On occasion I see others refer to the Joule Thief and flashlight (torch) page of talkingelectronics.com. He has many comments, some good, some not so good, and some that miss the mark. Someone recently mentioned the Circuit B, which is shown in one of the pictures. He talked about increasing the efficiency a lot by adding a capacitor. I found it amusing that he showed the wire on the windings to be 0.095 mm diameter. A millimeter is slightly less than 40 thousandths of an inch (forgive me for not using metric), and a tenth of a millimeter is less than 4 thousandths of an inch, so his wire is even slightly less than that. I’m concerned about using 30 AWG (0.25 mm) wire, which is ten thousandths of an inch, and he’s using wire that’s less than half that diameter! Then he winds 40 and 60 turns on the core, and the resulting DC resistance must be several ohms. The coil for a Joule Thief should have as low a resistance as possible, to minimize the IR losses, and help keep efficiency high. High resistance is bad for a JT. Use heavier wire to keep the resistance well below 1 ohm, and the losses will be much lower.
He gave two examples of “Poor Circuit Designs”. He said that the first one was a waste of a transistor. I don’t think the transistor was necessary, it has the base shorted to the collector, which turns it into a diode. The designer could have instead raised the 220 ohm resistor to 1000 ohms and it probably would have worked.
The second “poor circuit” is taken from this joulethief.com website, where it is being sold as a kit. This is a common circuit seen often in place of a conventional Joule Thief. The circuit works, but I would have made some changes to parts values, the most important being the 470uH inductor. Small inductors use fine wire and a 470 uH most likely has several ohms DC resistance, which is wasting power. I would have used a 100 uH inductor, which has heavier wire and lower DC resistance, and lower losses. He seemed to criticize the second transistor, which was most likely used to allow the designer to use a simple choke for the coil, with only a single winding, instead of a toroid core with two windings. The second transistor inverts the feedback from the coil, to keep the circuit oscillating. Nowadays a transistor is a few cents and is probably cheaper than the cost of adding the second winding to a coil.
He implied that the second transistor is wasting current and therefore inefficient. The 1k resistor he mentioned has only 0.9V across it when the leftmost transistor is turned off, and only 1.5V when it’s turned on. So somewhere between 0.9 and 1.5 milliwatts of power is being dissipated in this resistor, which is about 1 percent of the total power a typical Joule Thief uses from a 1.5v cell. That’s very minor. I would guess that the conventional 1 transistor JT wastes about the same amount of power in the 1k resistor it uses (but I have never investigated to find out what that power might be). He said that the JT with the feedback winding was more efficient, but with the very fine wire he used, I find that difficult to believe. In my measurements of conventional JTs, I’ve found that they are typically about 40 to 70 percent efficient, with a typical one around 50%. My Supercharged Joule Thief is much more efficient.
Pulsed LED is More Efficient He also makes the statement (I quote) “This proves the fact that LEDs driven with a pulse, are more efficient than being driven by a DC supply.“ That doesn’t agree with what I’ve found. In my experiment, I switched a LED between connecting directly to the coil/collector, and to a 1N5817 Schottky diode and filter capacitor to rectify and feed DC to the LED. The circuit and LED were the same circuit, just the switch was added. I measured more current to the LED when it was being fed DC from the diode/capacitor. Read the update at the end and the comment in my blog here. If you want more info about LED efficacy (efficiency) and more about LEDs (and lighting) in general, visit Don Klipstein’s LED web pages.
Transformer The talkingelectronics author goes on to talk about “the transformer.” He then contradicts himself by stating that the Joule Thief coils is not a transformer, but a flyback coil with a feedback winding, and I agree. But why did he call it a transformer to begin with? It isn’t a transformer, because the circuit will function without the feedback winding, as he showed in the second of the “poor designs”.
There are other confusing statements in this section. For instance, he calls it regeneration, but then later says positive feedback. One might call it regenerative or positive feedback, but regeneration is not the correct term. He then says that it “produces a very high voltage..”, which is untrue. If it were true, the circuit would be damaged! The coil stores energy, that if it did not have the LED to dissipate in, would increase the voltage until some other component breaks down. As long as the LED remains connected, the voltage will never be “very high”, the voltage will only rise to the point where the LED conducts. I think it’s confusing to the reader to make these statements.
“Flashing LED” The author continues with a section on a flashing white LED. It uses a second transistor with the collector unconnected, and the emitter to base junction reverse biased so that it zeners or breaks down, causing the LED to flash. If any circuit deserves the term poor design, it is this one. This circuit will never be found in any well designed project because it depends on a transistor’s function that was never meant to be used (the voltage is exceeding the maximum parameters given in the datasheet), and therefore is not controlled or tested when the transistor is manufactured. Thus the transistor cannot be depended on to flash, and this would be especially true at extremes of temperature. Furthermore, if current flows through the emitter to base junction during reverse breakdown, it will permanently damage the transistor, and the current gain will be permanently reduced. If you use the transistor for this purpose it is a good idea to cut off the collector lead so that it cannot be used as a transistor again.
He says it is difficult to visually compare the brightness of LEDs, and I agree. But the point is that one should not use the naked eye to make a brightness comparison of LEDs. I purchased a luxmeter for a reasonable price, I believe it was $40 U.S., and it allows me to make reasonably accurate measurements of the LED brightness. It allows me to make a comparison of different circuits connected to the same LED. If the experimenter thinks that that is too expensive, there is another way to compare LED brightness. Obtain a CdS photocell, and mount it on one end of a small light tight box, and mount the LED in a hole in the other end. Use a cheap DMM to measure the photocell’s resistance when it’s illuminated by one circuit, and compare that to the same LED on the other circuit.
Talkingelectronics certainly does a lot of talking, but I don’t believe it is substantially adding to the understanding of the readers.