Author Options:

Joule Thief Troubles Answered

I have built several Joule Thieves as seen in this video. None of them worked. Is the schematic wrong? I am sure that I used all the right components, and I wound the transformers right... What could be the problem? Could it be the ferrite core? I used one from an old computer power supply. Any help would be appreciated.


I could never get them to work. But I suspect that I always screw the toroid up, like connecting 2 wrong wires together so it doesn't all work right, or solder the wrong part to the transistor, etc.

The bifilar is tricky if you look at the signal with a scope the second winding is 180 out of phase. So the coils need to be reversed similar to the coral castle permanent magnetic holder. Which makes me wonder if the signal is now 360 out of phase and is that what makes the magic happen? The more you learn the less you know.


10 years ago

Hey, I finally got it to work! I guess the problem was the toroid. I used some thinner wire from an old printer cable, and I got on eleven turns. Expect a new Instructable soon!

That's great that you got it to work.....it was weird, but when I heard so many having troubles with that (both here and in the MAKE forums) I tried it, and it worked. Dumb luck I guess....seems like I wasted all my luck on that one project since none of mine from that time on has gone without a hitch first time through ;-)

on a project with a coil or a toroid, the problem IS always the toroid or the coil. Always.

I've "fried" a few transistors already, breaking "power contact" without the obligatory diode in place.....not on a Joule Thief, but other projects.....

i never had troubles with frying circuit parts, but almost 9 times out of 10 with toroids.....

I suppose it depends on the amperage being used.....I did a lot of Automobile modding when I had an old used VW....unprotected relays could reak havok on things, believe me :-) 


10 years ago

Did you try reversing the leads on one of your coils? What exactly did you use for your core and how many turns of wire did you use? The recent video used rather fewer turns of wire than Big Clive's original (?) directions. What about your transistor? Are you sure you used the right pins - there are several common pin-outs.

1. No, I didn't think it mattered... 2. I used a core from an old computer power supply, and I wound the wires 7 times. 3. I used a 2N3904 transistor, and I will try different pinouts. How can I tell what the right pinout is? What other transistors can I use?

> I wound the wires 7 times . Clive's plans call for at least 20 turns.

That looks like about 7 turns to me.


. I've never built one. Just going by what Clive said.

well, i guess go with the 20 if you have thin electromaggnet wire from the small transformers on the Power supply, but if you have the ones like cat5 wires, go for 8-10

Clive said 20 t, but EMSL says that far fewer seem to work. Did you save your first effort? I was quite serious when I said "reverse the leads" - it's almost easier to do trial and error than figure out which way is correct from the start. If you're sure the circuit is wired correctly, you just switch the leads of the primary and try again.

I made the joule thief shown on the site 1up mentioned (from Make), and I put eight turns on it. It works perfectly, I just made a bad solder joint and it fell apart, so it's out of commission for the moment. I was kind of surprised, really--nothing I make ever works right the first time.

If you used a real 2n3904 you should be able to use the documented pinout. You CAN use pretty much any general purpose transistor in this circuit. But your problem is probably (1) - yes, it matters a great deal which lead is which on the transformer coils; otherwise the current in the primary will be the wrong direction to turn the transistor off.

About the toroid, is there a certain way to wind it? As seen on this website, about 10 turns were wound with thinner wire. Mine looks more like the picture below. How should I do it?


What do you mean by "real"? And thanks, I'll try making another one.

What does the toroid do? I've made these before and loved them, just never understood how they worked (something that i hate)

current at first can only go into the transistor via wireA. When the transistor turns on current goes through wireB. When current goes through wireB it induces a reverse current on wireA, which turns off the transistor. When the transistor turns off wireB produces a high voltage pulse (Inductive kick). Before that voltage gets too high the LED (which is also a diode) starts conducting, taking up current and pulling the voltage down. When all of the current is used up, electricity can flow through wireA again. And thus the cycle starts again. I think, I mean it makes sense.


10 years ago

The coil gave me headaches the 1st time, but I found that the schematic wasn't clear to me and I had it hooked up wrong. Here is a rundown of the connections: Have the flat side of your transistor facing you. The - of the LED and the battery connect to the left leg. The resistor connects to the center. One of the single leads from the coil and the + of the LED connects to the right. The other single lead from the coil hooks to the resistor. And what tripped me up is that the battery + connects to the 2 connected coil leads. Hope it helps!


10 years ago

Okay, so I've built a lot of these (of course). Basically it works on the first time, every time, if you start with the right materials. The only odd thing that I have found is that certain ferrite toroids-- I think cheap high-resistance types designed for use as rf filters and not as inductors-- just never work. And if you really do get them out of old junk, you could end up with... junk.

So, get one new from Digi-Key or BG Micro, or if you must scrounge, start with one that was already wrapped many times with wire-- that's *usually* the right type, once you cut the wires off. If you find a toroid that just has one wire going through the center, or only one turn, it's often *not* the right type.

These things are not that sensitive to the number of turns, or to the thickness of wires, or so forth. You do need to make sure that all of the solder joints are correct, the LED is hooked up the right way, and so on. The transistor pinout in the circuit diagrams is correct for the 2N3904 and other "EBC" NPN's. If you really think that you've goofed up the wiring, go ahead and swap the leads as westfw says; however I've never needed to do that since I started using the wiring pattern shown on that web page with two colors of wire.