Joule Thief 101

[sm0ky2]

if you look closely to the image on the right, you notice the 6th harmonic representing itself as a positive feedback spike. this is not part of the original input signal, but part of the output of the tank.

[MileHigh]

Smoky2:

It sounds to me like you have a moderate case of "resonance fetish" for lack of a better term.  The idea that resonance is somehow special and suggesting that students of electronics are "indoctrinated to NOT observe (it)" is all part and parcel of the fetish.

For starters, I sometimes agonize about the concept of an inductor in self-resonance.  In the real world of electronics, that just means that the inductor is crapping out and failing to function as an inductor at the self-resonant frequency.  It's something that an electronics designer wants to avoid, not "take advantage of."  I also agonize about my comment about the inductor acting like a series LC circuit at self-resonance and having minimum AC impedance.  My first intuitive sense is that the model for an inductor is like a parallel LC circuit as shown in the attached diagram.  That means it has maximum impedance at self-resonance.

This is all backed up in this link:  http://www.cliftonlaboratories.com/self-resonant_frequency_of_inductors.htm

So I am unsure about this discussion and the answer is ultimately to be found in an actual circuit under test on the bench.  I will just repeat that it is essentially impossible to have a hypothetical discussion like we are having with no circuit, no explanation for how it operates, no schematic, and no timing diagrams.

Now, going back to the basic Joule Thief model, you seem to be implying that if it was operating in some kind of self-resonant mode of the main inductor that forms the transformer core of the device, then it would outperform a comparable Joule Thief operating in its normal mode at a given operating frequency as a switching device.  I have no data at all about that, but my instincts are telling me that that is highly unlikely.

Note that a Joule Thief when operating normally has nothing to do with resonance, which I assume you would agree.  It is just a switching device operating at a given frequency based on component values.

Going back to a Joule Thief operating in some kind of "self-resonant mode" there are lots of issues to ponder about that.  For starters, the transistor can only draw current through the main coil in one direction, but resonance means that current is supposed to flow in two directions, so that is somewhat of a paradox.   If the transistor is operating in its linear region, then that means within the signal there is a DC current drain from the battery through the main coil through the transistor to ground.  So that can't be good for the efficiency of the JT overall because the coil and the transistor are both acting like dumb resistors and producing heat.  Also, in true resonance, there is no magnetic field collapse that outputs energy into a useful load like the LED.  Instead, the magnetic field collapse goes back into the electric field inside the coil.

The bottom line is this:  I am sure you can mange to hack a JT circuit so that some kind of high frequency oscillation takes place and the LED lights up.  I am not convinced at all that that is related to the self-resonant frequency of the main coil of the JT transformer at all.  I would suspect that any operation at the true self-resonant frequency of the main coil would not really work, the circuit would crap out.  Rather, I suspect that the oscillation is based on some kind of positive feedback between the transistor acting as an amplification device and the JT transformer with some kind of capacitive coupling through a transistor junction being a critical element in the feedback loop.  No matter the case, an investigation into exactly why and how it is oscillating would require some pretty decent electronics smarts and very decent bench skills of which very few on this forum would be in a position to do.  I doubt that I would be able to do it myself unaided but I would be able to follow it and understand it.  It is highly likely that any kind of oscillation mode will light up the JT more efficiently than the JT operating as a simple switching device that energizes an inductor and then discharges it through a LED.

MileHigh

[MileHigh]

Smoky2:

Some more comments.

You make reference to resonant radio circuits as in a crystal radio, and I don't see how that applies here.

Modern day "joule thieves" place switching transistors in digital mode with inefficient diodes, successfully destroying the resonant effect.

Yes, but what is implicit in what I say about the JT in oscillation in my previous posting is that there is no "resonant effect" in the way you are suggesting.  There is no special advantageous "effect" that people are "destroying."  I don't know what you mean by "inefficient diodes."

There is still "some effect", because of the natural SRF of the circuit being the dominant factor between the inductor and the tank. but it is disrupted during each cycle, thus a heavier drain on the source than a resonant LRC would or should represent in ideal operation.

Well like I said, the "self-resonant frequency" of the JT circuit in normal operation has zero to do with resonance and what it really is simply an operating frequency based on component values.  It's like for a 555 timer where you select component values.  There is no resonance taking place in a 555 timer when it operates as an astable multivibrator.  So there is no partial "effect."

It is a simple concept, which Americans are indoctrinated to NOT observe.
they teach us these things are bad in circuits, and every way to get RID of this effect.
simply reverse your training to do the opposite.

Sorry but I think it is worth hammering home the point that there is no indoctrination, nothing "bad," no "effect" and no "negative training."  There is no true "alternative way to look at a Joule Thief circuit that 'they' 'don't want you to know.'"  Rather, if you want to learn and master switching circuits then great.  if you want to learn and master basic oscillator circuits using various feedback loops then great.  When you have a mastery of both types of circuits, then can you set up scenarios where switching circuits with built-in amplification will edge towards self-oscillation and understand the whys and the hows?  You bet you can do this if you climb up the learning and experience curve.

So anyway, I wrote this all up to demystify your prose and bring the discussion back to something more rooted in the mundane reality of life.  There is absolutely nothing special about Joule Thief circuits and there is absolutely no "hidden knowledge to uncover" about Joule Thief circuits.  If a Joule Thief circuit can be coaxed into oscillating, then the real thing to do is understand why if you are so inclined.  However, if you are just looking at the net result - how much perceived LED brightness do I get for a given amount of power input, I seriously doubt that any non-standard Joule Thief oscillation mode will outperform a Joule Thief in normal operation mode.

MileHigh

[TinselKoala]

https://www.youtube.com/watch?v=ekPh9p4YECE

Very nice! I wish I could grow peas that big in my garden!

I wonder how small your circuit would be if you put it on a circuit board, with input and output connectors so you could use 2 or 3 LEDs in series and run them off of assorted dead button cells without soldering anything.

Somewhere I drew up the circuits using two different types of chips from garden lights but I can't find them at the moment, I have so many different JT circuits in various places on this computer. It's getting hard to remember where I put stuff.

[tinman]

Brad:

Nice.  I have about 40 of those led garden light circuit boards lying around here, those chips they use make a decent JT.
Is that button cell 3 volts or 1.5?  Will it run those down to low voltage as well?

Bill

Hi Bill

Yes,the garden light IC's are extremely small. The battery is a 1.5v battery--that is why you need the small inductor. It will run down to a battery voltage of about .6v.

Brad