A Joule Thief is a simple circuit that takes advantage of the inductive kickback produced by an inductive element (a wound toroid in this case) to step up the voltage. This extra voltage at the output of the circuit can be used to power devices which would not work with the input voltage alone. In this case that voltage is used to light an LED, which would not be able to light up with a single AA battery as the voltage required to do so is 3.3 volts, while the battery can only provide 1.5 volts at its best.

The joule thief can operate with very low voltages, this means it can work with batteries which would normally be considered "dead", taking energy from them in a seemingly impossible way, hence it's name.

In this instructable I'll show you how to build a tiny joule thief mounted on a piece of copper clad board.

## Step 1: Materials:

• 2N2222 transistor
• 1k to 10k resistor (to select the brightness)
• Copper clad board
• 1x AA battery holder
• 1x LED (I used white, but any color will work)
• Solder
• 1x small toroid* (less than 1cm of outer diameter)
• Around 20-30 centimeters (8-12 inches) of thin copper wire
• Epoxy

*Small toroids like the one I used can be easily obtained out of small CFLs, not all CFLs have them, but there's a high chance you'll find one inside. Not all toroids will work, it's wise to test the circuit in a breadboard before assembling it.

## Step 2: Etch the Board

I first start by cutting a piece of copper clad board with the size of an AA battery, I will stick it to the back of the AA holder, so it must have the correct dimensions. In this case the dimensions are 1.2 x 5 centimeters.

To etch the board I stick some tape on both sides after scrubbing and cleaning it with alcohol, this will remove any oils left by my fingers which could ruin the process, it is also important to seal the edges too so the etchant doesn't eats them away.

Next I cut the shape of the circuit with an x-acto knife as seen in the pictures, the exposed areas will be eaten away. This disposition is fine as it is and I encourage you to use it, but you can always improve it's design.

To etch the board I use a mixture of Hydrochloric acid (20-30%) (also known as muriatic acid, pool cleaner or HCl) with hydrogen peroxide (3%) (also known as a common wound disinfectant) in a 1:2 ratio, 1 part of HCl per 2 parts of H202 . This solution can't be reused as it's unstable and will loose its power within some hours.

The etching process usually takes from 5 to 10 minutes.

After the board is etched I remove the sticky tape and I continue with the next step.

## Step 3: Winding the Toroid

Winding the toroid can be a little frustrating, specially if your toroid and wires are very small. First you must take a piece of wire around 25cm long (depending on the size of your toroid) and bend it in half so you have two wires parallel to each other, then you start winding those two wires at the same time (from 10 to 20 times) until you end up with something like in the image. As you can see, the wires should should not cross each other.

Next, you have to connect the two inner wires as seen in the image, these will be connected to positive. The remaining two wires can be connected to the resistor that will be connected to the base or to the collector, it doesn't matter which way around.

## Step 4: Completing the Circuit

The circuit is quite simple, but it can get confusing when soldering it onto the board.

I first tin all the pads on the board, this will avoid the copper from deteriorating and gives it a nice appearance, then I solder the toroid positive wire to the positive pad at the bottom, the two other wires will be connected
indistinctly to the pads at the right and the left.

I connect one end of a 2.2k resistor to the pad at the left and the other end to the "base" pad just above it. You can choose the brightness with this resistor. After finishing the circuit, if the current going in is higher than 30mA you can increase the value of the resistor in order to reduce it.

To connect the transistor I connect the collector to the upper right pad, the emitter to the upper left pad (the negative pad) and the base to the "base" pad just below it.

I finally solder a SMD LED taken from a LED bulb between the collector (positive) and the "emitter" (negative) pads. You can solder any kind of LED.

Once the circuit is complete I use acetone and a q-tip to remove most of the flux that was on the board, then I use epoxy to hold it to an AA battery holder, I use the remaining epoxy to seal the toroid and its wires, which can be easily damaged.

To finish it all I solder the positive wire coming from the holder to the positive pad at the bottom and the negative wire to the negative pad, connected to the emitter and to the negative pad of the LED.

IMPORTANT:I used a P2N2222A transistor, not a 2N2222A or a PN2222A, this means the collector and emitter are swapped, check your transistor's pinout before soldering it.

## Step 5: The End

This circuit uses around 20-30mA, this means it can last more than 50 hours on a fresh AA battery as calculated by an online battery life calculator. It's also quite bright, so it could be of some use in an emergency situation.

EDIT: I tried to see how much it would last on a fresh AA battery (a generic one), I connected it on Friday night at 24:00 hours and it's still running, without a very significant significant drop in brightness, two days later, Monday morning, almost 50 hours of continuous use. After the first day the voltage dropped from 1.48 volts to 1.20 volts, now it's at 1.05 volts.

EDIT 2: One day later it's still running, the brightness has dropped significantly, but it can still be used to see in the dark without problems (see 3rd picture) , the voltage is now at 0.78 volts. It's been running for three days straight.

EDIT3: The battery seems to have really died at some point during the Tuesday morning after 6 am, in total it's been running for almost 80 hours.

Thanks you for reading this instructable.

<p>what was the frequency of operation?</p>
<p>In this case is 44kHz, but it can depend on the toroid and windings used. </p>
<p>My jt works with 21 led's from led lamp. I use 4 bc 548 transistors, pc suply ferrite transformer and 54 micro henr ferrite coil to boost</p>
<p>That looks like a Joule thief on steroids! Nice job. </p>
Could you put several of these in series to get, say, 30v, with suitably uprated transistors? That would allow floodlight diodes to be used, but the current would be pretty hefty.
<p>You can search &quot;joule ringer by laser hacker&quot; it's a joule thief <br>working at 12V that can light up a few home led light bulbs using <br>only 200mA.</p><p>I did some circuits similar and got a CFL tube to light up at 1.5V with an amp draw of around 40mA!</p>
<p>Yes, I guess you could have a bigger toroid with more turns with a bigger transistor, but at that point I think it would be more efficient and reliable to get a boost converter. </p>
<p>Beautiful work</p>
<p>Thanks!</p>
<p>Instead of etching, have you thought about using wide Copper Tape and then cutting away the parts that are not needed? I was also looking at conductive paint to see if this would work as well. Paint it on a plastic sheet or even directly onto the back of the Battery holder? </p><p>Here is the link to Conductive Paint. I am not associated with the company that makes it or the store that sells it. </p><p><a href="https://www.sparkfun.com/products/10994" rel="nofollow">https://www.sparkfun.com/products/10994</a></p>
<p>I guess that would work too, conductive paint tends to have a big resistance, so I guess it wouldn't be as efficient. </p>
awesome! can i use 3w ultrabright led?
<p>Yes you could since the voltage drop would be the same, however it won't be as bright because this device usually outputs around 20mA. </p>
<p>Very nice! I have a question for you, You have one of the best images of how to wind the simple joule thief I've ever seen and I'm like to use it on my joule thief web page if you will allow me to do so. I'll also include a link to this instructable so show where it came from and to allow folks to get to your instructable from my page. My site is at: http:// cs. yrex. com / ke3fl / &lt;http://cs.yrex.com/ke3fl/&gt;- click on the &quot;<strong><a href="http://cs.yrex.com/ke3fl/#FEATURE" rel="nofollow">Feature Article</a></strong>!&quot; link and then click on the &quot;<strong><a href="http://cs.yrex.com/ke3fl/htm/JouleThief/JouleThief.htm" rel="nofollow">Joule Thief Experiments</a></strong>&quot; link.</p>
<p>Phil</p><p>i noticed on your Joule thief site you wound the white and green wires separately on the toroid. if you wire in parallel so they interleave you can up the efficiency, sometimes by as much as 40%.</p><p>This is largely a result of separating each winding by the wire in the neighboring winding. this reduces the capacitance between each loop. it will allow faster oscillation frequency and higher magnetic triggering between the 2 windings.</p><p>just a bit of food for thought</p>
<p>mickeypop,<br><br>Thanks for the information, I hadn't yet done any tests on parallel or seperate wiring so that's very good to know. Usually I wire them in parallel because it is easier to do &amp; takes less time that way.</p>
<p>Oh, that image you were referencing was used with permission from a wiki article, as the comment below the image states: &quot;To read the wiki article,<br><a href="http://en.wikipedia.org/wiki/Joule_thief" rel="nofollow">Click Here &quot;</a> I'm presently contacting Victor &amp; the original author of the graphic, that Victor modified, for permission to use it on my site. Again, thanks much for the additional information about separate &amp; parallel wiring. </p><p> <br></p>
<p>I've been separating my windings just a bit but it seems it's actually better to put them parallel to each other touching each other. I've seen people who even prefer to strand the wires together and then wind them like that, it seems to work better, at least for gate drive transformers. </p><a href="https://www.youtube.com/watch?v=Ozc_XsyiDrM">https://www.youtube.com/watch?v=Ozc_XsyiDrM<br><br></a>
<p>The image I used was a quick edit of this one: <a href="http://rimstar.org/sdenergy/joule_thief.htm">http://rimstar.org/sdenergy/joule_thief.htm</a></p><p>It<br> doesn't seems totally fair to me to distribute that picture, even though it's an edit the <br>original belongs to him, but liked your page so much I made you these:<br> </p><p>The second one has no background, so you can merge it with your page's background if you wish. </p><p>You don't have to quote me or add any link, I don't want recognition, seeing my work being shared is enough reward for me :)</p>
<p>Victor805,<br><br>Thanks again for the images. In this case it really doesn't matter if it is the RED or the GREEN wire that goes to the resistor and then the base of the transistor since it is the simple Joule Thief version. On the other hand, with the non-simple version it does matter which wire goes to the base of the transistor. It should be the wire with the fewer windings. You can read and see this on my Joule Thief Experiments page.</p>
<p>I know, that's why I added an &quot;or&quot;, as it doesn't matters. I'll read your website for sure, thanks! </p><p>If you need any image please let me know, I can do them in a couple of minutes. </p>
<p>Thanks very much Victor I'm emailing StevenD to ask his permission as well.</p>
<div class="comment-body"><p>It seems the uploader converted the .png <br>into a jpg leaving a black background, it also has lowered the quality <br>of the first a bit. </p><p>http://postimg.org/image/e54d4ny31/<br>http://postimg.org/image/tckcp0pxp/<br> Here they are uncompressed. </p></div>
<p>mi piace e di sicuro lo costuisco.</p>
<p>Well done. I rellay liked how you made the pcb.</p>
Thanks, I was too lazy to print one as I was running low on toner and there was a high chance of it being defective.
Nice to see that &quot;dead&quot; batteries can be used once again. Good job!
I know, right? I've tried to use it with old batteries which I was going to throw away (at a recycling point), on average I got from 2 to 3 hours of light before it started to fade and about 4 to 5 hours more with a very dim light until it turned off. It's certainly not as bright as a fresh battery, but for a battery that is considered to be &quot;dead&quot; is more than what I expected.
<p>You can get pre-made joule thief circuits from solar lights that no longer work due to dead solar panels. Just remove the wires to the bad panel or jump a LDR from + panel wire to the + battery to make it only come on when dark. Add a switch between minus wire and minus battery or the chip will eat the battery up.</p><p>http://petesqbsite.com/phpBB3/viewtopic.php?f=4&amp;t=3745&amp;p=22852#p22852</p>
<p>I like it!</p>
<p>The circuit will reduce the life of the transistor because the -ve spike is developed across the transistor base-emitter as well as the +ve that actually drives the transistor On . They don't like reverse voltage above about 3V.. You should have a decently rated diode eg 1n4001 in circuit with the +ve end to the Base and the -ve end to the -ve line.</p>
<p>I tested the thing with my <br>scope and both the base voltage and the collector voltage don't seem to <br>have negative peaks. The emmiter-base voltage for the 2n2222 is 6 volts,<br> so I guess it will be hard to damage it....</p><p>First picture is the<br> collector voltage with a peak of ~3.2 volts and the second one is the base voltage with around 1 volt peak. </p><div class="comment-body-container"><br> <div class="comment-photos"><a class="comment-photos-link" href="https://www.instructables.com/files/deriv/FSV/ZBHH/IJD78KEM/FSVZBHHIJD78KEM.LARGE.jpg"><br> <img alt="Picture of 20160113_232956.jpg" class="photo fancybox-thumb" src="https://www.instructables.com/files/deriv/FSV/ZBHH/IJD78KEM/FSVZBHHIJD78KEM.MEDIUM.jpg"><br></a><a class="comment-photos-link" href="https://www.instructables.com/files/deriv/FSC/9W8X/IJD78KFK/FSC9W8XIJD78KFK.LARGE.jpg"><br> <img alt="Picture of 20160113_232934.jpg" class="photo fancybox-thumb" src="https://www.instructables.com/files/deriv/FSC/9W8X/IJD78KFK/FSC9W8XIJD78KFK.MEDIUM.jpg"><br></a></div><br> </div>
<p>Pictures: </p>
very cool. thank you for the details. <br>now I just need to make this project and try a crystal power cell on it. I don't think It'll work, but I'd still like to try it! any thoughts?
<p>I've seen joule thiefs working fine with homemade batteries, for example in copper-zinc batteries, specially when the voltage isn't high enough. I guess it could work :P</p><p>Picture related, a lemon battery with a joule thief attached to it powering an LED.</p>
the fact it 'steps up' the voltage, dose it eat up batteries quicker?
<div class="comment-body"><p>Power equals voltage times current. In an <br>ideal case the power going in equals the power going out, although <br>efficiency rate isn't 100% as systems have many power losses.</p><p>This,<br> for example, implies that if I have a 1 volt battery with 30mA going in<br> and the LED fixes the output voltage at 3 volts, I'll have less than <br>10mA going though the LED.</p><p>Normally the LED voltage is lower when it's not running at full load, I've measured it to be around 2 volts.</p><p>In<br> conclusion, yes, the input current will be directly proportional to the<br> increase in voltage made by a circuit for a fixed output current, it will drain batteries quicker. But still, the rate of discharge of a battery used with this device is quite small. </p></div>