# Joule thief point

I have this circuit from a flashlight that runs an LED with just 2xAA batteries. Initially I thought it used a boost converter or a joule thief but when I opened it up I found that it did not. It depended merely on the 3V those two Alkaline or Zinc Carbon cells output in series to light up the LED at it's minimum voltage at 3V. Problems arise when the voltage of these primary cells drop after a while making the light very weak.

So I decided I want to modify it and make it run on two NiMh cells at 2.4V. I have attached the circuit below. Can I add components at just these 2 green points to make the joule thief work? Also there is not much space inside the flashlight to place a toroid, just enough space to squeeze a transistor and a couple of tiny components. I hear there exists Joule thief that uses discrete axial inductors and or ceramic caps?

Update:

Here is the latest Schematic. There is an additional point where I can add a component, that is the resistor point. I can de-solder the resistor (orange point) and add a component there if needed. The green points remain unused so anything can be added there.

• There are a variety of circuits now that appear to go under that name. It should be trivial to find them on the web, where you can also readily see their size as well as the stuff you need to do. You cannot add things to your green dots, though, to make this work. It's a different "black box" setup.
– jonk
Oct 2, 2017 at 13:02
• L=v*dt/dI, so if you have boost converter , you need a diode, L, C and logic level fET such that di>=20mA and determine f >1/dt such that L/dt>=2.4V/20mA and DCR of L is small, making part big. Oct 2, 2017 at 13:16
• I will update my schematic soon since there is also another point where I can add a component. Oct 2, 2017 at 13:25
• "joule thief" ha.. I learned a new term today .. thanks Oct 2, 2017 at 14:25
• @Trevor Glad to hear that. :) Oct 2, 2017 at 14:40

TI.com suggests this solution. They have complete BOM and layout. Good luck.

• That looks like a solar garden light circuit Oct 2, 2017 at 13:41
• could be, but can fit on 36mm² PCB with SMD , Ven is just the battery switch. Oct 2, 2017 at 13:48
• I also have a charge pump in mind. Do you think this is a better choice than a step up converter? Oct 2, 2017 at 14:41
• Only if you can beat 90% efficiency, which I doubt. Oct 2, 2017 at 18:14

Using inductive kickback, it should be possible to simply add a circuit between those two points. But I do NOT recommend it. It would be better to use a better solution.

The battery won't like this, but there is no way to protect it from between the two points.

simulate this circuit – Schematic created using CircuitLab

Not drawn is the clock source, but it can get its power from the same two points, but it needs to be able to run on low voltage.

The zener diode is there to limit the voltage for the clock source.

If you want to test this circuit, you should add a protection diode to the battery. And then you can go ahead.

L1 charges through D1, R2 and M1. And then discharges through D2, D3 and the load. While capacitors store voltage, inductors store current. If you let 15 mA (or something) flow through the inductor when it's fully charged, that same 15 mA will flow through D3, R1, LED and D2 as soon as M1 turns off.

The value for the inductor depends on the clock frequency and the load and the current. If you want good results you need to do some numbers, but pretty much anything will make the LED flash in a dark room.

• looks good, I'm up-voting. I'm planning of using a charge pump since it's small and uses fewer components. Oct 2, 2017 at 19:44

Not sure I can answer your question of how to put a Joule Thief in your project. But I did find a Joule Thief on Amazon that, when it was available, might have fit. Even though the product is currently unavailable, I included it here because many people don't know how small a Joule Thief can get. This is a single inductor Joule Thief with a 490uH surface mount inductor.

The size of the board is small, at 10mm x 16mm. The inductor is the first component on the left, right next to the input (Power) terminals.

The ones I purchased worked fine. I simulated this circuit with two 2N2222's and a 1nF cap. I think I've seen this schematic design elsewhere on the web, but I don't remember where.

• But, anyways, it is possible to manually build a joule thief even smaller. Big Clive show this one another photo. It is possible to build it, I've already a bunch of them. Jun 19, 2020 at 3:01
• @mguima -- Yes, I wanted to do what Big Clive did, but I didn't have that nice, small ferrite bead that he had. I'm curious, also, how much inductance he was able to achieve with it. Do you know how much inductance yours ended up with? Jun 22, 2020 at 1:31
• All the beads that I used for build joule thieves were collected from burnt CFL; for a while, I took apart dozens of lamps. About 5% of them had a small bead that, although bigger than Big Clive's beads, were small enough to fit in a flashlight lamp case (the small ones came mostly from known brands like Philips, Osram). It's been almost ten years that I've built a lot of joule thieves, but I didn't have an inductance meter at that time. Nowadays I have one and made some measures; take a look at this photo Jun 23, 2020 at 14:14
• Anyway, the common experience (my own experience and mostly the collective experience spreaded all over the internet) with JT's shows that, if the objective is just turn LED on with a basic JT circuit, the inductance value is not critical - whatever coil you use, the LED will lit. The inductance of both windings becomes critical, of course, if someone intends to tweak JT for maximum efficiency; in this case, everything matter: the different number of turns on primary and secondary, bead's size, wires AWG, and so on. Jun 23, 2020 at 14:29
• Here is my experience related to how inductance value affects efficiency. There are two ranges I like: 68uH-270uH for nice and bright, having as low an ESR as is practical. For these, I try to keep oscillation frequency above 22KHz to prevent high-pitched audible noise (by increasing the resistance determining the current). The second inductance range, 2mH-to-10mH, tends to make audible noise, but has much higher efficiency. I can't seem to make these inaudible, but some coupled-inductors (or common-mode chokes) are quieter than others, and/or can be placed inside of sound-deadening material. Aug 23, 2020 at 18:13