I was watching this Youtube video of a Joule Thief that is used as an EMF detector. Actually it is detecting the electrostatic field, not the electromagnetic field. If it had a pickup coil instead of an antenna, then it would be detecting the electromagnetic field. Anyway, it looks like he is trying to trace the wiring in the wall. That’s a very good idea. Except it wouldn’t work if the wiring was inside of metallic tubing or flexible metallic cable.
I thought that if the circuit was changed to turn the LED off when a field was detected, then the Joule Thief could be used for a night light. But then I realized that it would be easier if the circuit used a CdS photocell instead, which would turn it off when the light was on, and also when the ambient light was daylight.
The circuit is called a Joule Thief, but it differs from a true JT. The coil doesn’t have a feedback winding, and actually the whole circuit doesn’t have any feedback – it’s operating ‘open loop.’ The circuit depends on the AC power line field that is present at the antenna to give the circuit something to sense. It has no bias resistor on the input so it acts as a detector. In fact, the whole circuit has no resistors, which means there is nothing to limit the current. If the input gets shorted to the battery negative excessive current will flow and one, the other, or both transistors will most likely get damaged. It would be an improvement if there was a resistor between the antenna and the base, and another resistor elsewhere in the circuit to limit the current.
As the pulses of current from the first transistor pass through the second transistor, they are not enough to light the LED. When the current stops, the back EMF or inductive kick of the coil creates enough voltage to cause the LED to conduct for a brief time. It’s not very bright – there are only fifty or sixty short flashes every second.
What this all leads to is that the circuit would be better if the coil had a feedback winding and acted like a true JT, with the first transistor turning on the base bias to the JT. It could be more sensitive, especially if the two transistors fed a true JT instead of directly feeding the LED.
I drew up a schematic, but I’ve yet to get the soldering iron heated up and build it. All in due time. UPDATE: I built it and I’ve started another blog about it.