I was looking at two schematics online, one a NPN Joule Thief, the other a PNP Joule Thief. The gears started turning and I thought that I could combine them into one Joule Thief. At first I thought dual Joule Thief, but combined it, too, into Doule Thief. Why not…
One may ask why do this. Often JTs are connected to sources of AC such as a stepper motor. This is usually done by rectifying the AC to DC. But at low voltages the rectifiers waste a substantial part of the power. Making the circuit accept both polarities eliminates the rectifiers.
Several Circuits
There are several ways to get the job done. I’d say the most obvious way is to connect two separate JTs in parallel, one NPN, the other PNP. Each lights up on each half of the AC cycle.
But the coil and resistor are not polarized so it would be simpler if they could be used for both positive and negative parts of the AC cycle.
I could add a second JT to the same coil and add another resistor, to light the LED on the other half of the AC cycle.
But thinking about it, a second JT doesn’t have to be PNP; it could be NPN with the polarity reversed.
Essentially the transistors and LEDs are paralled (see note), so they can handle either polarity. But then both transistors would have to be the same – NPN or PNP. Can I do this? Suddenly I’m having doubts.
In order for the PNP to work, the base bias needs to be changed from one polarity to the opposite. I would have to use diodes to steer the bias to the base. I’m going to have to draw a schematic or two to get something that looks workable.
Devil In The Details
What I see is the circuit will obviously need a DPDT switch to switch the polarity of the battery. If I add a third pole to the switch, I can then use it to switch the polarity of the base bias (see schematic). Switching the base lead causes the wrong polarity transistor to have just the emitter and collector connected, which is the same as putting two diodes in series with their anodes or cathodes connected to each other. This prevents current flow in either direction, essentially an open circuit.
Instead, if I leave all the leads of both transistors connected in parallel, the emitter to base junction would conduct when reverse biased. Or would it? I’m going to have to experiment with this to see what happens.
Luckily I bought several of the 3PDT switches from Electronic Goldmine recently. I’m all set, with all the hardware. Need some bench time.
Note: Paralleling two LEDs ‘antiparallel’ is connecting them anode to cathode. This also applies to transistors and other polarized devices.
Do you have any schematics to illustrate the design?
I never did experiment with the ideas I brought forth. But as I said, just build two JTs and connect them anti-parallel. The positive of the first one connects to the negative of the second one and the negative of the first one connects to the positive of the second one. then connect these two to the 1 volt AC source’s leads, it doesn’t matter which way. One LED will flash on the positive half cycle, and the other LED will flash on the negative half cycle.