A high-power LED torch using a single AA-battery

 by qs
This high-efficiency design can power LEDs at 100mA of current and is not much bigger than the battery that powers it. The unique circuit uses 2 commonly available transistors (BC327 and BC337) in an oscillator to boost the 1.2 or 1.5 volts from a rechargeable (nickel) or alkaline battery to the 3+ volts necessary to light a LED. No fancy ferrite inductor is needed as we will be using the metal jacket around the battery as part of our coil.
 
Remove these adsRemove these ads by Signing Up

Step 1: The circuit and layout.

The circuit uses a BC327 (PNP) and a BC337 (NPN) transistor. They are connected as a blocking oscillator at about 100kHz through the 150pF capacitor. Each cycle starts with the coil being charged, then this charge, plus the voltage of the battery, about 3.5v, is applied across the LED. The 1N4148 diode takes this voltage and stores it in the .1uF cap, which is used to drive the transistors, resulting in a much more powerful and efficient circuit.

The 22uH coil consists of 35 turns of #28AWG wire over the battery, using the metal in the shell to multiply the inductance.

A simple diagram shows the placement of the parts.
1-40 of 93Next »
Havoc481 says: Jun 3, 2011. 8:15 PM
hay quick question i don't care if you think its stupid but what would happen if you where to apply 3.5 to 4v to this circuit what would the out put voltage be.
qs (author) in reply to Havoc481Jun 4, 2011. 6:52 AM
It will be zero.

The additional drive at the base of the transistor will not allow it to turn off and it becomes a short circuit, quickly draining the battery until it is destroyed.
Shalendar says: Jan 31, 2011. 11:36 AM
I'm having trouble finding 25mA 5mm 120000mcd white LEDs on any reputable site.
qs (author) in reply to ShalendarFeb 7, 2011. 6:57 PM
Any 25mA LED will work. The mcd rating is just how bright the 'spot' is.
Shalendar in reply to qsFeb 7, 2011. 9:36 PM
the 25mA leds i have didnt work. My guess is that the voltage was wrong. The LEDs i have are 2v. I'm pretty sure that the LEDs you used are at least 3v. I'll have to see if I can get higher volt LEDs and try it again.
Rusdy says: Apr 10, 2009. 4:47 AM
I wonder how efficient this circuit is (i.e. volts and current input and output)
qs (author) in reply to RusdyApr 10, 2009. 8:07 PM
With a new alkaline at 1.6v, it's 78% efficient, but drops to about 65% with a 1.2v Nicad - it has to do with the Vce drop across the transistors at 200mA.
Rusdy in reply to qsApr 11, 2009. 7:51 PM
I was a wee bit pessimistic when I saw the circuit. But after testing it myself in PSPICE, I stand corrected. It is pretty damn efficient for such a simple blocking oscillator circuit (as they're not really efficient). Those components choice, do give efficient conversion. At 1.2 V, PSPICE simulation does give around 40-50% efficient, and it's consistent down to 1V. The transistor does play a big factor here though (as you commented), definitely need high gain in high frequency, your typical hobbyst transistor (such as BC547/557) won't give the same efficiency (and less brightness as well for the LED). I wish I saw your circuit first, before I ventured out to LM2623 from National Semiconductor. It definitely gives 70-80% efficiency down to 0.9V. Unfortunately, it only comes in a surface mount (for a very good reason, with upto 2MHz switching frequency), so it's pretty expensive to work with.
qs (author) in reply to RusdyApr 11, 2009. 9:01 PM
I've been experimenting with the flyback circuit since the first Joule Thieves went around the web. You can see my efforts here on my website.

Normally I use low Vce transistors like the 2SD965 which really helps in the efficiency, but, in designing this circuit here, I wanted one that can be duplicated by the average hobbyist without access to the fancier equipment and components.

I'm glad you like it.
Sagar Gondaliya in reply to qsMay 16, 2009. 7:36 AM
your ideas are amazing. i was just wondering, i dont have a very good monitor so i cant read your schematics on the webite. could you maybe post some computer generated ones so i could read them? thnx very much in advance -a fellow maker
qs (author) in reply to Sagar GondaliyaMay 16, 2009. 10:16 AM
Tell me the ones you are having trouble with and I can try and send you better images.

Also, some of the circuits on my website link to larger ones too.
Sagar Gondaliya in reply to qsMay 16, 2009. 1:41 PM
im mainly having problems with the hack-lite one. thnx
qs (author) in reply to Sagar GondaliyaMay 16, 2009. 8:44 PM
Here's a bigger image. Or you can try this link which is accessible from the smaller pic on my website as well.

Let me know if this is helpful.

HakLite0000.jpg
Sagar Gondaliya in reply to qsMay 17, 2009. 6:26 AM
yea it is. thnx
zorba02008 says: Aug 11, 2010. 4:30 AM
thank you for reply
zorba02008 says: Aug 10, 2010. 3:11 AM
What will happen if we replace the 1.5 v battery by 3 v battery ? can i use a mobile battery instead of one AA battery (rated 3.6 650ma) ? How can inductance of coil affects the efficiency of this circuit (low or high)? thanks for your efforts
qs (author) in reply to zorba02008Aug 10, 2010. 8:16 AM
For Lithium (3.6v) operation, please see my response to Newzy.
zorba02008 says: Aug 4, 2010. 8:54 AM
i finished my one and add something i think it is very important (reflector) like in the images.......
Image.bmpImage2.bmpImage3.bmpImage4.bmpImage5.bmp
newzy says: Jul 10, 2010. 5:50 AM
@mchenson could you plz post all the details of the component that u used in your ckt..becoz u have not used c3 and value of c2 is much more.it will b helpfull.. @qs thanx for ckt..it is a nice 1..what is output voltage and current of this ckt??and if i use primary battery then 4 how many continuous hour will it last??can i add more led and 2 aa battery instead of 1??what modifications i have to made??? thanx a lot ..
qs (author) in reply to newzyJul 10, 2010. 8:39 AM
With Two 1-1/2volt alkaline batteries, the KISS design would be best: Direct connect the LEDs to the batteries. As long as they are White, Blue or Violet LEDs, their Vf is at or higher than the 3-volts supplied by the batteries. This ensures that the batteries are used 100% efficiently. It may not be cute or exotic, but it is the best solution.

If you are squeamish, a 22-ohm resistor in series with the LED will reduce the current (and efficiency) a bit to avoid overdriving the LED.

With 3000maH primary batteries, a single LED (@25mA) will run for (3000 / 25) = 120 hours.

qs
newzy in reply to qsJul 10, 2010. 8:41 PM
hii qs thanx 4 the reply..if u dont mind then can u explain me about theoretical calculation of this ckt??/like how to calculate voltage and current acroos o/p theoretically...it will be helpfull 4 me..thanx 4 ur cooperation..
joinaqd says: Dec 19, 2009. 4:17 PM
i have 2 questions:

1.) What colors are the band markings?

2.) If i have ferrite, how many turns would be needed to run this thing?
Sandivar says: Oct 25, 2009. 8:46 PM
Great circuit, and I look forward to building it just as soon as I can buy the required parts.

However, I do have a question:  You use a 22 uH inductor to test the circuit; why wouldn't you just use the same 22uH inductor wired into the circuit and skip wrapping the wire around the battery?

(And where can I buy some of those types of inductors, by the way?)
qs (author) in reply to SandivarOct 26, 2009. 9:53 AM
Any inductor of the correct value will work in this circuit, but inductors are not nearly as common or cheap as other electronic components - that is why I try to come up with home-made solutions for them.

One source I've used is eBay - there are a few Asian sources, like this one which has them at about 10c apiece.
carpeteknus says: Jul 29, 2009. 7:00 PM
Excelent ! lined up for my next weekend project :) Do you know how long will a new alkaline battery last? my rough math says around 20 hours for a 2000mA h(2000mAh/(4x25mA) ), but i'm not sure that's the complete discharge calculation
sign-up in reply to carpeteknusJul 29, 2009. 8:54 PM
Remember that the LEDs are run at 3.3v, while the battery is only 1.5v

Using the power calculation:
Input = 2500mAH * 1.5v = 3,750mWH
Output = 4 * 3.3v * 25ma = 330mW

So, to get operation hours: Input / Output = 11.4 hours. Of course there is also a penalty in the conversion (in-)efficiency we have to factor in, about 75% by my observations, which lowers the operation time to about 8-1/2 hr.
Biotele says: Jul 21, 2009. 1:23 PM
This is awesome ! You use the battery in place of the the ferrite. Does it have to be BC337/BC327, or could it be any type of NPN/PNP pair like 3904/3906?
qs (author) in reply to BioteleJul 21, 2009. 2:00 PM
To optimize performance, it really needs to have the components specified. However, that said, the real key is the NPN transistor. It has to have the power handling, gain-bandwidth and a low enough Vce for the kind of current draw we need. Ideally it should have been a 2SC965 or similar device designed for disposible flash-camera use, which shares most of the same requirements as our project, including a 1.5-volt power source.
Biotele in reply to qsJul 21, 2009. 2:54 PM
what is the function of D1? Is it to clamp the voltage at C2 to a certain level?
qs (author) in reply to BioteleJul 21, 2009. 6:01 PM
The diode rectifies the output voltage of about 3V to more than double the drive to the transistors, like a supercharger on a car engine. It also keeps C2 from discharging when Q2 (the BC337) conducts.
Davidl3 says: Jul 5, 2009. 8:09 AM
When you said "25mA 5mm 120000 mcd white LEDs," didn't you mean "25mA 5mm 12000 mcd white LEDs?"
imakethings says: Feb 24, 2009. 8:50 PM
what is the difference between PNP and NPN
tanmanknex in reply to imakethingsMay 15, 2009. 8:03 PM
a simple explanation is that they are built differently and are thus wired up slightly different.
qs (author) in reply to imakethingsFeb 24, 2009. 9:29 PM
It's probably too involved to get into here, but a good place to start is on Wiki, then follow some of their links to get more in depth information.
ralegg says: May 11, 2009. 7:03 PM
Could you list out which parts go to certain points in your electrical diagram? I am not very good with circuitry but I am trying, I have all the parts on your part list but your diagram has listed a 1N4148 diode, is this necessary b/c it's not on your list? Thanks
qs (author) in reply to raleggMay 14, 2009. 7:41 PM
Yes, the 1N4148 is required. I will add it to the list.
ML159 says: Apr 24, 2009. 5:26 PM
When I try to build the circuit the light just doesn't seem to light... I have to admit that I do not have every part specified and used different parts, is that the main source of my problem? I am using a bread board to test it and so far I have managed to get other "joule thief's" to work perfectly fine. I have been to you website and I have to say it is pretty amazing, kudos. I tried building those but i managed to get only one to work very poorly. Yet again I did not use the exact parts specified so that might be the problem. I am open to any suggestions as to why it doesn't work... thanks.
qs (author) in reply to ML159Apr 27, 2009. 1:22 PM
I think you have answered your own question. Keep in mind that we are trying to light up four LEDs at full brightness, so, yes, the components are extremely important, as you are finding out first hand. The BC327 and BC337 transistors are fairly easy to find, they are even available on eBay. Good luck, and keep at it!
ML159 in reply to qsApr 27, 2009. 4:11 PM
Thanks for your help! I think I am starting to get it to work.
dog812 says: Feb 27, 2009. 4:01 PM
Hey i was wondering.. Would a circuit like this work for my applications with my LED hula hoops? I would like to make them less expensive for people.. using a single AA or 2 AA in a drop in style holder would be awesome.. Please message me i would love to ask you some questions..
1-40 of 93Next »
Pro

Get More Out of Instructables

Already have an Account?

close

PDF Downloads
As a Pro member, you will gain access to download any Instructable in the PDF format. You also have the ability to customize your PDF download.

Upgrade to Pro today!