<snip>
    JS:  Thanks for the test on your device running on a single AA,  I really appreciate that. A regular AA non-rechargeable is 600 mA not the 2700mA like Xee uses, as that one has more than 4 times the output current.

According to wikipedia, the weakest non-rechargeable AA battery is 1100 mAh and to get a 600 mAh AA you would need to go with a NiCd rechargeable. Perhaps you are thinking of non-rechargeable AAA batteries which can be as low as 250 mAh according to wikipedia. I realise that wikipedia is not always the most reliable source, so if anyone can prove it wrong then you can feel free to just ignore this post. 

http://en.wikipedia.org/wiki/AA_battery
http://en.wikipedia.org/wiki/AAA_battery

  Thanks for the schematic Xee2 -- very clear.  Will try to replicate tomorrow..

@Nick-- agreed, there are LOTs of circuits out there.  Xee2's is particularly straightforward and attractive.  Note:  One could use a larger-capacity cap if you are worried about missing the "sweet spot", or run over a shorter time --  so that the input voltage is nearly constant.

@dimbulb: 

I have seen that amatuer radio qrp contest clubs
have come up with a meter.
they seem to be
measuring OU also.

Both your statements are intriguing -- can you provide more detail or links?

@Tektron - thanks, will look at this tomorrow as its getting late here now.

   Downunder and All:
   Thank you for the info on the AAs. I was not aware that they can vary from about 1500 to 3000 mA.  And the current is not shown on the batteries themselves. 
So, I'm afraid that unless we all use the same exact battery this test would not be very accurate either.
  But still, supposing that we use the most common and available AA,  a relative run time can be obtained.  A higher farad cap would also help. 
   The idea is to see if there really is any self-running aspect to this circuit, which can recycles the energy within the device, similar to the Joule Ringer.  And if that is not the case then, what advantage there might be over a regular Jtc.
   I personally feel that the additional energy that can keep a device running for months on end, does not come from recycled energy, but instead is drawn from the ambient.  But, that all needs to be proven.

   @ Pirate:
    Three leds were connected in parallel on Koolers BwJt, I don't know if that has any relation to the diode plug (can't spell it) that you mentioned. 
  I think that there may be a point where a higher draw is better, so long as there is resonance.  His circuits are charging instead of discharging the battery. But the other device in the same video does not use three leds, just a single one, and he mentioned that it was brighter than the one with three leds.
  He also mentions that his device are constantly going from 1.2 volts down to 0.6 volts, and back up to 1.2v, and back down the 0,6 volts.
Has anyone else found this to be the case???

Can you provide the link(s) for Kooler's DUT that you're talking about here?  sounds very interesting.

With the sj1 circuit, after 15 hours of running on a single AA battery, the measured voltage is:
Start: 1.623 V
15Hrs: 1.621 V

Hope that is useful to you.  I will keep it running this way a while longer.

Very low power Joule thief. Video at >>  http://www.youtube.com/watch?v=KIQ2D1pqZNc

@Xee2-- a most fascinating circuit, Xee2.  I'm enjoying studying it on my bench.

I had one of the Goldmine toroids already wound with 14 turns of 22 gauge wire, so I used that.  (Good thing I bought ten of these toroids months ago, as they are out-of-supply now.)

R = 2.3 Mohms
MPSA06
Green LED and red LED that I have gave similar Pinputs, using the cap/time method.

Here is what I found exciting -- the operation changes dramatically with Vinput voltage:

2.8 V - 1.65 V, green LED is brightly lit at the start, dims then goes essentially out at ~ 1.65V.  Power consumption is quite high:

Einput = 1/2 C V**2

So Pinput = 1/2 C (Vstart**2 - Vend**2)/time

        = 1/2 10mF (2.8**2 - 1.7**2) / 3.4 seconds   = 7.3 mW = 7300 uW

Not too exciting so far, but I noticed that the LED came BACK ON at approx 1.63 Volts!  Not bright, but clearly glowing.  Note that Xee2 on his schematic specifies 1.366V as Vinput, so this would be in the range of much lower power consumption, Pinput.

Note that the power consumption is hundreds of times LESS:

 Pinput = 1/2 C (Vstart**2 - Vend**2)/time

        = 1/2 10mF (1.431**2 - 1.243**2) / 200 seconds   = 0.0126 mW  = 12.6 uW  !!

The input power has dropped by a factor of over 500.
This is what Nick was talking about, there seems to be a "sweet spot" for operation.

BTW, I noticed the same pattern for the sj1 circuit a while back, dimming, then the LED comes back on and glows for a lot longer thereafter.  I thought it was just a curiosity when I discussed this with smartscarecrow last week, but now I see this as an important effect -- and very dramatic with Xee2's circuit, a huge reduction in Pinput.

With a red LED instead of Green, the Pinput is roughly the same (11 uW from 1.291V to 1.117V in 200s).  The red LED goes out at about 1.44 Vin (from the cap) and back on at lower Pinput, at about 1.406 Vin.

Now this is exciting -- but why the huge drop in Pinput at a critical voltage?? I really don't know, but would like to understand.  Will some of you jump and let's see if we can figure this out?  My GUESS is that the lower Pinput range is the most interesting in terms of seeking OU.

 I checked with my DSO -- before and after the transition voltage to lower Pinput, the frequency of operation is about the same -- about 48 Hz for my build.  (21msec = Period)  But above 1.7Volts, we see that the pulses are large and pulsed DC, whereas below the critical voltage, the pulses are smaller but have a significant AC component. 

Thanks, Xee2! 

@ JouleSeeker

My LED stays on down to capacitor voltage of about 0.8 volts

  Thanks, gentlemen. 
@Nick -- i hope Kooler is OK! 

I added a 1ohm resistor in series with the green LED here, and measured the voltage drop across this resistor (I will attach some waveforms soon), so that V = IR = I since R=1.

Further interesting -- the Pinput did not change much, perhaps down a little with the added 1ohm R:

1.261V to 1.200 V (on 10mF cap) in 80 seconds => 9.4 uW (compared with 11uW without the 1ohm R)

@ JouleSeeker

My LED stays on down to capacitor voltage of about 0.8 volts

Mine also, but the question is -- what happens to the LED lighting if you START at  2 V or above?  Does your LED go "off (or VERY dim)" at roughly 1.67 V  then clearly back ON at approx 1.648V  --as mine does ?

I see the same off-back on effect with a red LED, but at around 1.45 V...  so your device will probably show Off-Back on at some other voltage than 1.67/1.64 -- but probably it will occur, and in the range between 1.35 V and 2.5V.  Would you check?