Joule Thief is not working

[geratheg]

Applying 3 V, the LED lights up, but applying 1.5 V, no light.
Green wire is 22 ga, blue wire is 26 ga.
Using a 2N3904 Transistor and a Ferrite Toroid.

Am I doing something wrong?

[void_error]

Try to keep the circuit more compact and the inductor wires shorter. Also, the breadboard has some (small) capacitance between the strips so also take that into account.

Increasing the number of turns on the inductor could also help. I used 15+ turns on each winding and it worked every time with almost any transistor. Reducing the value of the resistor may also make it work, but it'll also reduce efficiency. Don't go below 100ohms.

[Asim]

Am i blind or the transistor and the led are not grounded ?

[geratheg]

I tried to do more turn, but the bulky green wire ahh!

Lol, I guess I may just use the same color blue wire and maybe mark one of them with a sharpie or something so I know which one's which. Don't have another color that's smaller, sadly.

Also do the wires need to be copper/ magnetic? My wires are just normal wires.

Am i blind or the transistor and the led are not grounded ?

What do you  mean by grounded? Look at the negative on the breadboard, it's connected to the LED and transistor.
I mean it worked with 3 V.

[Stonent]

Am i blind or the transistor and the led are not grounded ?

I'm wondering that too.

[miguelvp]

Am i blind or the transistor and the led are not grounded ?

I'm wondering that too.

It's the angle of the picture, the resistor is connected to the base (between the two white jumpers) and the ground to the emitter of the npn transistor.

Then the cathode of the LED is connected to the ground too. Could be he is using a pnp instead? or the LED is backwards? hard to tell from the picture.

Edit: image linked from www.evilmadscientist.com

[geratheg]

Does my circuit look correct guys? Maybe it's not enough coils on the toroid?

Hey what's a PNP? Maybe that's what it is?

(My transistor package says NPN.)

[miguelvp]

Here is where I pulled the schematic from, it has more information maybe it helps.

http://www.evilmadscientist.com/2007/weekend-projects-with-bre-pettis-make-a-joule-thief/

my guess, wire too thick, or LED is backwards but can't tell from the picture. Or maybe a busted LED?

Edit: Or as void_error mentioned the leads on your inductor are too long?

[geratheg]

http://www.instructables.com/id/Making-A-Simple-Joule-Thief-made-easy/?ALLSTEPS

5 wraps of wire.

Does it have to be copper wire?

[Richard Crowley]

Does it have to be copper wire?

That seems like a curious question. What would you propose as an alternative?

[IanB]

Well, silver wire, clearly...   

[geratheg]

Lol, silver wire. I'm new to electronics.

I got it to "barely" work using another toroid found from an old computer motherboard, which had one thick wire around it, so I just wrapped another copper wire around it.

Used it the toroid with the 1.5 V battery and  the LED turned on pretty dimly, and only lit up if I just touched the battery, but wouldn't stay lit. Didn't observe this before. Probably a toroid issue more than a transistor issue.

[IanB]

Your circuit is probably not oscillating very well, maybe not at all. If the circuit is working the LED should light up quite brightly even with a faded 1.5 V battery.

I've made one work with a toroid core, but I think they work best with a ferrite bead of the kind used for EMI suppression.

Make sure you have the polarity of the windings the right way round. It won't work at all if you have the wires connected the wrong way round.

[T3sl4co1l]

Is that toroid *really* white on three sides and gray or neutral on the other?

I don't think it's a ferrite toroid anyway, but that's good, because you don't want ungapped ferrite.  But you don't want too little permeability, or you need way too many turns.

I hate that stock circuit.  It doesn't have enough gain into the base.  The 1k is burning away all the drive.  I use this:

with little or no base resistor.  The RC time constant of Rbias and Cbb should be on the order of the oscillating frequency, so that base current runs out as the inductor fills up with current, then the transistor comes out of saturation, then turns off sharply.

Tim

[geratheg]

I bought the torroid, it said ferrite on the package.

Since I'm still new to electronics, now I'm curious as to how this works. I'm assuming through induction: the opposing magnetic fields induce current in each wire, so more current flows in each wire, thus boosting voltage. Is this the correct explanation?

And to be honest, I barely know how a transistor works. Just liked the idea of the Joule Thief and decided to replicate it.

I grabbed a book on electronics recently though. Still waiting on parts to start learning.

[miguelvp]

Think of the transistor like a faucet.

There is water pressure at the collector (source) but the faucet is preventing flow to the emitter (drain).
By applying pressure (opening the faucet) at the base (I guess is more like a pressure valve than a faucet but you get the idea) it allows flow (current) to go from the source to the emitter.

[IanB]

Since I'm still new to electronics, now I'm curious as to how this works. I'm assuming through induction: the opposing magnetic fields induce current in each wire, so more current flows in each wire, thus boosting voltage. Is this the correct explanation?

The very short and incomplete explanation is that current flows through the transistor and builds up in the primary of the coil storing energy in the magnetic field. The circuit is arranged so that as soon as the current reaches a maximum the transistor turns off and blocks the current flow. The energy in the coil now has to go somewhere as the magnetic field collapses so the current generated by the collapsing field flows through the LED. As soon as this has happened the transistor begins to turn on again and the cycle repeats.

[geratheg]

Thanks for the explanations.

I think I should try copper wire.
I have a T50-7 toroid, with about 7mm ID and 13mm OD and 5mm thick.

Which size wire should I buy? 24awg, 26awg, 28awg, or 30awg?

Possibly wire that could be used for many other applications?

Does the size of wires determine anything besides the amount of current? Does it for example determine the amount of induced current, or is it that amount of windings that matters more?

[Richard Crowley]

Yes, copper wire would be the typical "choice".  I strongly suspect that 99.99999....% of everything you see discussed here uses copper wire (or copper on PC boards).

The wire gauge only determines the current-carrying capacity.
22ga is a common general-purpose size, although that may be too large (or too small) depending on the kinds of things you will be doing.

For that circuit, the winding that goes to the base of the transistor can be very small (like 30ga) because there isn't much base current.
OTOH, the other winding supplies current to the load (the LED) and should be large enough to comfortably carry the expected load current.
Of course, using a larger wire gauge (as they show using the same gauge for both windings) does no harm for the base winding.

[IanB]

I think I should try copper wire.

You appear to mean enameled "magnet" wire, as opposed to plastic insulated wire.

The advantage of enameled wire is that the insulation takes up less room and the wire is available in smaller sizes. Both of these mean you can get more turns of wire in a given space.

For a joule thief, the nature of your toroid core determines how many turns of wire you need. Unless you have just the right core material you may need 20 turns or more, and this could be hard to fit with regular plastic insulated wire. This is why a ferrite bead is often the best choice of core--it needs fewer turns.

[T3sl4co1l]

That's what I thought that looked like...

T50-7 has very little inductivity, 0.0043 uH/t^2.  So you need 15 turns just for 1uH, and something like this might be more typical with 25uH, or 75 turns!  With only 7 turns on there, you've got more like 0.25uH, which probably isn't enough to run at any frequency (certainly not well).

On the plus side, you'll easily be able to get 75 turns in that core if it's, say, 34 gauge or smaller, which is fine for the current level.

Last one I made was a rather beefy:



Somewhere around a T30-35, not sure if it's standard.  Some grab bag item.  Think it was around 0.015 uH/t^2.  But, this is a much more powerful converter, so it uses a proportionally smaller inductance (rated for much more current), so that's good.  You can think of it as a much thicker, stiffer spring; it takes more force to push energy into it, but it stores more energy too.

20 turns, primary with 2 x 26 AWG wire (better than a single solid wire due to skin effect), secondary with 37 AWG (because it doesn't take much base current, or I'll just end up connecting resistance in series anyway).  That makes 6uH, which should be something like, from 1V supply, it charges to 2A peak in 12us, for an operating frequency around 40kHz.  (Actually I think the current is less than that, and the frequency higher.)

Rbb is switchable 100 ohms or 1k, for brightness control (or open circuit for "off").  Cbyp is, I think, 1uF.  There's also 47uF bypassing the supply.  The transistor is a PBSS4540 high current, low-Vce(sat) type.  As I recall, switching performance (as determined by an oscilloscope) was worse with any base resistance (besides the winding itself), so I left that off.  The output is simply strapped across a 1W white LED to ground.  It's pretty bright on a fresh AA cell!

Tim

[geratheg]

Wrapped 30 gauge wire, which took a while since it's so thin. Made 29 turns.

Still didn't work. I double checked the circuit, everything appeared correct.
I thought maybe it's the toroid.
But then realized the 1Kohm resistor affected the performance.

I kind of connected the transistor without a resistor (shorted it basically) and the light bulb just lit up.
The transistor did get hot, so I'm sure a resistor is necessary.

How do I determine which resistor value I need?

[void_error]

No resistor is a bad idea, the transistor could be damaged by excessive base current.

I slapped together s small circuit and it worked right away, with a wide input voltage range (see pics).

A 100-330ohm resistor should do. I used a 150 ohm resistor, 2x 8 turns on the inductor (took core out of a dead CF lamp) and a BC337 NPN transistor.

This is how your circuit should look like:

[T3sl4co1l]

No resistor into the base is a good idea.  No resistor at all is a bad idea.  Use the Rbb + Cbyp shown in my circuit.

Tim

[geratheg]

No resistor into the base is a good idea.  No resistor at all is a bad idea.  Use the Rbb + Cbyp shown in my circuit.

Tim

I actually learned something from your post: that toroids have different inductance values. Thank you.
Except in the specs it shows the letter AL. Can you help me understand what that letter stands for?
Here are the specs I'm looking at: http://kitsandparts.com/mtoroids.html

Also can you clarify what Rbb and Cbyp is? I'm new to electronics and don't quite know all components and abbreviations yet.

On a side note, the LED worked with a 270 ohm resistor to the base of the transistor. Is that value too low? I wonder why it doesn't work with a 1Kohm resistor like for you guys.