A "Joule Thief" or Voltage Booster powering 2 white LED's with a single, "dead" AA battery

My first "Joule Thief" powering 2 white LED's with a single, "dead" AA battery

I’ve been wanting to do this for many months.  I finally got some scrap parts together and made my first Joule Thief.  This circuit milks the last bits of energy out of your “dead” batteries.

See the Video, pictures, circuit diagram, and observations…

A "dead" AA battery.  The tester has its needle in the red section indicating it is too low for normal applications.

A "dead" AA battery. The tester has its needle in the red section indicating it is too low for normal applications.

I took a “dead” AA alkaline battery out of my scrap battery collection, connected it to the circuit, and watched it power up (2) white LEDs.  This circuit uses common parts and was basically free to make since I used salvaged parts from scrap.  This little circuit gave me a feeling of euphoria when I saw it working.  It’s like magic!

VIDEO: Joule Thief from Scrap Parts

This circuit is also a kind of “boost converter”.  I originally found a version of the circuit from a video by morpher44 titled “Joule Thief Circuit – Driving CFLs with 1.5v“.  Then I found the website where the term “Joule Thief” originated.  [ http://www.bigclive.com/joule.htm ]  The original circuit was apparently first published in a magazine dated 1999.

Circuit Diagram

Joule Thief Circuit Diagram

Joule Thief Circuit Diagram

Parts used:

  • Transistor – I used On Semiconductor BC547B.  I suspect many different kinds will work, including: BC547B, BC549, BC337, 2N3904, and 2N4401.
  • Capacitor:  1,200 pF (marked “122″)  [ Capacitor Markings ]
  • Resistor:  960 Ohms
  • Toroidal inductor [ home-made toroidal inductor with 20 turns ]
  • LED -  any color (I tested with red, white, blue, and infrared)
  • Battery – So far I have seen this circuit run with a battery voltage as low as 0.45 V and as high as 9 V.

Power Consumption

Here’s a setup where a single AA battery is powering (9) LED’s.

A single battery at 1.25V powering all (9) LED's.

A single battery at 1.25V powering all (9) LED's.

The transistor I am using, On Semiconductor BC547B, has a maximum continuous IC of 100mA.  That’s not much, but plenty for a few LED’s.  But what is the current draw while powering (9) white LED’s from a single AA battery?

(9) LED's lighting up a cubby hole.

(9) LED's lighting up a cubby-hole.

I thought I better get the meter out and measure the current draw to make sure I’m not frying the transistor.

Current draw from a single AA battery:  24mA.

Current draw from a single AA battery: 24mA.

No problem.  The current draw is well within the transistor’s specifications.  I realize that a multimeter does not measure the current spikes, as an oscilloscope could.  What this reading tells me is that even if the current were 4 times higher during the ON cycle, the transistor would still be OK, as the transistor I am using can handle 100 mA continuously.   This transistor can handle a peak current of 200mA, as long as it is not continuous and as long as the transistor does not stay ON for long.  [ Source:  BC547B data sheet (PDF)]  [Edit, Aug-15-2014:  The transistor is still holding strong after 3.5 years of use!]

Based on these two measurements, we can calculate the total power draw from the AA battery, in milliWatts.

(9) LED’s lit up by this Joule Thief consume 30mW.

How much current does the Joule Thief provide?

From my tests, it doesn’t matter how big the battery is, whether it’s a tiny button cell or a big D cell.  The current appears to be determined by the voltage of the battery.

When I connected a 9V battery just for a second, the LEDs were super bright and blinding.  To safely connect a load to the Joule Thief, I think it’s wise to make sure it oscillates, which means the battery voltage should be less than what the load requires.  This will protect the load from getting too much current.  A 9V battery is a bad choice for an LED.  A single cell battery at less than 2 V is good for most LEDs.

One thing I like about the Joule Thief is there is no wasted power on a resistor limiting the current flow.  Compare this to a voltage booster that supplies a constant voltage higher than is needed, which would require a resistor to limit the current flow.  The Joule Thief requires no resistor to limit current flow.

What kind of LED could be used in the Joule Thief?

Any LED may be used with the Joule Thief. The toroidal inductor acts like a transformer to boost the voltage to whatever the LED needs.

“The magnetic field collapses, inducing however much voltage is necessary to make the load conduct”. [ Source ]

ANY LED could be used, no matter how large or small the forward voltage is, since the field collapses once that path in the circuit is opened, creating a current flow custom made by the circuit for the LED installed.

LED Color Affects Current

As the color of an LED varies, usually the forward voltage varies.  The forward voltage of my collection of LEDs are all different, mostly depending on the color.  As the forward voltage changes, the current changes.

(2) white LEDs used 11mA.

(9) white LEDs also used 11mA.  The color is still white, and these LEDs run at the same voltage, so the current flow through all LEDs is unchanged.

(1) red LED used more:  12mA.  Red LEDs have a lower forward voltage than white LEDs.  It still seems ironic to me that a single red LED uses more current in this circuit than (9) white LEDs.  The red LED also makes much less visible light, so it seems like a waste of current.

(1) IR LED used the most:  31mA!  Infrared is nearly invisible to the naked eye and yet uses the most energy, because the forward voltage is the lowest of all the colors I tested.  (My IR LED has a forward voltage of 1.5V.)  A lower forward voltage results in more current making it through the Joule Thief as it powers the LED on and off more per unit of time.  The LEDs with the lowest forward voltage have the most ON cycles, allowing more current to pass through the circuit.

Buzzer works with a diode

When I connected a buzzer as the only output of the Joule Thief, it would not make a sound.  I verified that the buzzer worked with the batteries alone, but when the buzzer is in the Joule Thief circuit it remained quiet.

I fixed this by adding a diode and capacitor.  The buzzer made sound if I altered the circuit so that there is an LED (or diode) in front of the buzzer and put the buzzer in parallel with a capacitor.  The capacitor is probably not necessary but the diode is important to rectify the current.

Lowest battery voltage that powers an LED in the Joule Thief

When the battery gets down to about 0.45V, it is still lighting the LED’s, but at a much dimmer level.  If I disconnect that battery and immediately reconnect it, the LED’s will not turn back on.  It seems the voltage is not high enough to get the oscillation started.  (This circuit is an oscillator.)  If I set that battery aside for the day and come back to it, the voltage in the cell increases again, maybe to about 0.65V.  Now the same battery will start the Joule Thief circuit oscillating again and will light the LED’s.  So, if you want to keep the LED’s running, don’t bump the circuit and disturb the oscillation.


This is an amazing little circuit and has great potential.  I’ll be exploring this further as a way to make use of low voltages.


Related Topics and Further Study

Handmade Toroidal Inductor Here at ElPerfecto.com

Wikipedia page on Toroidal Inductors:

Forum #1 – Tons of comments at overunity.com:

Forum #2 – Useful comments at energeticforum.com:

Forum #3 – Interesting comments:

Websites with Joule Thief experiments:


  1. Watson says:

    When the supply voltage to the JT becomes excessive (more than 5V), the excessive voltage will damage the transistor. The circuit may still work, but the transistor’s current gain may be permanently reduced, and it takes only a fraction of a second.

    JTs are fun to experiment with, but connecting parts up to it in a haphazard way is asking for trouble, and the typical failure mode is you “let the smoke out”.

    I suggest you use different transistor types. The BC547 and 2N3904 are poor choices for a JT. Sort of like putting a little 4 cylinder engine in a race car. Try the BC337 or 2N4401, which will increase the LED brightness considerably.

  2. Earl says:

    Thanks for your comment. I confirmed that the gain of my transistor went down since I put it into this JT circuit. The gain went from 318 to 246. There may have been some slight damage to the transistor after attaching a 9V battery for several quick connects. But I’m learning a lot by simply trying whatever pops into mind.

    How did you come to the conclusion that 5V is the limit for this circuit?

    Yes, the transistors you listed probably would be better choices but this is what I had on hand. I am going through my scrap electronics looking for other transistors to try. However, my first JT is working like a champ, so I’m glad I made it.

    What transistor specification is the critical one? Is it the maximum frequency that it can switch, the breakdown voltage… or something else?

    I’ve read several of your blog posts and am really interested in your experiments. I like the one about trying a non-ferrous toroid and sharing your results. I’ll be back to your site to read the rest.

    Thanks for stopping by.

  3. marshalll says:

    can i use this to boost the output of a peltier device??

  4. Earl says:

    I think many different DC sources could be used, including the Peltier cooler (a thermoelectric device). In my tests, my Peltier produced up to 1.05V, which is plenty for the Joule Thief to work with. [Related Post: Peltier Cooler Generates Electricity]

  5. Earl says:

    You asked me on another page about the coil. You can see the details about the coil at this post:

    Handmade Toroidal Inductor


  6. danilo says:

    hello another question,,I have a working JT,,but when I’m measuring the CE voltage (using DMM) with or without LED, I can only measure above 0.8V..I know my JT is working because it can light up my LED with source voltage of above 1.1V.. how can I measure? the output voltage correctly..

  7. Earl says:

    Because the voltage is ever changing as the current pulses through the load, an oscilloscope would be used to measure the voltage accurately.

  8. Watson says:

    The reason the JT is limited to 5 volts is that with the same number of turns on the primary and feedback windings, whatever voltage appears across the primary winding also appears across the feedback winding, but opposite in polarity. If the collector has 5 volts, the base has negative 5 volts. The typical silicon transistor has a maximum of negative 5V across the base to emitter junction (or sometimes -6V), the circuit is limited to 5V. You can change this by changing the ratio of turns in the windings. You could use 20 turns for the primary and 10 turns for the feedback, which would allow 10 volts on the primary.

  9. ramon says:

    one joule thief with battery 3.7v and a light bulb 220v

Leave a Reply