Is joule thief circuit gets overunity?

[poynt99]

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

[TinselKoala]

@PW:
The sequence of events you have described is exactly why I have asked Lawrence to provide the input and output Voltages on one scope, and the input and output Currents on the other scope. This will allow "us" to observe the time sequence, the relationship between the input and output pulsations. We cannot see this in Lawrence's present data, with input and output data on separate instruments. A 4-ch scope will of course show these relationships in real time, but Lawrence only has the two 2-ch scopes to work with.

Another bit of data that would be quite instructive to "us" would be some actual measure of LED light output. This doesn't have to be quantitative, just synchronous. This would allow "us" to observe the relationship between the actual _illumination_ of the LED and the pulses coming from the various parts of the circuit. Again, there are not enough channels available to see this in real time, but an experimental session could be devoted to showing, say, the input or output voltage or current trace on one channel, and the trace from a photovoltaic cell taped to the LED on the second channel. This will answer the question of "when" the LED is actually ON, in relationship to the input and output signal pulses.

I have found in the past in pulsed circuits that people sometimes believe that a component is "on", when in fact it is "off".


@PW, .99:
In Lawrence's recent scopeshots of Board 80, the one that I have edited to show the _true_ baseline for the CH2 current measurement.... note that the scope is reporting + 40.00 mV RMS for that signal...... which barely ever actually goes positive, never even reaching 40 mV positive that I can see, and extending deeply negative during its spikes.  RMS, of course, is mathematically always a positive value..... so how are we to interpret a _positive_ RMS value for a signal that is actually almost always, if not strictly always, negative? The RMS value gives the wrong direction for the current being measured in this channel, doesn't it?
Lawrence has explained that the RMS boxes are a "legacy" from his early work. I hope it's clear now that these values are not relevant and may even be confusing the issue, due to things like the positive RMS value for a clearly negative-average signal. They should not be displayed at all. If the scope has a simple "average" parameter available, that might provide some rough and useful information. An artificially positive RMS value does not.

@Lawrence:
Once again.... I ask you to provide Input and Output Voltages on ONE SCOPE, and the Input and Output Currents on the OTHER SCOPE, so that we can see the TIMING relationship between the input and output pulses. I am NOT suggesting that you always do it this way; what you are doing normally is proper for your power measurements. But we need to see at least one set of screens taken with input and output voltages on one scope, and input and output currents on the other, for the timing relationships.

This TIMING relationship is what PW has described in detail above, and what I have greatly simplified when I said that the dips in the battery voltage trace are due to the LED turning on.
As PW has explained there are a couple of intermediate steps in that process, and the light from the LED is probably not happening at the exact same time that the dips in the battery voltage are happening. Nevertheless, the dips represent power being drawn from the battery, and this power eventually is mostly dissipated in the light from the LED. We really would like to see some traces of INPUT and OUTPUT on the SAME scope screen at the SAME time, so that these relationships can be made clear.

If you can arrange it, a photovoltaic cell (solar cell) simply taped in a stable manner to your board's LED will provide a voltage signal when the LED is actually ON, and can be simply hooked to a scope channel probe directly. This measurement should be done with the FLEET board's _Input_ Battery Voltage on one channel and the Photovoltaic output on the second channel. You can get a photovoltaic (solar) cell very cheaply by taking apart one of those self-charging garden lights that cost a couple of dollars -- and you'll also get a couple of white LEDs. A bargain however you look at it. Or, a simple photoRESISTOR in series with a power source like a battery can be used to provide a signal that is on when the LED is on. There is also a photoresistor inside many types of garden lights, but some simply use the PV cell itself to detect day/night.

[TinselKoala]

An even better, temporary, arrangement that might work to show the time relationship between the LED output and the circuit's pulsations, would be to use an optoisolator chip's input stage as a substitute for the board's LED. Then one can simply use the phototransistor output from the optoisolator to monitor the actual "LED" turning on and off. This method doesn't have the potential to actually measure the board's LED light output, though, but it will give the most accurate timing data, I think.

The H11Dx series of optoisolators is cheap and robust. I think I have an H11D4 that's not in use, I can try this on my 2n2222 JT test bed to see if it will work. The phototransistor output stage can be powered by a separate battery or power supply for the few moments the testing will take.

[TinselKoala]

Some oscilloscopes, like my venerable HP180a with the 1821a horizontal timebase/delay generator, have a trigger _output_ that can be used to synchronize other instruments.
It would be nice if the Attens had a trigger output. Then you could use the trigger out from one to trigger the other one, and achieve "manual" synchrony that way and reassemble a "4-channel" set of traces in the spreadsheet.

[TinselKoala]

Sorry to make so many posts in a row, but I'm having my coffee and thinking hard about this, also warming up my scopes.

By the way, Lawrence, even DSOs should be "warmed up" for a few minutes and allowed to stabilize before taking quantitative measurements.  Believe it or not, we have actually encountered.... persons.... who will turn the scope on, and as soon as it passes its internal self-test they make a measurement and then immediately turn the scope off, to .... save power or something. Then to make the next measurement, they turn the scope back on...... '

But the point of this post is to request that Lawrence select ONE "good" performing JT / FLEET board for his own testing and reporting purposes here, so that we can always refer to that board's performance traces in our discussions. This jumping around from board to board, with their individual differences, is not helping to clarify matters in the question of OU performance very much.
But it might also be helpful to have the WORST performing of all the boards to use as a control or comparison, as well.