Maximum voltage achieved with my Joule Thief was 20.5V.

Maximum voltage achieved on the Joule Thief's output was 20.5V.

My first Joule Thief had an LED as the load.  When I replaced the LED with several different capacitors, they all charged up to the same low voltage:  0.4V. This surprised me at first because the LEDs ran at 2 to 3 Volts, but the capacitors did not get charged that high. 0.4V is also about the lowest voltage I have seen the Joule Thief work and light up an LED. Probably not a coincidence.

Capacitors can achieve much higher voltage when a diode or LED is included.

Capacitors can achieve much higher voltage when a diode or LED is included.

Capacitors do not get a full charge without a diode in place to rectify the current. You can use an LED as the diode. When I used an LED, I got a capacitor to charge up in series with the LED, as shown in the sketch above.  Adding the LED increased the max voltage across the capacitor from 0.4V to 20.5V.

I now assume I could intermittently power a load with this Joule Thief up to about 20V.

I have successfully powered a couple of 12V LED strips with a single AA battery and the Joule Thief.  The light output was dim, though.  But I guess we could go as high as 20V with this circuit.  However, I would not choose LEDs at that higher voltage output, since the light output gets too dim at the high voltages.  Maybe the resistor could be lowered to allow more current to flow.  But then we have to make sure the transistor has a high enough rating for IC.

Replace Battery with Capacitor?

Can a Joule Thief work with a capacitor as the power supply?  When I replaced the battery with a charged capacitor, the LEDs did light up, but not for long… just about one second.  A higher capacity capacitor (or super cap) would probably increase the time the LEDs stay on.  But then you have to be careful about the voltage level of the capacitor.  The voltage of the power supply should be less than the LED forward voltage to protect the LEDs from burning up.  But if you are not using LEDs as your load, I think you could use a higher supply voltage.  How high depends on how much current you will generate and what current your components are capable of handling, including your transistor, gauge wire, and the power rating of your resistors.


There are so many possibilities for the Joule Thief and I’m just getting started.  If you want to keep up-to-date, you are welcomed to subscribe to the RSS feed.  And I always welcome your comments and feedback.


  1. Watson says:

    You’re really pushing the envelope. What I mean is that the LED is typically rated for 5 volts reverse voltage, but you’re subjecting it to much more than that: 20.5VDC. Actually more than that because it drops a few volts, so that has to be added, too. Maybe 24 or more volts peak. Yet the LED still withstands 4 or 5 times its rated reverse voltage. I’ve read that the LEDs can withstand several tens of volts, as much as 50 or more volts. I’ve never tried to test that claim out, but now you’ve done it for us.

  2. Earl says:

    That’s an interesting point. I admit I didn’t think anything about the reverse voltage, but now that you mention it, I’m surprised my salvaged LED held up.

  3. Nipuna says:

    Actually the LED isn’t getting any reverse voltage. at the point it stops conducting, the transistor starts conducting- essentially putting a short circuit across the LED.

  4. Abraham says:

    You certainly are “pushing the envelope”, Watson was right there… he just didn’t understand the circuit well enough to recognise that it’s the transistor under stress, not the LED.
    The voltage spike that appears on the collector is duplicated at the base, inverted. So the reverse base-emitter limit (6V for the BC547B) is being exceeded by a large margin if your transformer windings are equal. You may find the gain of the transistor has dropped as a result, I’m surprised it’s survived at all. If you want to use the circuit to generate voltages above 6V it would be prudent to alter the turns ratio to reduce the feedback voltage, i.e. more primary turns or fewer secondary turns.

  5. Abraham says:

    Ooops! Apologies to Watson.
    I shouldn’t have written that in a rush. Closer inspection of your circuit diagram reveals that he is, indeed, correct and using a LED as a rectifier diode in that way is placing it under the same reverse voltage stress as the base-emitter junction of the transistor.

  6. Earl says:

    Hi. Since you had a change of mind, would you like me to remove the comments to help reduce confusion?

  7. Abraham says:

    No need on my account, Earl.
    I don’t think it will confuse anyone… and as for the slight embarrassment of having to acknowledge my mistake in public, it won’t kill me!
    I see this is a fairly old thread, are you still experimenting with circuits derived from the Joule Thief? It’d be interesting to know how long the transistor and LED lasted when subjected to voltages so far in excess of the manufacturers ratings.

  8. Earl says:

    Hi, I still have an old Joule Thief circuit running 24 hours a day. It is currently lighting up a gold-colored LED from a single AA battery. The components are durable, and they were scrapped part to begin with. To see some of my other Joule Thief projects:

  9. rhum says:

    hi sir,

    i have done your joule thief with 4.7uF/10v cap, it gave me 22v, but how can i use it for some applications? ei. simple audio amplifier? i tried to use it but it didn’t work the way i expect… help me how to rework it.


Leave a Reply