternative for those with a single scope: take input and output _current_ at one session, then take input and output _voltage_ at the other session. At least that way you will be taking the input and output readings simultaneously (or nearly so, separated by the scope's sample interval).