LTJT - poynt99 Tests Assembled Unit

[poynt99]

Pics of the assembled unit that arrived yesterday.

I added battery holder leads roughly the same length as what Lawrence had in his photo.

There was one cold solder joint that was loose and unconnected. I did a quick job of getting the solder to flow on the mag wire, careful not to disturb the joint too much. Another cold joint was left alone, as it appears there must be some "good" contact there, good enough for the base drive to get through anyway.

.99

[poynt99]

Some test results of this unit:

TEK00017.PNG indicates an average INPUT power of 63.31mW.

TEK00019.PNG indicates an average OUTPUT power of 237.1mW/10 = 23.7mW.

The light output from these tiny super-brights is impressive.

Lawrence, let me know if you feel I am doing anything wrong. The Po/Pi ratio is quite low compared to your findings.

.99

[MileHigh]

Some test results of this unit:

TEK00017.PNG indicates an average INPUT power of 63.31mW.

TEK00019.PNG indicates an average OUTPUT power of 237.1mW/10 = 23.7mW.

The light output from these tiny super-brights is impressive.

Lawrence, let me know if you feel I am doing anything wrong. The Po/Pi ratio is quite low compared to your findings.

.99

Lawrence's claim that this pre-assembled LTJT demonstrates over unity in it's original configuration has been refuted.

There are important lessons to learn here with respect to how to take proper measurements, to understanding how electronic circuitry works, and to understanding the energy dynamics of electronic circuits.

MileHigh

[ltseung888]

Some test results of this unit:

TEK00017.PNG indicates an average INPUT power of 63.31mW.

TEK00019.PNG indicates an average OUTPUT power of 237.1mW/10 = 23.7mW.

The light output from these tiny super-brights is impressive.

Lawrence, let me know if you feel I am doing anything wrong. The Po/Pi ratio is quite low compared to your findings.

.99

Dear Poynt99,

Now try the following:

1.   Use a DC Power Supply*** and lower the voltage.  In our test, the Vrms was lowered to 80mV.  The Vpp was 600 mV.  Please compare your results with the ones I supplied.

2.   Please take the Vpp readings as well.  That will give grounds for clearer comparison.  When the Input Vpp was 600mV, the Output Vpp was 10.4V.

3.   The dip into the negative region for our Power Curve was much higher.  If the correct explanation for negative area were – more feedback to the source, your feedback to the source is much lower than ours.  May be the lowering of the input power will explain that.

Continue to have fun.

You can start to assemble the other LTJT and vary the parameters much more.

*** You can get a really cheap DC Power Supply for less than the Federal Express Mail Price.  I paid US$45 for shipping.  I believe the cheap DC Power supply from China cost less than that.

[Harvey]

Hi Guys,

Just wanted to point out the non linearity of both the phase angle and the Voltage, Current curves (and thus the resulting power).

Q. Can you identify where the load LED turns on and where it turns off?

Q. Is it possible to have a negative power where both the Voltage and Current are positive? (Hint, notice slope of power trace compared to Voltage and Current)

Q. How do we know if there is capacitance involved? (Hint "ELI the ICE Man") Where is this capacitance if it exists?

Q. What would happen to the PO / PI ratio if we used a One Megohm output CSR? How about 100K? 100 Ohms?

Q. Would using a 0.1 Ohm output CSR allow more current to flow in the output stage?

Q. Does a single test in a qualified lab substantiate or refute a claim?

Q. Is it ok to ground the output section to the input section through the scope grounds?

[poynt99]

Hi Guys,

Just wanted to point out the non linearity of both the phase angle and the Voltage, Current curves (and thus the resulting power).

Q. Can you identify where the load LED turns on and where it turns off?

Positive current indicates when the LED is forward-biased, and the negative current when reversed-biased. These LED's are quite "leaky" in the reverse bias condition, which is why there is power dissipation even when the LED is "OFF". So in terms of the question, the load is always ON in this case, it's just a matter of degree, depending on the bias.

Q. Is it possible to have a negative power where both the Voltage and Current are positive? (Hint, notice slope of power trace compared to Voltage and Current)

No. p(t) is simply the product of v(t) and i(t), and the basic math rules apply.

Q. How do we know if there is capacitance involved? (Hint "ELI the ICE Man") Where is this capacitance if it exists?

It can be argued that capacitance is always involved. There is inter-winding capacitance, capacitance in the junctions and leads of the transistor, capacitance across the LED junction, capacitance to the bench and environment, etc.

Q. What would happen to the PO / PI ratio if we used a One Megohm output CSR? How about 100K? 100 Ohms?

There should be better power transfer as the impedance gets closer to that of the input side, i.e. about 1 Ohm or so in this case.

Q. Would using a 0.1 Ohm output CSR allow more current to flow in the output stage?

In theory, yes.

Q. Does a single test in a qualified lab substantiate or refute a claim?

In a perfect world, yes. However, someone should always double-check the test setup, the methods, and the results.

Q. Is it ok to ground the output section to the input section through the scope grounds?

In theory, yes.

In both the simulation and the actual test, it made no difference to the wave forms or measured power levels.


.99

[wattsup]

@poynt99

Maybe some dumb questions but here I go anyways.

I see in the photos that you have three probes on the device at the same time. On a low voltage pulsing device such as this any probe may effect the overall function simply by extending the conductive paths.

With the three probes (1,2 and 3) on and while looking at the three waveforms....

1) what happens to waveform 2 and 3 if you removed probe 1
2) what happens to waveform 1 and 3 if you removed probe 2
3) what happens to waveform 1 and 2 if you removed probe 3
4) Under the three above conditions, can you notice a change in the brightness of the LEDS.
If the answer is yes to 1, 2 or 3, then the probes are effecting the circuit.

5) Then, is it possible to use probe 1, grab the screen, remove probe 1 and use probe 2, grab the screen, then remove probe 2 and use probe 3 and grab the screen again. This way only one probe is on the device at any given time.

6) Is it possible to probe without the ground leads of the probes connected?

7) Also, are you probing before or after the feed LED. Seems to me anything before the feed led or on the feed led is irrelevant since it is only the energy after the feed led or across the drive coil that should be considered and compared to the output.

Maybe one last thing or definitely a repetition. If you have enough juice in there to light up LEDs, then you have more then enough to pass a germanium diode and into a capacitor tank of low voltage but high enough uF. I would really consider running this on a capdcap method because it seems those LEDs will skew the results. Or at best remove the feed LED and replace it with a germanium diode going in the same direction as the LED was.

wattsup

I know these are many points so there is no real need to answer them all.

[poynt99]

wattsup,

There are 4 probes used at all times for these measurements. Two for the INPUT power (scope 1) and two for the OUTPUT power (scope 2). The probes are placed as per this drawing, and they are indicated by P1T/P1G, P2T/P2G (T being TIP, and G being GND) and so on.

The tests are being performed as per the parameters measured by Lawrence and his team of experimenters. The device should not be modified from the original in order to properly compare results.

.99

[ltseung888]

DC power supply used in Hong Kong

It is likely that the results of this particular prototype were obtained with the DC Power Supply as shown in the diagram.

We used both AA battery and the DC Power Supply for our testing.  When the Input Voltage was way below 1.0V, there was a good chance that the DC Power Supply was used.

However, we do not necessarilyy need to use the DC Power Supply to get very high COP values as shown in the Hong Kong University testing in the second photograph.  If Poynt99 can get a DC power supply that can drop to 600mV peak-to-peak, the chance of hitting pseudo resonance is much higher than using an AA battery.

If not, Poynt99 can build a prototype from the components supplied and try to produce the same result as that at Hong Kong University.  Tuning can be fun.  FLEET is not the very tolerant Joule Thief!

Continue to have fun.

[poynt99]

I have a variable DC power supply.

I will connect it up and see how low a supply voltage the assembled unit will still operate. Then I'll take some measurements again.

.99

[ltseung888]

I have a variable DC power supply.

I will connect it up and see how low a supply voltage the assembled unit will still operate. Then I'll take some measurements again.

.99

Great.  Looks like you are the only one in OUR forum who has the necessary equipment to have fun.

[MileHigh]

Just for reference, I am reposting Lawrence's data for the fully assembled prototype that Poynt is testing:

Dear Poynt99,

This particular prototype showed Tseung FLEET Comparison Index of 64 rms when built on Oct 10, 2010 in Hong Kong.  It is soldered and not much tuning can be done.

The details are as follows:

Windings:
11 turns Joule Thief type
22 turns Transformer type
0.5 mm diameter wire
Frequency = 147 KHz

Input
Channel 1 (Instantaneous Voltage)
         Vpp  = 600 mV
       Vrms  = 80 mV
Channel 2 (Instantaneous Current, 1 ohm resistor)
         Vpp = 188 mV
       Vrms = 60 mV

Output
Channel 1 (Instantaneous Voltage)
       Vpp  = 10.4 V
     Vrms  = 3.4 V
Channel 2 (Instantaneous Current 10 Ohm resistor)
      Vpp  = 324 mV
    Vrms  = 92 mV

Calculations:
   Input Power pp  = 0.6 x 0.188 = 0.1128 watt
  Input Power rms = 0.08 x 0.06 = 0.0048 watt

  Output Power pp  = (10.4 x 0.324)/10 = 3.369 watt
Output Power rms = (3.4 x0.092)/10 = 0.0306 watt

COPpp   = 30
COPrms = 64

The verification is simply to display the waveforms and check whether the true COP is greater than 1.  The actual numbers are unlikely to come out identical as in all pseudo resonance experiments.

*** Please do not modify this particular prototype as it may be shipped to other sites for additional verification later.  Use the components package to build another one.  Thank you.

Lawrence

Lawrence,

Assuming that the copy/paste above is for the LTJT that Poynt is working with, I have some questions about this data:

Who generated this data?  Was it you or Aaron Quant?  Perhaps somebody else?

Input
Channel 1 (Instantaneous Voltage)
         Vpp  = 600 mV
       Vrms  = 80 mV

Whether you used a 1.5-volt AA battery or a power supply set to 600 mV as the power source, quoting peak-to-peak or RMS values for a DC power source doesn't really make sense.  As you can see in Poyn't waveforms, the output voltage for the battery is more or less a constant DC value.  The voltage drops slightly when Joule Thief is drawing current from the battery.  This is all normal.

So the question for you is why are you quoting peak-to-peak and RMS values for what is supposed to be a DC voltage source that powers the Joule Thief?

Thanks,

MileHigh

[PhysicsProf]

  I went to the waveforms posted by .99 in order to check things another way.

Both the Input and Output power waveforms display a characteristic triangular pattern, for which it is rather easy to calculate the Energy per cycle.  The energy can be calculated simply from 1/2*base*height, where the base is the length of the pulse in microseconds (Time) and the height is the Power, in mV*V or V*V.  For the input, this is approximately:
Einput ~ 1/2 * 4 uS * 150 mV*V ~ 300 mV*V * uS

The output power waveform also shows a small amount of power over about 4us, and the total Energy per cycle is roughly:
Eoutput ~ 1/2 * 1.2 uS * 1800 mV*V/10  + 4uS*15mV*V/10 ~ 114mV*V * uS

The ratio Eoutput / Einput by this method for one cycle is approximately 114/300 = 0.38 .

Using the average Power input and output method, .99's method, we get PoutAve/PinAve = 23.7/63.1 = 0.38 .

I conclude that the numbers check out and I do not challenge .99's method.  

However, his coefficient will improve if he does as I suggested, for instance-- simply cutting one wire to the LED connected to the transistor, which LED is quite superfluous IMO.  However again, I understand that Lawrence has requested that nothing be changed in this prototype so that it can be sent to others for further testing.

[poynt99]

OK,

I've tested this unit on my other bench using my 1GHz scope and variable power supply.

The device would operate, i.e. "oscillate" only down to a (RMS or MEAN) voltage of right around 0.6VDC. Below about 0.58V, the unit would cease oscillation.

Further to MH's comments on the given INPUT data, I concur. Vpp measurements should not be used in any of these measurements, as the true COP will be skewed significantly as a result.

For instance, While I had about 600mV as the supply voltage, and the scope was displaying the supply voltage (using either RMS or MEAN), changing the measurement to a "Vpp" setting changed the display to about "200mVpp". It would appear that when the scope is told to measure P-P, it uses an "AC" coupling, and discards the DC voltage component in the process (and this makes perfect sense). So this is only now measuring "noise" on the DC supply and does not represent an accurate value to use for any calculations.

I have to conclude Lawrence, that I can not verify your results or claims in this case, using this particular unit. It appears that some incorrect assumptions may have been made also in terms of using Vpp and RMS in the calculation process, and as we've been through the debate about correctly processing the p(t) wave form, similar rules apply to the use of Vpp and RMS where it is not warranted.

It is best advised, that the individual voltage and current "measurements" displayed by the scope, be ignored or not used. The focus must remain on only taking the MEAN of the p(t) wave form, for that is the only computation that provides a true indication of power.

.99

[poynt99]

Thank you professor for verifying my data using the good old fashioned method you described. 

I will be building one or two more of these devices, and I shall leave this assembled unit as is.

I also hope to have a little surprise that some may find quite interesting, exciting, and even sobering.

Regards,
.99

[PhysicsProf]

  Thank you, .99, for your measurements and your willingness to build/test further devices along this line.

  Thank you, Lawrence, for your willingness to send prototypes to people for testing, and your willingness to debate regarding the proper methods for measurements.  This shows your sincere desire to learn the truth about what is happening in these circuits.

[ltseung888]

OK,

I've tested this unit on my other bench using my 1GHz scope and variable power supply.

The device would operate, i.e. "oscillate" only down to a (RMS or MEAN) voltage of right around 0.6VDC. Below about 0.58V, the unit would cease oscillation.

Further to MH's comments on the given INPUT data, I concur. Vpp measurements should not be used in any of these measurements, as the true COP will be skewed significantly as a result.

For instance, While I had about 600mV as the supply voltage, and the scope was displaying the supply voltage (using either RMS or MEAN), changing the measurement to a "Vpp" setting changed the display to about "200mVpp". It would appear that when the scope is told to measure P-P, it uses an "AC" coupling, and discards the DC voltage component in the process (and this makes perfect sense). So this is only now measuring "noise" on the DC supply and does not represent an accurate value to use for any calculations.

I have to conclude Lawrence, that I can not verify your results or claims in this case, using this particular unit. It appears that some incorrect assumptions may have been made also in terms of using Vpp and RMS in the calculation process, and as we've been through the debate about correctly processing the p(t) wave form, similar rules apply to the use of Vpp and RMS where it is not warranted.

It is best advised, that the individual voltage and current "measurements" displayed by the scope, be ignored or not used. The focus must remain on only taking the MEAN of the p(t) wave form, for that is the only computation that provides a true indication of power.

.99

Dear Poynt99,

Please display the waveform and the mean value when the DC voltage is dropped to 600mV.  The final COP should change significantly.

Let us focus on the waveform first.  That reveals most detail.  Please keep the voltage and current waveforms.  Their shape and values are significant in our analysis.

Thank you in advance.

@PhysicsProf, thank you for your support and great work.  At this moment in time, I would like to "hit on" a pseudo resonance condition where the apparent COP is greater than 1.  There may be disputes.  But if we have the waveforms, we have something more solid to discuss.  Poynt99 now has the same components as you.  It will be interesting to see the results from your two different builds.  Beijing also has the same components so we have more checks.  I still have my best prototype with me and that prototype has been double, triple checked.

Hitting pseudo resonance is still an art.

May God Guide us all.

[MileHigh]

Lawrence,

You can ignore my questions if you choose to, but note for certain that the analysis of any circuit has to be based on understanding the concepts and making proper measurements and understanding and interpreting the evidence.

Hitting pseudo resonance is still an art.

There is no "pseudo resonance" going on with a Joule Thief and there is no "art."  The circuit operates based on well understood principles which have already been discussed.  You can expect that the analysis that you see done by Poynt and hopefully others will elaborate on what we have seen so far.

This can be a beneficial learning experience for all involved, and that should include you.  I am not getting a sense that you want to learn and understand.  Please open your mind and try to absorb the information.  Sometimes it may not be what you were expecting or hoping for, but you should still use the experience as an opportunity to learn and grow.

MileHigh

[poynt99]

Lawrence,

The wave forms are very much the same as when the supply is around 1VDC. There is no significant amount of power going back to the source.

I did not propose we not show the voltage and current wave forms. What I suggest is that we not use the "Measure" feature to display what the computed RMS or Vpp values are, because they are of no real value and they can lead to erroneous assumptions.

The only computation that needs to be displayed, is the MEAN of the p(t) wave form.

.99

[poynt99]

Here are the scope shots of the low voltage run. The "in_mean.bmp" file is black and white because the floppy would not hold all 4 bmp files. I had to use a smaller file format (.tif) for this pic.

in_mean.bmp indicates an average INPUT power of 11.2mW. Note that the supply voltage is 576mV.

out_mean.bmp indicates an average OUTPUT power of 32.4mW/10 = 3.24mW.

.99

[ltseung888]

Lawrence,

You can ignore my questions if you choose to, but note for certain that the analysis of any circuit has to be based on understanding the concepts and making proper measurements and understanding and interpreting the evidence.

There is no "pseudo resonance" going on with a Joule Thief and there is no "art."  The circuit operates based on well understood principles which have already been discussed.  You can expect that the analysis that you see done by Poynt and hopefully others will elaborate on what we have seen so far.

This can be a beneficial learning experience for all involved, and that should include you.  I am not getting a sense that you want to learn and understand.  Please open your mind and try to absorb the information.  Sometimes it may not be what you were expecting or hoping for, but you should still use the experience as an opportunity to learn and grow.

MileHigh

Dear MH,

We are looking for pseudo resonance behavior.  It is like the two tuning fork experiment.  If the two tuning forks are 5% different, no sympathetic vibration will occur.  When your experiments are not tuned to correct resonance, you may conclude that sympathetic vibration is false.

I know that you said that you did not have oscilloscopes.  There is no possibility that you could have ever observed the pseudo resonance effect.

Please give time for Poynt99, the PhysicsProf or the Physics Society to reproduce the same result as those at Hong Kong University and the many teams in Hong Kong and China.

The pseudo resonance behavior normally shows itself with a significant dip in the input power waveform to the negative area.  The Voltage waveform will not be smooth.  There will be significant pulsing.  The Current waveform will show large negative values.  So far, the waveforms from Poynt99 have not shown these signs yet.

Another hint is the very large increase in Vpp in the output (10 times or more).  Poynt99 will not display those values because he believes that it will mislead the researchers.  My opinion is different.  When it is hunting for pseudo resonance, every small hint will help.

Wait for Poynt99 to build his own LTJT similar to the PhysicsProf on a breadboard.  He can then observe the change in waveform (similar to the above described) with small changes in hole positioning, wire length etc.  TK already showed a sample.  The moment that Poynt99 could show vast differences in waveforms himself; the analysis will take a completely different turn.

When I have seen dozens of working FLEETs and have one good one with me, I do not care about blind guesses???  I ran two successful workshops at Hong Kong University using my tested equipment.  The trouble with resonance is that a small change will shift the condition.  I shall just wait for Poynt99 or others to hit a pseudo resonance condition.  (My first hit with the Tong Wheel took a few weeks in 2009.)

Once I get back to Hong Kong in a few months, I shall be able to lay hands on the tested and trusted equipment again.  The waveforms do not lie.  The 108 LED Christmas Tree does not lie.  (I am sure the Joule Ringer does not lie also.)

[MileHigh]

Lawrence,

What you are calling "pseudo resonance" is almost certainly something else.  Sometimes circuits can exhibit what's called "metastability" where you typically see unexpected high frequency oscillations.  Sometimes the cause can be a cold solder joint which makes an intermittent connection.  Note there were two cold solder joints on the sample that you sent to Poynt.  Even if you did indeed observe an effect related to metastability, there still won't be any over unity to be found in that situation.

If and when more data comes in, and if it is quality data, you will see that there is no over unity associated with the Joule Thief circuit.

Also there is a very simple explanation for how a Joule Thief can light up 108 LEDs.  Inductors have an innate capability to do this.  One more time this is not an over unity phenomenon.  Certainly the Joule Ringer does not lie.  The Joule Ringer always runs out of energy and has to be recharged.

This is all about understanding and the pursuit of the truth.  Sometimes the truth is not what we were hoping for but it doesn't matter, it's still the truth.

MileHigh

[poynt99]

The pseudo resonance behavior normally shows itself with a significant dip in the input power waveform to the negative area.  The Voltage waveform will not be smooth.  There will be significant pulsing.  The Current waveform will show large negative values.  So far, the waveforms from Poynt99 have not shown these signs yet.

Lawrence,

Do you perhaps mean something like this?

.99

[ltseung888]

Poynt99,

I was hoping to see waveforms closer to the one below which I posted a few times.

The characteristics are:

1.   The Input Voltage  Waveform showed sharp spikes or pulses.
2.   There is a significant negative area in the Current Waveform.
3.   The Vpp Voltage Output is more than 10 times the Vpp Input.
4.   The Area covered by the Output Power Waveform is much larger than that of the Input Area.

When I have such prototypes, I can use them to light up > 100 LEDs; and produce long lasting devices; recharge the Input batteries etc.

When you have built the new LT-JT from components on the breadboard, let me know.  I shall walk you through the various possible tuning steps.  You can skype me.  My skype name is Lawrence_Tseung.  You can also try some of the suggestions from PhysicsProf first – such as removing JT LED; changing the 100-10 ohm combination to 1 ohm etc.  Please leave the existing prototype as is.  Make your changes on the new prototype.  Thank you.

@PhysicsProf, do you mind sharing your private email to me related to the prototype that has reasonable Index but may suddenly fall below 1.  On restart, rises to reasonable value again? 

[MileHigh]

Lawrence,

Do you perhaps mean something like this?

.99

That's a fascinating one Poynt.  It looks like when you lower the supply voltage below a certain threshold you get a different oscillation mode at about 330 KHz, which is very fast.  I can see that when the transistor switches off, that some of the energy stored in the core gets pushed back into the supply rail.  Did you increase the output impedance of the voltage source also to simulate a nearly dead battery?  That's one part of explaining the spikes.

I think  can see how it is shutting itself off too soon to run at 330 KHz.  As the current increases the 1-ohm resistor raises the potential threshold for the base input to conduct and that chokes off the base current, causing a premature shut-off.

What I can't figure out is how the EMF or current related to the discharge goes back to the voltage source.  If L2 pulses high on the "dot" then L1 pulses high on the "dot" also and that's connected to the voltage source.  However, there is no reference between ground and the "dot" of L1 to give L1 the ability to kick-back to the supply.  The discharge glitch is about 200 nanoseconds long.  The only path that I can see would be if the base-emitter junction briefly conducted backwards, and that would let L1 kick back some energy into the supply.  Perhaps that's possible because the diffusion layer of the diode has a population of electrons and holes in play and there is a short window of opportunity for reverse conduction?  (It's been a long time since I delved into the semiconductor physics for an NPN transistor)

I am really going out on a limb here.  It takes some serious analysis work on the bench to figure out the cause of a glitch.

MileHigh