Joule Thief

[gyulasun]

Hi Conrad, I have sent you a Personal Message.  Gyula

[conradelektro]

Simple high frequency (115 KHz to 125 KHz) air core Joule Thief:

Resonanceman told me a few weeks ago in his Reply #13352 that various air cores (even some wire coiled in a heap or twisted wire) will work.

So, I built a very simple air core coil (just L1 + L2, no secondary L3) with some left over wire from a rather old telephone installation.

It swings with about   115 KHz to 125 KHz  (as I like it) and can drive 5 LEDs very brightly with about 1 to 1.3 Volt (100 to 150 mA power consumption because the LEDs really shine).

This is geared to a single AAA or AA rechargeable battery and could be turned into a simple flash light. You can use 7 LEDs with 1.5 Volt, or 3 LEDs with 0.6 Volt.

The point of this post is again the high frequency of an air core Joule Thief (in comparison to a toroid Joule Thief).

I also like the idea that one can take a card board tube instaed of hunting for a toroid. Winding a spool is also easier (either by hand or with a crude machine turned by hand) than fiddling wire through a toroid.

My next attempt will be an air core with 250 mm diameter and 250 turns of 0.2 mm enamelled wire. According to my secret calculations  (or rather unfounded estimates) it should light CFLs. The big question and my motivation is its frequency?

Some one in this forum told me about this calculator (sorry I forgot who it was):
http://www.dl5swb.de/html/mini_ring_core_calculator.htm (very good, I use it a lot for my air core coils)

Greetings, Conrad

[Pirate88179]

Conrad:

Nice job on the air core.  The only potential problem I see with this approach vs the ferrite toroids is the efficiency will not be as good.  The ferrite demagnetizes very quickly and causes very good spikes as the field collapses.  These spikes are what we are using to drive stuff and, I could be wrong, but I don't think you will be able to run as many leds for as long a time without a ferrite core.  Be interesting to see though.

My JT are silent after adding a strong neo to raise the freq. I only needed to do this to 2 of them as most of them were already way above hearing frequencies.  Even my cat cannot hear them.  Maybe you can try different windings and tuning to get them to be quiet?

Bill

[conradelektro]

@ Pirate88179 and all experimenters who like to think about frequency in a Joule Thief


I have not studdied electronics, but I try to learn as much as I can about it. In the last months I found out (from this forum, from articles on the internet and from books):

- A Joule Thief circuit is a resonant circuit with a coil and a capacitance: the coil is L1 and the capacitance is given by the base resistor together with the base/collector of the transistor.

- The frequency limit of the Joule Thief circuit depends basically on the inductance (Henry) of one loop of wire over the core.  Higher inductance implies lower frequency.

- For practical reasons one needs at least about 5 turns of L1 on a toroid and 15 turns of L1 on an air core.

- Because a good toroid induces a much higher inductance in 5 turns of wire than an air core induces in 15 turns of wire, one can reach much higher frequencies with an air core (up to 200 times higher frequencies).


Efficiency:

- A good toroid (with high permeability) needs few turns of wire to produce a high Voltage on L3. In this sense it is "efficient". (A very small space is occupied by the coil, very little wire is necessary.)

- An air core needs many turns of wire to produce a high voltage on L3, or one uses a bigger core (which is possible up to a certain practical limit). In this sense an air core is "inefficient". (A big space is occupied by the coil, lots of wire is necessary.)

- But the "energy" transferred over equivalent coils is roughly (some practical limitations apply) the same, no matter what core one uses.


One more thing to understand:

- A toroid might produce 6.000 spikes with a 1000 Volt per second on L3; this will need a certain energy on the input.

- An air coil might produce 60.000 spikes with 1000 Volt per second on L3; this will need ten times more energy on the input.


No we can ask the right question about efficiency:

Will 300 LEDs driven with 60.000 spikes of 1000 Volt per second shine ten times more brightly than 300 LEDs driven with 6.000 spikes of 1000 Volt per second?

For all practical reasons (within sensible margins) they will shine ten times more brightly and will use ten times more energy.


Remarks:

- If one uses a toroid with the same low permeability (Al value) as an air core, one will need the same high number of wire-turns as with an air core and the same size of core.

- The Tesla coil shows that an air core can also reach much higher Voltages than a toroid. This is because of its size. One could also do it with a very big and very low permeability toroid. But this would be even "less practical" (but not "less efficient").

Greetings, Conrad


P.S.: I do not want to be the smart Alec, I just cite conventional theory as found in books.

[jadaro2600]

@ jadaro2600

It would be very interesting to know at what frequency the Cold Cathode florescent tubes in your 17 flatbed scanners are driven?

Since I got interested in the frequency of Joule Thief circuits I found out that the conventional "CFL, FL, neon an cold cathode lamp drivers" (I looked at), are driven with a frequency in the audio range (which I find annoying).

Greetings, Conrad

I'll turn one on and see - to note however, none of the driver boards are making any noise what so ever.  Most scanner's stepper motors make the sci-fi noises.

There are two transistors on the circuit I've measured, across the E-C junctions, they're running at ~56.6kHz.  E-B is at ~58.2kHz

I haven't measured the voltage at the tube, but it's connected like a trigger coil would .Voltage there causes the meter to malfunction, so I would need to build a series capacitor bank to measure it.   ...in fact, this model is connected ... coil-lead 1 - tube lead 1, tube lead 2 - capacitor lead 1, capacitor lead 2 to coil lead 2.

coil-tube-capacitor-coil



- - - - - -

I think it would be nice to tune this a bit, or rebuild another one using some of the circuits here on the forum.  The NPN PNP configuration uses very little ma ..

These particular 12 inch CCFL's are trigger-on 1000v, staying on at around 600 or so volts ( as was found through researching the available data on the replacement lamps.  It seems they're available in a variety of lengths and cut-on voltages.

This may be a very nice source of light for a joule thief to power.