Introduction: 10 Year LED Flasher

This LED flasher will blink continuously for 10 YEARS on a single 1.5v AA alkaline cell. This is possible by achieving an average current draw of 35uA (microamperes) .

Step 1: This Is the Complete Circuit Diagram of the Flasher

Picture of This Is the Complete Circuit Diagram of the Flasher

Main components:

IC1: CD4001 CMOS quad NOR gate

Q1: 2N4401 NPN tansistor

T1: 1/2" toroidal bifilar transformer ( 2x25T)

Bat:: 1.5v AA alkaline cell

Led1: high efficiency led

Step 2:

Picture of

IC1 CD4001 A&B are wired as astable multivibrator with frequency=1Hz . This is the flash rate.

Step 3:

Picture of

IC1 CD4001 C&D are wired as monostable multivibrator with pulse width = 1mS (1 millisecond) . This is the time the LED will be ON during each second of time. IC1D biases transistor Q1 through R3 in a modified "Joule Thief"(blocking oscillator) circuit. With this configuration, IC1-D "gates" the oscillator on & off.

Step 4:

Picture of

When you connect the 1.5v cell for the first time, IC1 will not operate because it needs a minimum 3V to run. At this moment Vdd is 1.5v minus 0.3v (voltage drop of D1). By momentarily connecting jumper J1, transistor Q1 will be biased through R3 and will begin to oscillate. High voltage (flyback voltage) spikes at collector of Q1 will light LED1 , and will charge capacitor C4 through diode D1. Voltage at C4 (>3v) will feed IC1 which will start running and will run until battery is exhausted ( 10 years from now ! )

Led "clamps" voltage spikes to its forward voltage V(f) . If you use a RED led, add a 1N4148 diode in series, or the voltage at C4 will not be sufficient to feed IC1 CD4001. ( red leds have a lower V(f) ).

Step 5:

Picture of

Current drain of the circuit with the LED "on" is 35mA. Since the duty cycle is 1/1000 , average drain is 35uA (microamperes) . Typical AA alkaline cell has a capacity of 3000mA/H at low current drain. This is equivalent to (3000*1000)/35 , or 85,714 hours, or 3,571 days, or 9.8 years! For more details, watch the video.

Step 6: Full Video Tutorial


ham-bob (author)2017-01-14

PLease help - when I test the right half as a jewel thief - it works fine. When I test the left half with 3 volts It also works ok. But when I combine the 2 halves i do not get enough voltage to run the ic. My toriod is a 1/2" id one with 29 tuns. Please help - what can I do next? Ive tried 3 torrids and 2 leds.

Thanks much.

simpletronic (author)ham-bob2017-01-15

Are you using a red LED? Red leds have a forward voltage drop of aruond 1.6V, not enough for the CD4001. I have posted an update to this circuit

with a PCboard download. This new circuit has no LED f(v) limitations. Greetings!

ham-bob (author)simpletronic2017-01-20

An update! On a proto board i now have this working!

There are 3 of us old-far** working on this. What would

make one have 5 volt on the cap and another 3.5 volt ??

simpletronic made it! (author)ham-bob2017-01-31

Hi! Answering your first question, when you measure voltage with a VOM (specially analog VOM) a small current flows through it. This small current slowly discharges the cap. As for the second, bear in mind that you are not measuring a fixed voltage but rather a sawtooth wave (see simulation image) with a rapidly varying voltage. A VOM (specially digital) doen not update display fast enough to reflect actual voltages.

ham-bob (author)simpletronic2017-01-16

no - I am using a white LED.

I now have tried 3 different torids and can not get enough voltage to run the IC I have a electrolytic cap - will a tantalum be better?

I am using the revised or updated version of the unit.

simpletronic (author)ham-bob2017-01-17

Standard electrolytic has higher leakage than tantalum but should work nevertheless. Value of cap should not be larger than 10uF, or will never reach the required voltage. Ferrite cores are conductive... make sure windings are not shorting through it. Does led light when you short the header pins?

ham-bob (author)simpletronic2017-01-17

Yes it does - when I put a meter across the cap the voltage drops rapidly

What causes this?

Kiểm toánP (author)2016-12-11

what value of L1 and L2 (henry)?


Kiểm toánP (author)2016-10-28

Thanks for sharing :)

lyoppi (author)2016-10-05

Please upload circuit design too along with wiring daigram

simpletronic (author)lyoppi2016-10-13

I have published a circuit update and a PCBoard download.

and a youtube video:


michael.richey.566 (author)2015-10-28

I'm a big blinky light fan - and this seemed like a really cool project that would give me and my son something to do.... So I ordered all of the parts and sat down to build it and I can't get it to work. The 2n4401 transistor gets hot and the light never blinks.

So, I loaded up the "tutorial" video and after watching through it half a dozen times, I took screenshots of the overhead view and started comparing it to the schematic.

They don't match.

1. There's an unaccounted for jumper between 3 and 5 (where a 0.1uF cap should be)
2. That same 0.1uF cap is between 2 and 4
3. The 4.7M resistor that's supposed to be between 1,2,4 and 5,6 is actually between 1 and 3.
4. the jumper that's supposed to be between 3 and 8 - is between 4 and 8.
5. the 15K (brown-green-orange) resistor between 12,13 and 14 in the schematic isn't brown-green-orange, and it's in series with another resistor.

I don't mean to be a pill, but I'd kinda like to see this thing work outside of a video.

simpletronic made it! (author)michael.richey.5662015-10-28

Hi Michael!

This circuit works fine so don´t give up yet! Let´s troubleshoot:

If transistor is getting hot, that means there is a large current flowing through it. Possible causes:

(1) the pins (EBC) are not wired correctly - check with datasheet.

(2) jumperJ1 (read description) is for start-up only. It should be removed immediately after. When jumper is connected, transistor and transformer work as a regular Joule Thief, and led should light immediately unless transformer polarity is wrong. Try reversing the 2 wires of one of the windings. (only 1 winding). This would also cause heating. Transistor could also be faulty. check it or replace. Check transformer for short-circuits (ferrite core is conductive).

The 2 resistors you see in prototype are a 10k in series with a 4k7 (which sum 15k as in schematic.)

IC1 is a CD4001(quad nor gate). All 4 gates are identical and can be interchanged with identical results. That´s what I did in my prototype. If yo stick to the pin numbers in the shematic, it should work perfectly.

Hope these tips help! Greetings!

After really painting up the toroid - I'm getting the suggested result, button press = led on, but I'm still not flashing. When I press the button, I'm only getting 2.5v to pin 14 - so I suspect that my toroid isn't adequate to this task. I was able to find 1/2 inch OD x 1/2 inch width - so my coils are more oval than round and I'm only able to get 17 or so turns. I've ordered some that are more like the ones you used.

Is there a test I can perform to determine if my winding is adequate or is it more of a trial and error kind of thing?


2.5V on pin14 is a bit low but may not be the problem. Are you using the right diode? ( 1N5819 schottky has a very low forward voltage drop) . Led "clips" the voltage spikes (50-60v) to a voltage slightly above its forward voltage somewhat as a zener diode would. Red leds and some green have a low Vf (forward voltage drop) in which case you must add a diode or two (1N4148) in series with the led (0.7v). I made my transformer with a core salvaged from an old PC motherboard. Folded 1 meter of enameled wire in half and wound the 2 coils at once. cutting the end separates the 2 windings. Nothing special ! . Recheck the wiring of the CD4001 gate section.

Could you post or PM me a good close-up of your build ? Greetings!

I'm trying to make this work with a single 3mm ultraviolet LED, which according to the supplier has a Vf of 3-3.6v.

I just noticed that while I'm pressing the button - there is a slight pulse in the LED.

simpletronic made it! (author)michael.richey.5662015-10-31

Michael, at first sight I see a missing connection. This is the path of the 1ms pulse from CD4001 out to the led driver. No wonder it didn´t flash! .

UV leds emit little visible light so stick to green-white-blue (highest efficiency)

Ican´t clearly see the wiring of the gates, but start by adding that wire . Good luck!!

The gate wiring is as follows:
1-2, 1-4, 3-8, 5-6, 9-11, 12-13 - jumpered
.01uF cap between 3-5, .01uF cap between 10-12
4.7M resistor between 1 and 5
15k resistor between the 5819 diode output and 12
14 to the 5819 diode output
7 to 1.5v+

I added the missing wire from 11 to the trigger and I've got the same situation - I press the button and the led turns on - but no flash.

I replaced the UV with a red LED - but it's the same

typo - those are 0.1uF caps between 3-5 and 10-12

pin 7 to 1.5v+ ?. should go to 1.5v- (0v bus) but it looks OK in image. Don´t use a red led (too low Vf).

Do the following test:

with the 1.5v cell connected, connect a higher voltage battery (e.g,6v) directly to CD4001 power pins ( - to pin 7 and + to pin 14 ) . Do not connect jumper J1 during this test - do not exceed the maximum operating voltage for C4.

If led doesn´t flash, there might be a fault in the CD4001.

OK, I grabbed an 18650 3.7v battery and tried what you asked and nothing flashed. First I disconnected 7 and 14 from the 1.5v rails, added a pair of leads and connected the 18650. The LED didn't even turn on. CD4001 operating voltage starts at 3v, so this should be fine.

So, I replaced the CD4001 and got the same thing...

I connected my multimeter to the -3.7v and the pin 11 output and I get a pretty consistent 1.6-1.8v - I'm not seeing a 1Hz signal, it seems to be just on

simpletronic made it! (author)michael.richey.5662015-11-03

Leave the 4001 connected to the 1.5v rails during above test, otherwise you are isolating it from the rest of the circuit. (1.5v cell connected also)

Digital multimeters have a very slow display update and will not catch a 1/1000 second pulse. Try with a led in series with a resistor (e.g.4k7). You should see a 1Hz square wave (50/50) blink at pin 8 or 3, and a very short-faint blink at pin 11.

Yonatan24 (author)2015-08-07

I don't know if you know this but batteries have internal resistance so the battery won't be able to survive ten years even when there is no load.

I didn't understand the part where it the battery charges itself? Perpetuum Mobile?

Duracell seems to believe their batteries can survive 10 years. They even made a guarantee saying as much.

powerpc (author)2015-05-21

Very very very cool!!!!

simpletronic (author)powerpc2015-06-18

Great !!

Testas.T (author)2015-05-17

I'm very interested how many current it draws, because in the begining you said that it consumes 35uA and at step 5 you said it consumes 35mA

simpletronic (author)Testas.T2015-05-18

35mA is the current draw when the led is ON, which is during 1mS(one millisecond) of every 1000 milliseconds flash cycle. The time the led is off (999mS of every flash cycle) the battery current is 0, and the circuit is running only from the charge of C4, Therefore the average current from battery is 35mA/1000 or 35uA. Greetings!.

kasbah (author)2015-05-16

Have you got a part number for the toroidal transformer you used?

simpletronic (author)kasbah2015-05-16

No, it's just a 1/2"frerrite core I salvaged from an old motherboard. Just wind 1meter of enameled wire folded in half , and then separate the windings by cutting end. Most cores are electrically conductive, so It'a good idea to paint it with enamel, or clear nail polish. Tutorial for transformer here:

nqtronix (author)2015-05-14

I'm impressed.

At it smelled like bullshit, but wow, you have thought about evverything!

Yes, I did some research and this are the numbers I came up with:

The quiscent (standby) current of the CD4001 is roughly 0.25uA per gate, resumting in 1 uA total. The leakage current of a tantalum is 0.5 to 1 uA (I picked a few random parts on digikey and compared the datasheets), make sure you don't get a tantalum polymer type, their leakage current is about 10 times a big! You could also get a ceramic capacitor, they have almost no leakage at all. The value 10uF is perfectly selected, it allows a 2uA drain (1uA leakage + 1 uA due to the CD4001) for 1.4s, longer than a full refresh cycle.

Maybe the most critical part is the 1N5819 as it's reverse current can be up to 35uA @2V & @25°C (source: page 5), but I'm sure you've done the homework on that as well.

If you came up with this circuit, congratulations, I haven't seen such a elegant, well engeneered in the last month on instructables. I can't wait to see with what you'll come up next.

simpletronic (author)nqtronix2015-05-14

Thanks for your comment! It's really rewarding. The idea of using ceramic for C4 is really good. All the leakage currents you point are correct, but needn't be taken into account: If you look closely at circuit diagram, you will see that CD4001 is powered by capacitor C4 and NOT directly by the battery. The same applies for the diode reverse leakage... in fact we could even say that capacitor "recharges" battery during the led OFF phase through diode's reverse leakage (voltage on cap is higher than the battery's) . Battery is ONLY used during the Led ON phase (1mS) which is when C4 is recharged, and peak current is 35mA. From there on, the rest of the cycle the whole circuit operates exclusively from the C4 charge which will be refreshed on next cycle. All leakages, led current and energy dissipation come from that 35mA peak current during the 1 millisecond Led ON phase.

nqtronix (author)simpletronic2015-05-15

Sure, the LED can only be on for 1ms, this is the key of the crazy battery life.

I didn't thought of the battery beeing "recharged" with the diode leakage current, I though I was missing something :)

The capacitor leakage however is still relavant, because the current is "bypassing" the battery: The faster the capacitor discharges itself during the "off-period", the more current (and thus power) it requires to be recharged in the "on-period". It does not matter which load you connect to it, everything from a simple resistor to it's own leakage counts.

simpletronic (author)nqtronix2015-05-15

Your reasoning is totally correct, if we use a ceramic or polyester for C4 to avoid leakage (as you suggested) we will be "treasuring" a few microJoules and that can translate into more battery life by increasing R3 (maybe 1 more year?), or brighter apparent led flash by increasing pulse width (increase R2), or higher flash rate (reducing R1). Greetings ! .

samuel123abc (author)2015-05-14

This is so cool! Good job!

Thank you Samuel !

vinz3nt (author)2015-05-13

this reminds me of the first Logitech optical mouses, when these came out it shipped in a box with a flashing led, I kept it and it flashed for years!

raptor_demon (author)vinz3nt2015-05-13

or the original pink Floyd pulse CD case. that thing blinked forever

vinz3nt (author)raptor_demon2015-05-13

Now that you mention that! A friend of mine used to have it, forgot about that one!
I have to build one now and put it somewhere, I'll post back in 10 years.

DIY-Guy (author)2015-05-13

This looks like a candidate for a multi-LED unit run off of a D battery.
Beautiful work!

np_grum88 (author)2015-05-13

You should try to make it more compact for a led throwie

About This Instructable




Bio: Music: my profession for over 40 years... Electronics: my beloved hobby always.
More by simpletronic:Tic Tac Toe ( 3 in a Row )USB Powered Traffic LightPhantom Power Blocker (protect Your Dynamic Microphones)
Add instructable to: