Joule Thief 101

[sm0ky2]

I have a book by John D Lenk   Simplified design of switching power supplies

It has some data charts of some inductors from 10uh to 3.3mh and the SRF of 45mhz to 360khz respectively

Listing under must meet criteria....

Stray capacitance - The inductors self resonant freq must be 5 to 10 times the switching frequency


The book explains SRF briefly...

"All inductors have some distributed capacitance that combines with the inductance to form a resonant circuit. The frequency of this self resonance should be between 5 and 10 times the switching frequency(but not an exact multiple of the switching frequency!). As the inductance value is set by circuit requirements, the SRF is determined by distributed capacitance(a higher capacitance produces a lower SRF).

When SRF is low, the normal linear ramp of the inductor current is preceded by a sudden jump in current when the switching transistor turns on. This results in so called switching losses that lower the regulators overall efficiency. As a result, distributed capacitance should be kept at a minimum so that the SRF will be high and will not seriously affect the inductor current. Distributed capacitance can be lowered when the toroid is wound, either by overlapping the ends of the winding somewhat or by leaving a gap between winding ends(rather than ending the winding at 1 full layer)."

Mags

@ Mags

this is an excellent example.
Along side these spikes in current, is an associated drop in voltage.
We are trained to ignore these relationships when we examine certain phenomena.
But they hold true in every case, regardless of what we do with the electricity. run it through an inductor a capacitor a resistor a transformer a transistor and back to your meters and these factors always remain proportional.
Then we are given workable solutions that dodge resonant frequencies. Why?
because there is a crossover point at the SRF.
In one state (lagging) the spike is along the current domain.
In the other (leading) the spike occurs in voltage.
Now consider a "mostly dead" battery, and the available current from its' depleted output.
And the inverse situation, where current does not spike, but voltage does.
current incurs an associated drop from the source.
This is a function of the SRF of the coil vs the SRF of the ferrite.
If the two were perfectly balanced, there would not be a spike in current, nor a spike in voltage.
But rather, both amplitudes would peak at their maximum value, one slightly out of phase with the other. This cannot generally occur, because of non resonant parameters in the rest of the circuit.

[Magluvin]

Hey Smoky

So what was described in the power supply book only touched on the coils SRF, not the cores SRF.

So if we want to bring these 2 together, can we lower the core SRF or use other materials that have a lower core SRF?  Or is it that we have to make the coil with an SRF of the core, or say an SRF that is a lower multiple of the core freq.

Thanks for all the knowledge on this stuff.  You seem to know it all pretty deep.

Mags

[sm0ky2]

Hey Smoky

So what was described in the power supply book only touched on the coils SRF, not the cores SRF.

So if we want to bring these 2 together, can we lower the core SRF or use other materials that have a lower core SRF?  Or is it that we have to make the coil with an SRF of the core, or say an SRF that is a lower multiple of the core freq.

Thanks for all the knowledge on this stuff.  You seem to know it all pretty deep.

Mags

Lowering the SRF of the core material is not possible. Ferrites that have a lower SRF, like raw iron or magnetite, also have a very high hysteresis. The materials we use in modern inductors is a ceramic embedded with very fine particles, allowing for a more pure and clean alignment of the magnetic domains. Naturally, the smaller the particles used, the closer we get to a pure atomic induction response, thus the SRF of the material approaches the self resonance of the atoms. We aren't quite there yet, but our technology is getting pretty close, and with nanotech we expect to be able to create ferrites with even higher SRF frequencies.
If you can imagine the future of a microwave oven using only a toroid and a simple oscillating circuit.

The problem with building a coil that has an SRF as high as the ferrite core
comes in two forms: one being the very low capacitance value required, we would almost need a superconductor (or at the very least, some gold coils!!)
The other being the switching rate of the transistor. Transistors of this nature can be very expensive and hard to find. Generally a "JT" uses a transistor with a relatively low range of freqs. when compared to the SRF of the core material.

So, to answer your question (which it sounds like you already have on your own)-
The latter of the 3 options, is what we choose in practice. We already have the tools to make this possible, with minimal alterations to the circuit.

[sm0ky2]

in the simplest form, this can be achieved with a variac or high-precision variable resistor,
some short-range trim pots will do the trick. This goes in place of the base resistor in the JT circuit.

With the normal JT set-up, the SRF of the coil will drift over time. I believe this is due to the drop in voltage from the source battery, but I have not invested a serious bit of time into examining this situation.
My brother reports the drift taking place over a 3-4 month time period with some of his simpler JT "nightlights". The light dims and they need to be "re-tuned", they then return to normal operation, until some time when the SRF drifts again to an effective amount.
He has shown this to me on the scopes, so there is no doubt that the drift occurs.
I only speculate that it is from the battery voltage, because we know this to drop over time.

[MileHigh]

Smoky2:

The information in Magluvin's book is pertinent and says it all.  You stay away from the SRF of the coils in a switching power supply because at the SRF the coils crap out and don't function as coils anymore.  You even stay away from having a harmonic of your excitation frequency line up with the SRF of the coils.  The excitation is a pulse train with sharp edges so naturally the signal is very high in harmonics.

A coil at it's SRF is either dead and blocks AC if you model it as a parallel resonant tank or it's dead and offers no resistance to AC if you model it as a series LC tank.  In either case the inductance is nowhere to be found.  Above the SRF it just looks like a capacitor.

Why should a coil at its SRF enhance the performance of a Joule Thief when it is effectively dead and not functioning properly?  You are just playing the resonance fetish game.

Your discussion of reflections and stuff like that only occur at super high frequencies.  You would worry about that when you design a motherboard for a PC with a 4 GHz clock speed and perhaps a 1 GHz memory bus clock, but not for a Joule Thief.

My gut instincts are telling me that your brother is playing with hacked Joule Thieves that are running as oscillators, but they are not running at the SRF of the main coil that forms the JT transformer.  I will repeat to you again in plain English, the main power coil craps out at the SRF and the inductance disappears.  So my feeling is that you are leading yourselves down a garden path.  If you really wanted to be sure you could inject a signal into the coil and look for the SRF.

In broad general terms, the "buzz" about a coil operating at its SRF on the free energy forums is a bunch of BS.  You are effectively turning the coil into a piece of wire or an open circuit.  There is nothing exciting about that.  There is no "secret sauce" related to hacking a JT and turning it into an oscillator and running it at the SRF of the main coil.  There is a very decent chance that the oscillation would in fact die at the SRF because the main coil of the JT becomes inert at the SRF frequency.

MileHigh