Joule Thief Is Not Working

[(*steve*)]

How do I determine which resistor value I need?

Something less than 1k, and something more than zero

Because the inductor is essentially an unknown, the required resistor can't easily be calculated. If you have more 1k resistors, try 2 in parallel, or three. If that still fails, try 220 ohms, or even 100 ohms. I'd be looking at the highest value which gives a reasonable light output.

As discussed above, the circuit is hard on the transistor, a higher base resistor will reduce this a little.

 

[BobK]

Mine works with a 1K and a little less bright with a 10K, so the resistor is not very critical.

Bob

 

[geratheg]

Tried it with a 270 ohm resistor, worked perfectly.

I wonder why though? Why doesn't a 1k resistor work for me, but it works for you?

 

[(*steve*)]

Perhaps the transistor has a much lower gain, perhaps the core has very low performance, perhaps you're not holding your tongue at the critical angle :-D

If you feel like experimenting, try turning the transformer around. swap the windings you use for the collector and base. Depending on the exact configuration of both windings, one way may work better than the other.

 

[geratheg]

Turned the transformer around, didn't work very well.

 

[BobK]

How many turns did you end up with? And how did you wind it?

The one I wound has 22 turns with the wire doubled, then the loop cut to make two windings that are perfectly interspersed.

I am using a 2N3904.

Bob

 

[geratheg]

29 turns. 30 gauge wire. Perfectly wound, side by side wires.

 

[BobK]

Funny then that it needs such a low resistor, since mine works okay with 10K. It must be the transistor. Can you try a new 2N3904 or PN2222?

Bob

 

[geratheg]

I've tried another 2N3904 with similar results. I contacted an ebay seller that sells Joule Thiefs who said that it may have been the toroid which has very low inductivity.

I ordered some of these: http://www.ebay.com/itm/8-pcs-10x6mm-ESD-EMI-ferrite-cores-toroid-cores-18mm-outer-dia-/141281353212

The seller even has joule thief kits for sale at decent prices, and I think this is what he used for his toroid Joule Thief. He also has toroidless joule thiefs and such. Interesting.

He also has various cheap components. The metal film resistors he sells are really cheap.

 

[(*steve*)]

I think I mentioned previously that the core might be the problem.

The blocking oscillator is a great thing to experiment with, but I'm not sure it's a great practical circuit.

It's great strength is that you can get it working with the combination of almost any transformer and transistor, but this comes at the price of efficiency and stress on components.

 

[(*steve*)]

More on the core. Saying "ferrite core" doesn't tell us a lot. There are many many different types of materials used. They will differ in the range of frequencies they work best at, losses will differ, and the amount of inductance you get for a given number of turns. The last variable is one of the most significant.

The quality of your resulting transformer (and I mean this literally -- there is a measure of this called Q) will vary significantly if you use a ferrite core designed for rfi supression as opposed to one designed for use in a high frequency inductor.

Beyond this there are issues with core saturation which, while they often won't stop a blocking oscillator, can make the transistor run very hot. (auto correct just turned transistor into transformer, which could have been unfortunate had I not caught it)

 

[geratheg]

The toroids that I linked to in the previous post work amazingly well. 6 wraps of wire, could probably use less wraps, but with 6 wraps the LED lit up brightly. As far as resistor values, used various up to 10K into the base and the LED was still pretty bright.

Also for experimental purposes, I tried something else. If I plugged the negative lead of the LED into the base of the transistor, it would still light up though not as bright.