I’ve read quite a few of the blog posts (here is one) and seen quite a few of the Youtube videos that use a JT for a charger. Basically it’s a regular JT, with a diode from the collector to the charging battery’s positive terminal, and the other negative terminal connected to the positive of the running battery that is also connected to the coil.
When the coil is wound bifilar, the same voltage appears at both windings because the windings are equal in turns. So if you get a high collector voltage, the voltage will be high at the winding that goes to the base bias resistor and pot, and then to the base. But a high voltage is not needed at the base, it is turned on by current, and needs only a fraction of a volt to a few volts.
The usual circuit uses a resistor and potentiometer between this winding and the base to adjust the current. But the resistor and pot are dissipating power that could otherwise be used in charging the battery. Also, the base to emitter junction has a maximum reverse voltage of 5 volts, and it is bad for the transistor if the voltage goes more negative than this.
To ‘harvest’ this wasted power, the feedback winding (that goes to the pot. resistor and base) should be more closely matched to the current needs of the base. This is easy to do, by reducing the number of turns so that the winding puts out lower voltage and the current stays the same, which means the total power used by the resistor, pot and base is reduced. Less power is then wasted. The pot will have to be set at a lower point, but the wasted power will then be available for the charging.
The feedback winding’s lower voltage means that the base will not be exposed to high reverse voltage, and the transistor will have less risk of being damaged. The next question is how many turns do we have to remove to get the voltage down to what it should be.
If a 12V battery is being charged, the actual voltage will be 13.8V, but the charger will have to put out more like 15 or 16 volts. So the feedback winding should have no more than 5 volts. This is about 1/3 of the output, so the winding should have 1/3 of the number of turns. It would not hurt to have 1/4 the number of turns because you can adjust the pot to compensate. Oops, I forgot to include the running battery’s voltage – the two are in series, so the collector voltage is more like 28 to 32 volts. So the ratio should be more like 6 to 8 turns for the primary winding for every one turn of the feedback winding.
I have often seen this circuit built with a 1N4004 diode for the rectifier. The 1N4004 series of rectifiers were made to be used at power line frequencies; when they are used in a high frequency circuit such as a Joule Thief, they cannot recover quickly enough and power is wasted in heating up the diode. It is better to use a fast recovery diode such as the UF4007, or a 1 amp Schottky diode that can handle at least 100 volts.
If for some reason the circuit is running and the battery being charged is disconnected, the output voltage can go much higher. That’s why the experimenters put a neon lamp across the transistor, to protect it from high voltage. But the base will still get a voltage more negative than minus 5V. So a good way to protect the base is to put a LED between the base and emitter, with the cathode or flat spot toward the base. If the voltage goes more than 3V negative, the LED will light up and shunt the current away from the base. Blue is a cool color to use. 😉
Back to experimenting…