Introduction: "Joule Thief" Circuits, Crude to Modern...

Picture of "Joule Thief" Circuits, Crude to Modern...

It seems that many "Joule Thief" circuits depend on a clunky (bulky and heavy) toroid or "donut" that has to be carefully wound with copper wire. But now there are several very small 4 legged ICs available that do the job using only a simple inductor, single cell battery and a LED. In effect, the 4 legged IC replaces the clunky toroid.

I came across these ICs when I disassembled some solar powered yard lights. I looked for a toroid but only found a four legged IC and a part that looked like a resistor but actually was a very physically small inductor (coil). Both of these parts along with wire attachment points were soldered to a small circuit board. I was able to remove parts, attach wires to them and assemble them on a Radio Shack type of "Breadboard" to test and better understand this circuit.

But then I created a very crude and minimal circuit to better understand some of the key parts of a "Joule Thief."

Step 1:

Picture of

I used a Yellow LED that requires 2 volts (or a little more). My 1.5 volt "Rocket Battery" has been worn down to 1.4 volts. As a result, the LED is off and is not even close to conducting any current. Points A and B on the inductor coil L are at pretty much the same voltage, 1.4 volts compared to "ground" or the minus of the battery.

When the switch is pushed and and held ON, briefly, current flows through the coil and creates a magnetic field around the coil. Points A and B are still positive with point A being slightly more positive than point B.

But when the switch is released and turned OFF, the magnetic field suddenly collapses and creates a 1.4 volt voltage with a reverse polarity. This means that point B is now 1.4 volts higher (more positive) than point A. It is as if the coil has become like a temporary battery connected in series with the actual battery, presenting 2.8 volts to the LED. The LED reacts to this by flashing on for a very short moment. Pushing the switch again repeats this cycle. If I could push the switch rapidly enough, the LED would appear to be solidly ON.

The pictures that follow will reveal how simple it would be to recreate this. The coil or inductor is 12 feet of 24 gage wire wrapped (200 turns) around a 1/4 inch diameter soft iron nail.

Step 2:

Picture of

A crude "Joule Thief" circuit was created and hooked up as shown but no current is flowing.

Step 3:

Picture of

Now the switch is held down and current is flowing through the coil. The switch only has to be turned on momentarily to create a magnetic field. Also, the switch is shorting out the LED.

Step 4:

Picture of

But when the switch is released and turned off, the LED lights very briefly.

Step 5:

Picture of

Again, if I could only turn on and off the switch in a very rapid manner, I would have a working "Joule Thief" like circuit.

And it turns out that the IC I mentioned at the beginning of all of this does just that. This IC turns the LED on and off somewhere between 50,000 and over 100,000 times a second, making the LED appear solidly ON.

With the QX5252F IC connected as shown, Pin 4 on the IC very rapidly connects and disconnects to Pin 3, the negative on the battery, causing the LED to repeatedly flash similar to the crude "Joule Thief" already mentioned. But here the LED flashes so rapidly that the LED appears completely ON, more so than some florescent lights that have a flicker to them. 

Step 6:

Picture of

The picture here reveals the QX5252F IC at work, keeping the LED flashing so rapidly that it appears to be solidly ON.

The bulky, handwound coil-on-a-nail can be replaced with the tiny inductor shown between the IC and an LED.

Step 7:

Picture of

Although the IC "Joule Thief" seems to work with my crude, handwound coil on a nail, I actually use a very small 330 (orange-orange-brown) inductor that has the size and appearance of a small 1/2 watt resistor.

Step 8:

Picture of

So you can see that all of the parts of the circuit using the QX5252F IC would fit on a small circuit board with the whole apparatus taking up much less room and having much less weight than a circuit using a Toroid with two copper windings on it.

Step 9:

Picture of

Although the toroid or "iron donut" has served us well and deserves much respect, the tiny inductor along with the QX5252F IC may very well serve as an advancement to some of the readers here.

Note: If the battery is replaced by a rechargeable battery, a solar cell can be connected to "Unused" Pin 1 and Pin 3 on the IC. During the day, the IC will turn off the LED while also trying to charge the battery, depending on sunlight. During the night, the IC will turn on the LED. But this is another story since this operation takes us away from our "Joule Thief" discussion... 


pruthvi-in-instructs (author)2016-11-07

how can I increase the output power

uc1 (author)pruthvi-in-instructs2017-11-14

According to theory and also the datasheet of the QX5252 you can reduce the inductance of the inductor (L), as the inductance reduces the maximum current through the inductor increases and the maximum current through the diodes can also be increased. Look at the datasheet please. Also another factor in increasing the power output is to increase the battery voltage, causes similar effect to decreasing the inductance.

MAYBE, you could take several separate joule thief circuits, each with an added output diode and connect the diodes to a common point for power output. The voltage wouldn't increase here but the current might, hence an increase in power out. I have never tried to increase power out so all I'm offering here is just talk....

Dave Kruschke made it! (author)2016-04-29

I looked through my notes on "Joule Thief" circuits and found
that a modification to a solar garden light that I did over three years ago is
similar to what Denis Rev, below, has come up with. The difference between what
Denis has and what I have is trivial. Moreover, both the inductors and
capacitors can vary (100<L<1000uh and 1<C<100uf) and the circuit
will still work. But either a Schottky diode OR and a regular LED are needed (I
didn't try other diodes) in this circuit. I prefer an LED to a Schottky because it's common and cheap and also gives more light. My outside night time solar light ran
over a year before I had to replace the AAA nicad. The second AAA has lasted
over two years...

sitnah (author)Dave Kruschke2016-07-13

Hi Dave,

I'm a newbie to all of this. I think I did the wiring right for the slow color flashing LED circuit you show, I'm including a drawing of what I did.

My problem is when the switch is turned OFF and the sun is shining on the solar panel, the LED stays lit.

Putting a smaller solar panel (1.5V 200mA) on the circuit, and the LED stays off in the sun.

But I still wonder if it is still putting out power which is just not enough to make the LED light up, meaning power is still going thru it somehow?

Did I do something wrong?

tkennaugh (author)sitnah2016-07-14

From the circuit iIt looks like the ic relies on the input voltage from the solar panel being less than the forward voltage of the led. You will therefore need a panel with a maximum voltage around 2.5V. You would be able to increase the capacity of the solar panel by connecting multiple 1.5V solar cells in parallel.

sitnah (author)tkennaugh2016-07-16

Turns out it will light up in the sun if the solar panel is any higher than 1.5V. Even at 2V the red flashed off & on (no blue or green).

Dave Kruschke (author)sitnah2016-07-14

The suggestion by tkennaugh below looks pretty solid, although I haven't tried it. You could connect a 1 to 10 k ohm potentiometer across two 1.5 volt batteries hooked up in series. Then connect the pot adjustable wiper lead to the solar cell input with the negative of the batteries going to the negative of the IC. Of course, remove the solar cell. Using a multimeter, you could adjust the voltage going to the IC and see what voltage causes the LED to light up, just what tkennaugh is talking about below. Wiring the solar cells in parallel may be just what you need to do...

sitnah (author)Dave Kruschke2016-07-15

Thanks to both of you for the advice. Guess I'll go with a smaller solar panel since it needs to fit in an old round solar light top I have.

Amazingly, it will actually light up 5 color changing LED's before the blue starts to "fade out". I'm impressed !!

JohnO163 (author)2016-06-30


My wife got a little cheapy, solar light (2 led) cement animal from Menards. It has never worked. It is a seasonal thing so Menards have no more. I thought I would see if I could fix it. I replaced the 1.2 volt watch type battery. Nope, no help there. The solar panel puts out a uniform sign wave at 50 mV p-p. Is that too low? Is the solar panel bad? This was checked with an oscilloscope. With the new battery in place, I measure 1.2 volts dc at the leds with the switch turned on. 0 volts with the switch turned off. I tried a digital Fluke meter and a scope. I see no sign of any kind of AC waveform present at the leds. Is the QZ5252F chip bad? I know that I am knuckle head to spend time on a $3.00 cement porcupine, but I am the curious kind. Thanks for any light you can shed on this.


Lowell, Indiana

Dave Kruschke (author)JohnO1632016-06-30

First thing, I took a 1 inch+ by 1 inch+ solar cell from one of those $1 Walmart garden lights and measured the unloaded voltage while holding the cell one foot from a 60 watt incandescent light bulb. My multimeter measured 1.5 volts, well above 50 mv. I believe that the 5252 IC requires at least 0.9 volts from the solar cell to charge the battery. If the voltage is less than 0.9 volts, the IC acts like it is dark outside and turns on the light connected to the charged battery. If the battery is not charged then there is no light. If you took a 1.5 volt alkaline battery, size AAA or larger and connected it directly to your small 1.2 volt nicad, it might charge up in less than an hour, enough to power the IC that will turn on the LED(s) for a while. Some of the solar cells have poor connections on the backs. The wires look "soldered" to the cell but I suspect that what looks like solder might be some kind of conductive glue that may not be reliable. Anyhow, go to Walmart and get one of their $1 solar garden lights to help nail down your problem. I bet you will make progress without even turning on an oscilloscope...

JohnO163 (author)Dave Kruschke2016-07-01

Thanks Dave!

Great minds think alike.. We had a old $1.00 solar pole light with a failed battery. I brought into work to examine it and than I read your email this morning. Not knowing any better, I used the o-scope on the direct output of the little panel and only looked at AC. I used the Fluke meter on every thing else. The 60 mV p-p is normal when comparing the two little solar panels. Had I checked the DC right at the panel, I would have found 1.5 to 2 volts (depending how close my light source was to the panel). Both panels checked the same. I am not familiar with solar panels (as you can tell) so this was a fun learning experience. The four pin IC chips used on the solar products had different numbers. I could not find the pin layout for the HW012. Both solar units had different style 1.2 volt batteries. I threw dice and installed the chip with the different numbers. One of the posts on your site said to be careful on chip replacing because the writing on these Chinese chips could be on either side; no marking for Pin #1. I followed your sites advise and just did a quick ohm job between the two different style boards. I checked to see what pins were connected to what part of the circuit. Sure enough the writing was on the opposite side of the HW012 as compared to the QX5252F. Its works like a champ now! Thank you so much for sharing your technical expertise without hesitation. Have a Safe and Happy Fourth!

John with the lit up porcupine....

jburroughs1 (author)2016-03-31

Dave, I've been using the "nocturnal solar lamp" instructable and when I first built them they worked flawlessly. I was just trying to prototype before I built them into a container and they charged up during the day and lasted all night till the morning came and they turned off. Now they only come on for shorter and shorter periods of time. The battery shows 1.4 volts and the Solar Panels are 3 volts an 70ma. I have a 330uh inductor and a 0.5W straw hat. I can't work out why this is happening. Any ideas?

I seem to remember that an inductor that is between 100 uh and 200 uh works "better," than 330 uh, that is, it results in more current. But if your prototype worked flawslessly for a while, I'd say something has CHANGED and it probably wouldn't be your 330 uh inductor,.Is the solar cell getting the same amount of light? Is the solar cell free of any coatings, dirt, etc? Is the battery "cycled out?" You might have to get identical parts, create another successful prototype and then swap out parts. We have a cheap solar light that comes on every night but I don't think that it stays on all night. After the first year and a half, the battery was replaced because it was "cycled out" - recharged too many times. The new battery made the light work just fine and it continues to work. I also seem to remember that a "cycled out" battery would charge up quicker than a good battery but then it would discharge quicker as well. This is very true with the rechargeable battery in our camera, for instance. Please let us know what you find out...

There are two observations. All the parts are new. The inductor is not the "seafoam green" but brown and I can't seem to find what those differences if any are. I have some smaller inductors coming. The first day or two no doubt were sunny and they certainly got a good charge but if I measure the batt voltage I get more than a volt(AA). One I had on a bread board and the other I soldered up. Neither will stay lit for more than a few minutes now. In fact to get the one that is solders up I have to flash the light in the dark room on to trigger it. Any thighs would be appreciated. I'll put together some others and see what happens.

While you are waiting for inductors, you can consider the following: When measuring battery voltage it might be more realistic to put a resister across the meter leads to simulate current draw by the LED. I've seen white LED current values from 10 to 25 ma, which, by ohm law might suggest an equivalent resistance of 300 to 120 ohms. For sure you could put a 1000 ohm resistor across the battery when measuring it's voltage. Another thing you could do is charge up your battery using a D battery with some kind of resistance between the D battery and your rechargeable battery. If you succeed in charging the battery, you could put that in your lighting circuit and see how long it runs...


I found the spec sheet for the 5252f says it switches off at .9v so that must be what is happening so I'm going to replace with a new akaline battery(its raining today) and see how long it lasts. I'll also build a new circuit with a switch from the battery so I can save the batt and turn off the light. I'm hoping this will be the fix! Thanks for your help and I'll let you know what happens this weekend

pierr2323 (author)jburroughs12016-04-27

Hello J, I read that you got your hand on the spec sheet of this 5252F. Would you mind letting us know how we can get this spec sheet? I tried looking up the WEB unsuccessful. Thanks, Pierre

Dave Kruschke (author)pierr23232016-04-27

Try searching: chinese 5252F solar ic as this led to several spec sheet options for me...

pierr2323 made it! (author)Dave Kruschke2016-05-17

Thank you for the tip. I found pretty much what I needed as far as spec sheet. I even was able to make a little EXCEL sheet comprising a compilation of different spec sheets Data and some schematic. I am glad to share it for anyone interested. This attachment is an EXCEL file.

pierr2323 (author)Dave Kruschke2016-05-17

Thank you for the tip. I found pretty much what I needed as far as spec sheet. I even was able to make a little EXCEL sheet comprising a compilation of different spec sheets Data and some schematic. I am glad to share it for anyone interested. This attachment is an EXCEL file.

Dave Kruschke (author)2016-05-03

Nah, I haven't really gone further than you. And the "design" you refer to isn't even mine but something I saw somewhere on the internet and simply tried out. But your post made me think more about why a diode is needed to make the capacitor/flashing LED combination work. Remove the "diode" and it doesn't work. Looking at the schematic, one sees that pin 2 of the IC grounds out the capacitor every cycle, IF THERE ISN'T A DIODE. This would prevent the capacitor from building up a charge large enough to operate the flashing LED. So some kind of diode is needed in this circuit. Perhaps using a Schottky diode would make the circuit more efficient but I don't really know...

DenisR1 (author)2016-05-01

Hi Dave many thanks it seems that you have gone further with your design I havn't tried your method with a red led instead of a Schottky diode but i'm sure others have found ways of making a very bright led from a aaa battery, I didn't think that I found something that hadn't been done before after all I'm quite new to electronics.

Kind Regards



DenisR1 made it! (author)2016-04-28

Denis Rev

Sorry I'm Not sure how to make a drawing yet but i'm learning, dpp is a drawplus program I will attach a pdf file that should open easily.


DenisR1 made it! (author)2016-04-27

Denis Rev

This is my first comment on a 5252 design by adding a schottky diode and a electrolytic capacitor it will supply enough power to trigger a three colour auto changing led I found this by experimenting with different components.

Dave Kruschke (author)DenisR12016-04-27

Sounds like you did some good work. However, I have no idea of what app to use to open a *.dpp file (I've never seen one of these)...

MarcelG6 (author)2015-10-30

pls note that the Paulmann solar cube contains this chip, nimh battery,solar panel and 4x led. i have a few of these in the garden, and on 2 locations, there is too much shadow, so decided to modify 2 lamp units for wired power supply. after opening them, i found out it uses the 5252F chip.

Dave Kruschke (author)MarcelG62015-10-30

Yes. So far, every solar "garden light" I've taken apart uses the 5252F chip OR IT'S EQUIVALENT. The bolded text is to emphasize that the pinout of this four legged IC can vary from that of the 5252F chip. Before removing this IC, it is helpful to trace the runs on the small circuit board and make a schematic that reveals the true function of each of the four pins of the IC...

GeorgiR1 (author)2015-09-15

Because LED is supplied with a pulsing voltage around 200KHz DVM (Digital Voltage Meter) will only show voltage close to a battery level voltage. Around 1.3-1.4v. Take electrolytic capacitor 10uF 20vdc ( or close to that ) and connect across LED. Positive (+) side of a capacitor to a positive side of LED. And now you can measure 5v to 7v (or whatever) on your LED.

brhodewalt (author)2013-07-25

Also, in your schematic in Step 5, what is the purpose of the connection to pin 2 on the IC?

Dave Kruschke (author)brhodewalt2013-07-26

Pin 2 is for the connection to the positive terminal of the 1.5 volt battery.
Pin 3 is for the connection to the negative terminal of the 1.5 volt battery.
Pin 2 and Pin 3 are basically for the battery power supply.
Pin 4 gets repeatedly grounded to negative, connecting the battery directly to coil L momentarily. When pin 4 ungrounds coil L, the magnetic field created collapses and and results in voltage in series with the battery to flash the LED...

Pin 1 monitors voltage from a solar panel or other battery charging source. No connection or a low DC voltage will enable the LED driver chip. An LDR (light dependent resistor, transistor or diode) from + battery to pin 1 will turn the chip off during the day.

Here are some ways to use the chip. The first schematics do not use pin 1 at all. The second one is for a power failure circuit.

Hi, I'm curious about this power failure circuit -- how will it protect the battery from over-charging? Based on the QX5252F datasheet, while it has over-discharge protection, it doesn't say how it would handle over-charging. I assume it doesn't (though I hope I'm wrong). And since this circuit will always be connected to the AC mains, how would it affect the battery in the long run?

I don't really know for sure about overcharging the battery but I can relate my experience with a cheap solar light over more than the last three years. This light ran over a year when it was first deployed. But then it sort of quit. I took it apart and replaced the small Nicad battery with an unused battery from an identical solar light and now It has been running for a couple of years. So what is happening? This light gets indirect sunlight all day long and it gets direct sunlight for at least 1/2 day. After sundown, the light stays on for at least several hours. In other words, the solar light continues to operate. Maybe the ic protects the battery from overcharging or maybe the solar cell output isn't strong enough to overcharge the battery. Either way, I don't see any "overcharging" issue at this time...

Thanks, Dave. Yes, I was also thinking that the solar cell output wasn't strong enough to overcharge the batter. The reason I asked is I found BurgersBytes' power failure circuit quite interesting. It uses AC as an alternative to charging the battery via QX5252. It also opens up the possibility of charging it via USB (5vdc).

Still waiting for my QX5252F's and solar cells from Aliexpress :-(

Patrick JamesO (author)2015-08-17

Hi, will a 1/4 watt inductor be okay assuming I'll be using 330uH?

I think that 1/4 watt inductor would be more than enough. If a 3 volt LED draws 20 milliamps, that would be about 6/100ths of a watts.

But, ... you could just try the 1/4 watt inductor and see if you can feel any heat. You won't hurt anything. You could also try different inductors, say from 100uH to 1000uH. I seem to remember getting a brighter LED using a 100uH inductor.

Thanks, Dave. I ordered a bunch of assorted 1/4 watt inductors last week. I can't wait to try them out. Looking forward to more of your instructables.

harmhero (author)2012-07-20

I build a tiny Joule Thief, it works grea on one AA battery!

But what I don't understand is why the output voltage is the same as the input, 1.5 volts?

BurgersBytes (author)harmhero2014-05-25

The voltage is doubled by oscillation and cutoff. The extra voltage cannot be read on a DC meter, but a scope will show it as AC. 3 volt LED's can be run. Be careful with yellow and red colors.

And big Ted says always remember to plug your 'scope in first ?

Dave Kruschke (author)harmhero2014-05-27

Try measuring the voltage with your multimeter set to AC instead of DC...

Dave Kruschke (author)harmhero2013-05-20

Ha! I just happened to think: What did you use to measure the voltage when you measured 1.5 volts? The output of the Joule Thief is not a steady DC voltage but a voltage that fluctuates rapidly. Some meters only measure the average value of a fluctuating voltage so while the average reading might be 1.5 volts but the peak voltage (that is high enough to light the LED) is maybe 2.0 volts or higher. However, this momentary peak voltage might not be seen on the meter display. I suspect that using an oscilloscope would reveal these peak voltages I'm referring to here.
Dave Kruschke

Dave Kruschke (author)harmhero2012-07-21

Could you describe your circuit or better yet, provide a schematic?

On the other hand, without further info, I wonder what your output voltage would be when your battery voltage depletes to 1.0 volts. Sooner or later, the battery voltage will drop but if you are still getting 1.5 volts, something is working right to at least some degree...

peterbryenton (author)2015-08-12

Pleasing to see some analytical and critical thinking combined with a bit of deconstruction to build understanding, thank you. I'll have a go at building this myself now.

Logical steps for explaining and well illustrated too.

Jan May (author)2015-04-21

I love this instructable. I am very much interested in building my own solar garden lights with a 5252F IC,
4v x 40mA solar cells, and "fairy light" strings with about 6 small
LED's in them.

I still have a couple of questions about the 5252F / XXXX chips though.

First of all, does anyone know the differences, if any, between all the variety of XXXX numbers for the 5252F chips, like 1101, 1106, 1120, 1201, 1305, 1306, 2105, etc? I see all kinds available on Ali Express.

Also, does anybody know if they provide any sort of charge controller logic, so they won't overcharge a NiMH AA battery? I am concerned I might have purchased overspec'd solar cells and theoretically could overcharge them.

The possibility of using a temperature sensor and an ATtiny85 to implement my own dT/dt monitoring sounds kind of fun (I could blink the lights in time with the temperature-dependent cricket sounds too), but I'm pretty new to electronics and the complexity of all of that might keep me from ever finishing (or even starting) the project.

DBoulanger (author)Jan May2015-05-02

Jan, as per the QX5252 datasheet I found on the net, the XXXX number represents the "Lot Number" and the "Year". Quite a few manufacturers are using the "Lot Number" as a "Week Number", therefore based on the number written on the IC, you can determine when the chip was manufactured. For example, 2105 would indicate week #21 of year 2005.

Personally I don't think there is any charge controller logic, so some experiments should be done to play safe. The only thing I can confirm is the fact that there is an "Integrated Schottky Diode" between pins 1 and 2 of the TO-94 package, whereas it would be located between pins 6 and 7 on the DIP8 package. So taking in account the diode's voltage drop and the voltage of the solar cell, I doubt the solar cell would go over the battery's voltage specification.

Also, for NiMH batteries, the charging time is roughly 15 hours at a C/10 rate, therefore if the battery has a 1000maH capacity, it can easily cope with a 100ma charging current. Choosing an inductor that would allow a safe charging current, will make sure that everything operates smoothly.

Using the temperature sensor approach with a uC, is actually fairly complex. I'd prefer the easiest approach, which is making sure just enough current is provided for the charging process. Also, the overall idea is to keep a minimal component count, therefore adding a uC complicate the things since you have to provide power to that IC too, which wouldn't leave you much for the led.

This circuit is based on a QX5252F and I'm actually experimenting with a YX8019, which comes from a different manufacturer but uses a similar concept. Some specs aren't as clear as I would like them to be, but that's OK, if some magic smoke starts to float around, I'll know something didn't go as planned. Just kidding !

Have fun.

Dave Kruschke (author)2015-04-26

First, Jan May, I don't know of any differences in the 5252 ic. But other ics similar to the 5252 can have different pinouts.

I'm not aware of any "charge controller logic" - perhaps the solar cell in a typical garden light isn't strong enough to damage a rechargeable battery. Our "garden lights" have lasted over a year before I changed out just one battery that didn't seem to recharge anymore. Let us know what you find out if you set something up to test this further.

I can understand your talk of harvesting different temperatures and applying the results to some kind of cricket like audio output. But I don't understand what you intend to do with dT/dt, the RATE of temperature change.

In any case, I think you will find a lot of this stuff easier than it sounds after you actually try out some hardware...

JRV31 (author)2015-03-26

Very good instructable, I built it.

I liked the inclusion of the simple circuit in step one.

Thank you.

About This Instructable




More by Dave Kruschke:Picaxe Tea BrewingPicAxe LED dark detector"Joule Thief" circuit in a Flashlight Bulb? Yes!!!
Add instructable to: