My watsonseblog dated 2011 Apr 12
2012 Apr 12 JT Toroid Turns, Inductance
I have some answers to the questions I read here.
What are the inductance ranges that work in the Joule Thief? What are the limits to the choice of turns?
What is the maximum number of turns and the minimum number of turns needed to get the Joule Thief to function?
Let’s start with the frequency, since that sets limits on the inductance and turns. For the lower limit, the frequency can be as low as a hundred or so Hertz, which is where flicker starts to become noticeable to the human eye. For the upper limit, anything below about 450 kHz should be okay as long as it doesn’t interfere with anything. The frequency could go higher, but the efficiency starts to drop and the RF interference could cause stations in the AM band to have interference.
To reach the upper frequency limit, the inductance could be as low as 10 microhenrys. This could be just one or a few turns for a high permeability (high mu or high μ) core, or a few tens of turns for a low permeability core or air core. In other words, the inductance of the coil depends on the number of turns, which is dependent on the type of core material, the size of the core, all of which determine the permeability.
You might think that I should also include the size of the wire. But I’ve wound many coils with a bifilar (two wires in parallel) winding using different size wires such as 24 AWG and 30 AWG. When I measure the windings, their inductances are identical, even though the wires are very different sizes.
It’s possible to get a JT to light the LED with an air core coil of just a few turns, but the frequency might be several megahertz, and the stray capacitances of the circuit may cause a loss of efficiency. A common transistor such as the 2N4401 or PN2222 can oscillate as high as a hundred or more Megahertz. But there really is no point in going this high.
To reach the lower limit, it would take a hundred millihenrys or more, but with this high an inductance, the coil would become very large. The hundreds of turns needed would be very high DC resistance if the wire is thin, so heavier wire would be needed, and the coil would have to be physically larger. The coil might be as big as a golf ball (1-3/8″ or 35mm) or even much bigger depending on the core used. This would be too big, considering the circuit itself would fit on a small coin. Remember that a JT needs about a quarter amp peak current at 1.5 volts, and that is the same as a 6 ohm resistor on a AA cell. The coil must have a DC resistance that is much lower than this for a reasonably bright LED.
I hope this answers the questions. The Joule Thief is tolerant of a very wide range of coil inductance and turns, as long as the DC resistance or losses do not cause the LED to dim. And the overall size of the circuit will put a limit on the coil size, number of turns and inductance. This holds true for silicon small signal transistors. Some transistors such as silicon power transistors and germanium transistors have much lower maximum frequency, which could be as low as a few hundred kilohertz or even less. This limit could greatly reduce the maximum JT frequency and hence the minimum coil inductance. I’ve found that germanium power transistors can’t go above a few kilohertz. The coil has to be high inductance, several tens of millihenrys, and physically large with heavy wire.
Back to experimenting…