Picture of High efficiency regulated Joule thief
     Normally the Joule thief produces output voltage, which value is difficult to predict. Without load (the LED) I have measured voltages over 30 V. I wanted to create a Joule thief, which can be used to supply some small electronic devices, but having well defined and stable output voltage. There are known some solutions in which instead the LED load, a one-diode rectifier is used, and the output voltage is stabilized by the use of Zenner diode. I did not like this solution, because through the Zenner diode flows always a constant DC current, what  reduces drastically the efficiency of the device and empties fast the supply battery. I was looking for other, better solution of the output voltage stabilization (limitation).
     To try my solution, at first, I needed to build the standard working Joule thief. How to do this - there are a lot of articles and internet sites (for example this). How to find the needed parts? - I knew that inside the high efficiency lighting bulbs are some parts, which could be re-used. I had a defect bulb and I carefully cut the plastic box. From there I extracted the voltage converting board. On these PCB's can be found some very useful stuff : HV diodes, chokes, HV capacitors, HV transistors..etc (HV means high voltage ~ 400V). I took the ferrite toroidal transformer, cut and removed all its wires. After that I disassembled the choke. I took around one meter enamelled wire from it, and winded it around the ferrite bead. Because the wire was fold, I winded simultaneously two coils having ~ 50 turns. Having the main part of the Joule thief (the transformer) ready, the remaining work is not much.. The only tricky in the design is to connect both coils in the correct way. (see the mentioned link for additional information). So designed the Joule thief was able to produce 34 V voltage measured on the collector node of the NPN transistor (2N2222) without any load, when supplied by 1.2V AAA battery (filterd with 2.2uF capacitor).

Remove these adsRemove these ads by Signing Up

Thank you (Krb686, Milen) for this discussion. I learned a lot from your comments.

Krb6861 year ago
Hey this is very cool! I love JTs and have messed around with some myself. I am curious why your circuit is as complex as it is. You have potentiometer controlled shut off, and your storage capacitor discharges through R2 and the potentiometer so that your JT must turn on and fire every so often to recharge up to the necessary level. Here is a very similar circuit I built that instead uses a zener diode runover turning on a Q2 to shutoff the JT.

Notice how the transistor leaks much less current than your potentiometer, and the JT turns on less frequently so that it wastes less current.

Hey I'm not trying to bash your design just so ya know, I've never built yours and maybe it has advantages I don't know about, but you could try mine out and see what you think! Good work by the way

Are you still active I'd like to ask a question?

Yep, whatsup?

I came across this discussion and i was trying to replicate your circuit but when I went to the letsmakerobots forum I was confused by the schematics. They seem different from the one here. Are you using a zener diode in the regulation circuit and what value was your capacitor? It looks like its a 10 microfarad. It just seems like the schematics are different between here and the robots forum. Just wondering if there was a finalized version.

Hey there, yes unfortunately the website layout on letsmakerobots is pretty terrible, especially the comment section, so I can see how you were confused.
Anyways, I do have some finalized designs that I put more work into, but what exactly are you looking for? Just a basic Joule Thief circuit?

Here is one:

Or if you are looking for a circuit that boosts the voltage of a capacitor to a desired level using a JT-like circuit and then is regulated, here is one of my finalized versions of that.

That is similar to Milen's version but a bit different and uses a zener to turn on another BJT that shorts the base of the first BJT.

Thanks! I'm looking for a JT circuit that will give me a good flow of current with decent efficiency. I just bought an oscciliscope and an LC meter so I can fine tune everything. I want to drive two 1 watt leds in series.

Also when you open those links to the circuit simulator, be sure to slow the simulation speed down and hit reset. I believe when you open those links the circuit begins as if it were at steady state after running for a long time so you won't see the circuit operator until you hit reset.

Milen (author)  Krb6862 months ago


I have closed this project, but you can ask me...

Milen (author)  Implicate_Order2 months ago
Here some answers to your question:
There are a lot of versions of the JT schematics and there is not a finalized version. You can google and you can find different and different ones. In some of them a zener diode is used to fix the output voltage. In my circuit the output voltage is clamped by MOS load transistor or is regulated by changing of the duty cycle (operation/non operation). Please read carefully the text - there is explained how it works. The performance can be improved by addimg of some additional devices (for example some PMOS switch interrupting the base resistor current path)
I use 2.3 or 3.3 uF capacitors to charge.
Before making the circuit - you can try to simulate it - I have included the simulation circuits. If you have problems with the simulator - there is yahoo LTspice group, where a lot of stuff is explained. Or you can contact me. In the simulations you can change the device parameters and to play a lot with the circuit. When you reach some satisfactory simulated performance, you can try to implement this simulated version by real devices. Its performance will not differ from the simulated
Krb686 Krb6861 year ago
Forgot to mention also, all of my work was spurned by a discussion with some at the letsmakerobots website. It is all here in this thread if you would like to see it, with the final results in a [warning] very long and detailed post on page 2.

One of the most important things I learned from playing with it though, is that changing your winding ratio so that the winding on the base has more turns than the collector allows your JT to start from lower voltages, and increases the efficiency and potential power output.
EmOSinXRhEnZ2 months ago

Hi.. would you please make a tutorial video on how to make these? it would be much appreciated.. thanks!

achand810 months ago

Can the output voltage be filtered using a bigger filter cap like 1000uF?

Milen (author)  achand810 months ago

I tried with bigger capacitor, but the JT had difficulties to go in the oscillation mode. The big capacitor is clamping strong the potential at the collector of the transistor. But with some additional devices, I think that this can be done. Simply this big capacitor must be isolated from the small one, and after the voltage on it have reached some value slowly to be connected. This can be done by the use of PMOS transistor, which gate is controlled with RC chain with big time constant. At the startup the PMOS transistor is open and the big capacitors is isolated,...with the time the potential on its gate becomes low and the transistors Ron starts to decrease until it reaches low values. Then the big capacitor appears connected in parallel with the small one... you can try this...but I suggest you first to simulate it, and if only works in the simulation, to implement it phisically.

dyntema1 year ago
Hi, one critical remark: when your end-voltage is reached you short the basis of the transitor to ground, hereby passing a lot of current through L1 and R1. in fact the schematic presented here will probably drain more current while doing nothing than when active.Try to switch the emitter of the transistor (instead of the basis) with the FET and invert the driving signal for the FET. And R3 is pretty useless, you can leave it out and short the terminals to ground directly.
Milen (author)  dyntema1 year ago
Hi, you are right... :When the basis resistor is too low, the current to ground could be this case i was intending to put in series with the resistor additional PMOS switch, controlled by the same signal, which controlls the NMOS switch. I did not put because, I did not have any available with enough low Vth, able to work with supply lower than 1.5V. I think some suitable DMOS transistors are available.
In the current implementation the DC current ( when 1KOhm resistor used ) is < 1.5mA...what is less comparing with  the load current. Off course the efficiency can be increased when higher resistor is  used and high beta oscillator transistor. If high power solution is needed, than the basis resistor should be reduced and then to reduce the power consumption the DC path from the supply to ground through the resistor should be interupted in some way - can be PMOS switch, thyristor or some other switching device.
Great design, I've been thinking of various ways of doing this, but this is COOL,so simple !! CHEERS!
The Schmitt trigger regulates the output voltage so it never reaches the max. it shuts down the Joule Thief when the set voltage is reached...