Is joule thief circuit gets overunity?

[picowatt]

TK,

If you look at the B2 trace for Board 80, the LED is on during the narrow (c.a. 70us) positive going peaks during which time the LED is turned on.  As the LED turns off (because the voltage across it is dropping below its turn on V), the current indicated by B2 drops to zero.  This causes the rate at which the voltage is falling on B1 to slow down, hence the somewhat exponential rolloff.  As soon as Q1 turns on again, the voltage at B1 rapidly falls to zero (actuallyit falls to Q1's Vce).

Looking at the waveforms for Pin and Pout for Board 80, it looks like the circuit is 75 to 85% efficient or there abouts.

PW

[TinselKoala]

@PW:
You are confusing me with your "B"s, I don't have the diagram constantly in front of me, so I like to refer to the four possible traces as "input battery voltage", "input current", "output voltage" and "output current". The A1A2A3A4B1B2B3B4 assignments can be read from the schematic when they are needed, as long as we all agree to use the same schematic and the same probe orientation as given.

I've made a short video showing the response of an NPN phototransistor (NTE3037) coupled to the LED of my PCB 2n2222a JT. I verified the output of the phototransistor by also using a photovoltaic cell to show that the PV gives the same kind of waveform as the PT when stimulated by the LED, just at lower amplitude.

I've compared the phototransistor response to both the "input battery voltage" and the "output voltage" taken at the normal points as indicated on the schematic. The deepest part of the valley or dip in the Input Battery Voltage trace coincides with the moment of LED turn-on, and the LED, with a depleted battery, isn't at constant brightness during its "on" time, as evidenced by the nonhorizontal tops of the positive going narrow peaks in the output voltage trace.

This isn't a quantitative measurement, of course.... that would take some calibration effort but is clearly possible. This is just designed to look at the timing of the events. I've been meaning to do this step for some time, to verify just where in the LED light curves, the peaks and valleys of the electrical signals occur.

The video is processing and uploading now and should be viewable in half an hour or so. It will be at:
http://youtu.be/E4k2bsyca4I

ETA:Golly... I've just finished writing the notes/description, and the thing has already had three views. Please, be sure to read the description again, there might be additions that you've missed, whoever you are out there in the audio radiance.

[ltseung888]

I am in phoenix until apr 16, but I am following along.

@poynt99,

While we wait for your results, I shall go to Shenzhen and see if I can borrow some 4-Ch Oscilloscopes.  Hopefully, I can get their experts to calibrate and do the actual hook-ups for me.  PhysicsProf has the 2-CH Atten Scopes.  He will be able to double check my results in the latest corrected way.  I am sure that I can have Mr. Zhou use his many different models of Osilloscopes to display his results (Board 101-200).

My plan is to get the Hong Kong Universities, especially their Physics and Electrical Engineering departments involved again.  They have better scopes and most probably the high-end 4-Ch scopes.  It will take them a few minutes to hook up any one of these "guaranteed OU" boards and record the waveforms and perform the oscilloscope analysis as described.  (I shall update and provide the latest procedure to avoid confusion).

My job is to sow seeds.  We now have the Boards and Procedures that appear to show OU.  We should have more experts join the game.  They will know their oscillocopes better.  They will be able to produce better JTs including better toroids, more layers, more LEDs etc.  Some will no doubt use the 2n3055 with the higher power so as to work above the noise level.

In October 2013, my filed patent with BSI Hong Kong will be published.  In the patent, many conceptual applications are described.  However, I believe many researchers will think of same or similar applications before that date.  The floodgate is open.  With your expected results (and those from PhysicsProf) and others, I feel that much more resources will pour into this lead-out energy research. 

Thank you to you, Tk and others for spending so much time and energy to make such progress.  God Bless.

[TinselKoala]

Well, I appreciate the acknowlegment, but what are you going to do "if" we find that your circuit isn't OU after all, nor is it even especially efficient? (I believe that PW's estimate is "ballpark" accurate.)

Meanwhile I realized that I haven't shown the test point locations on my PCB JT. Except for the use of a metal can 2n2222a, this circuit, even including the toroid windings and material, is, as far as I can tell, the exact same as your boards (without the capacitors). I'm using a battery that is even less powerful than a depleted AAA, though.
Anyhow, FWIW here are the locations of the test points, corresponding to the letter-number system used on Lawrence's most recently posted schematic.

Legend:
A2A3, B2B4 == common circuit reference point, all scope ground leads go here
A1 == input battery voltage
A4 == input current (inverted probe, so negative means "conventional" current flowing)
B1 == output voltage
B3 == output current

[picowatt]

@PW:
You are confusing me with your "B"s, I don't have the diagram constantly in front of me, so I like to refer to the four possible traces as "input battery voltage", "input current", "output voltage" and "output current". The A1A2A3A4B1B2B3B4 assignments can be read from the schematic when they are needed, as long as we all agree to use the same schematic and the same probe orientation as given.

I've made a short video showing the response of an NPN phototransistor (NTE3037) coupled to the LED of my PCB 2n2222a JT. I verified the output of the phototransistor by also using a photovoltaic cell to show that the PV gives the same kind of waveform as the PT when stimulated by the LED, just at lower amplitude.

I've compared the phototransistor response to both the "input battery voltage" and the "output voltage" taken at the normal points as indicated on the schematic. The deepest part of the valley or dip in the Input Battery Voltage trace coincides with the moment of LED turn-on, and the LED, with a depleted battery, isn't at constant brightness during its "on" time, as evidenced by the nonhorizontal tops of the positive going narrow peaks in the output voltage trace.

This isn't a quantitative measurement, of course.... that would take some calibration effort but is clearly possible. This is just designed to look at the timing of the events. I've been meaning to do this step for some time, to verify just where in the LED light curves, the peaks and valleys of the electrical signals occur.

The video is processing and uploading now and should be viewable in half an hour or so. It will be at:
http://youtu.be/E4k2bsyca4I

T,

Really, you can't manage A1=Vin, A2=Iin, B1=Vout, B2=Iout with the A3-4 and B3-4 being the scope grounds connected to the bottom of the input CSR?

OK, I'll quit usig them.

Your video shows that the LED is on when the voltage at Vout is hi, which can only happen when Q1 is off.  that is as I stated previously.

Regarding the second part of the video showing the Vin ripples and Vout:

As Q1 turns off, and Vout subsequently goes hi, the current draw on the battery is switched off.  Vin immediately begins to rise as the battery recovers from the load applied when Q1 was turned on.  This battery (Vin) recovery is the rising portion of the "apparent" negative peak during your Vout "hi" time .  The positive most portion of Vin is the maximum recovered battery voltage.  When Q1 turns on again, Vout is pulled lo.  At the same time, currejt is being drawn from Vin which slowly discharges the battery, hence the slow downward ramp on Vin until Q1 again turns off releasing the load on the battery and again allowing it to recover.

Again, if you look at the second part of your video, you will see that Vin begins to drop as soon as Q1 turns on (and Vout goes lo) and immediately begins to recover as Q1 turns off (and Vout goes hi).  So, actually, the poitive going portion of the Vin ripple coincides with Q1 turning off, Vout going hi, and the LED turning on.  The negative going portion of the Vin ripple, which is much slower and drawn out in time, coincides with Q1 being on, Vout being lo, and the LED being off. 


Your waveforms don't look as sharp and "spikey" as Lawrence's.  What do your Vin and Iin traces look like?

PW