Paul discussed some points regarding the low voltage joule thief.
(I’ll have to get his permission to put the paragraph here.)
First I have to admit that I’m not an expert when it comes to electromagnetics and toroid cores and transformers. The reference data books give standard information such as the transformer’s voltage step-up is equal to the ratio of the turns. Great, but the stuff about B-H curves and the like is a bit out of my league. I’ve read about this but I don’t claim to understand it well enough to make accurate judgments regarding cores, etc. That said…
I tell people that the Joule Thief uses a coil, but it doesn’t transform. The output of the JT is taken from the same winding as the input, and the second feedback winding is there only to keep the circuit oscillating and is not needed for the actual DC to DC conversion.
However in the case of the low voltage JT, the output is not taken from the same winding, it is taken from the same feedback winding that’s connected to the gate of the JFET. So the transformer terminology may apply in this case.
The JFET is not like a transistor which requires input current to get output current. The JFET requires a change in input voltage to get a change in output voltage, like a vacuum tube or valve. Also, with zero voltage on the gate, the JFET is already conducting current. To get the JFET to switch off requires a negative voltage on the gate to turn it off. The JFET operates differently than a regular transistor, which I will call a BJT, short for bipolar junction transistor.
Back to the coil. In order to get enough voltage to switch the gate with a supply of a hundred or less millivolts, the ratio of the primary and feedback windings has to be very high. I used 2 turns on the primary and 400 turns on the feedback winding in my low V JT. That brought the overall gain up to where the Low V JT would oscillate at well under a hundred millivolts.
But the others who were experimenting used a 2SK170, which could operate below 30 millivolts. The objective was to get the LED to light with the heat from a human body. But if you want the LED to light up bright enough to illuminate something, it will take several tens of milliwatts of power, and at 30 millivolts, that’s much more than 1 amp of supply current. The 2SK170 is only capable of a few milliamps, so it will not come even close to making the LED bright enough to illuminate anything.
Paul said that one could not get much work done from a 40 millivolt thermocouple, but the thermocouple is generating enough power (many amps?) to energize a solenoid and hold the armature in against the spring pressure of a few ounces. And it does this as long as the pilot light is lit in the gas heater or water heater. The problem is just getting that very low voltage, very high current stepped up to enough voltage to be useful in lighting an LED or running a microcontroller, etc. And that’s what we are trying to do.
I’ve used a MOSFET to step up 0.7 volt at more than 1 amp to enough power to drive a high power LED. So I figure that it can be done at an even lower voltage.
Remember one important point: In a JT, increasing the number of amps at the sacrifice of inductance is better because the energy stored is equal to the current squared. Increasing the number of turns stores more energy, but the decreasing current is also squared, so you lose more.