How long bright flashing LED last on AA cells?

Well, the venerable LM3903 gets a couple of years out of an alkaline D cell, and IIRC 6 months out of an alkaline AA. Since your voltage is lower and your amp-hours are about half, and self-discharge is much higher in rechargables than primary cells, probably 3 months.

If you make your own flasher electronics, using inverter techniques so that the flasher will continue to operate until the battery is almost totally discharged :- up to several years.

A suitable circuit will work for many days, if not weeks, from cells normally considered "spent".

Key problem here is the words 'rechargeable NiMh'. The _self discharge_ of many rechargeable batteries, exceeds the current consumption that the circuit could be made to have (depends how fast you flash the light). Many batteries of this sort, especially the high power designs, have poor performance in this regard. This is why non rechargeable batteries are often preferred in units designed to run for a long time.

Best Wishes

AAA alkaline used in Pink Floyd`s Pulse could last a couple of years:

I'm surprised that so many people have an answer when you have not told us how much current this LED will draw in your application. The life of your battery depends on the average amount of current it must provide. While we can guess at the duty cycle for pulse operation, the LED current that is needed can vary by a factor of 10, 100 or even 1000 for different applications. We cannot use the term "super bright" to determine the operating current since is not an industry standard definition. One manufacturer's "super bright" LED may be rated for a very different current than "super bright" LEDs made by another company. In addition, many of the examples given for flashing LED systems that have very long battery life obviously run the LEDs at far less than rated current.

Just to give one example, if you used a Lumileds Luxeon Star, your 1000 mAh battery will last for less than 3 hours of continuous operation, less than 12 hours if the LED is operated at 50% duty cycle and less than 24 hours if the LED is operated at 25% duty cycle. If, on the other hand, you used a 20 mA LED then your 1000 mAh cell will last about 200 hours IF you run the LED at 20 mA when it is on and it is operated at a 25% duty cycle.

To get the long operating times mentioned in previous messages obviously requires that the LED current be far lower than 20 mA, which I would not call "super bright".

No. As you start by saying, operating time does depend on the average current (and self-discharge rates). A long operating time can be achieved with the LED current at 20mA or more when on, if the duty cycle is low enough. A very low duty cycle is pretty common for a flashing LED running from a battery supply.

Making a small inductor the load of a multivibrator oscillator, with the LED in parallel with it, can produce a train of "super bright" flashes even from a supply of

In message , Adam Aglionby writes

Looks like the humble LM3909 which is obsolete now. Quite an expensive assembly for an album novelty.

Ironically it's cheaper to use a microprocessor now. You set the watchdog to wake the processor up after a specific delay and the program consists of turning the LED on, running a short blink time delay, turning the LED off and going to sleep. The watchdog resets the processor and it starts again. You can set the LED current (resistor) on time and off time as desired.

Does require more than one battery though.

What duty cycle would you use?

Used to have one lying around AFAIR it was a LM3909, on a short pulse long break cycle.

Build quality was nice for a CD box, but novelty sleeves are much rarer nowadays with smaller artwork space, lensatic 3D Raven LP by the Stranglers was another nice one. Soft rock group made a shaped disc in shape of Africa the continent, problem was Cape Horn jammed on some turntables tone arm pivot.

Adam

There was a Logictech mouse sold some years ago with a flashing LED in the box. It used two very nice AA Ultra Duracells in conjuction with a little COB assembly-- a CMOS ASIC under the epoxy. Just three additional parts- the LED itself and two resistors to set frequency and LED current. Duty cycle fixed. Output was an NMOS open-drain with about 18 ohms Rds(on) typical.

BTW, despite the voltage doubling drive circuit, the LM3909 was not capable of driving many modern non-red LEDs from a single 1.5V cell because of the low output voltage.

For hobbyists anyway.

I don't know of any flash parts that have an on-chip voltage doubler, but there are plenty of mask-programmed micros and ASICs. I'd love to hear of any mask programmed parts..

Best regards, Spehro Pefhany

It depends on the application :)

For example:

For a torch with a super bright LED running off a single cell, I would probably have an adjustable duty cycle from 5 - 50%. That would give adjustable brightness (and run-time).

For a "walker on road setting", one 10 mSec flash, 3 times a second. Cyclically varying the repetition rate is more likely to get it seen.

For a "power on" indicator where power is limited, one 5mSec flash every

That's quite a large duty cycle range. 50% to 0.05% or a range of 1000. Which do you think the OP was referring to when he asked about his battery life?

If he had to ask the question posed, I shouldn't imagine that he knew either.

All that was clear was that he wanted a bright LED to flash and wanted to power it from, ideally one, but if necessary more, AA NiMH cells - and was concerned how long it would keep flashing. The logical inference was that he wanted it to flash for as long as possible.

He know knows that, if he wants, he can keep the LED flashing for months of a single AA NiMH cell...which I think rather answers his question. Perhaps. ;)

In message , =?UTF-8?B?UGFsaW5kcuKYu21l?= writes

I'll have a guess at the application.....

They probably wanted to add a flashing LED to a fake camera that could run for a long time on a battery.

In reality, a short duty cycle blinking LED on a camera immediately says FAKE! :)

If that's true, it won't deter a thief. Bear with me here, the logic gets interesting.. >:)

If it won't deter a thief, it's very important that cameras with flashing LED's on them actually contain lenses and working circuits to gather video evidence of the events that they might seem to encourage. Therefore it makes sense to always have something real in there, so in practise, you'd be wrong. :)

In the end it comes down to two types of criminal (occupying various points on a single line continuum actually). Smart, and dumb. The smart ones will think this logic through and know that there is no easy guide, a small stub sticking out the top might be a radio aerial that sends a keep-alive pulse that would set an alarm off if it was sabotaged to prevent transmission, so a lack of wires would deter a smart criminal, but not a very stupid one.

So, the LED makes sense, it will deter the smart ones, who will usually have easier ways to make a dishonest living, and there's not much you can do about the dumb ones, they're hard to predict. They're usually opportunists though, so if everything else looks like a deterrent too, they'll leave well alone. I've had an empty bell-box on a wall over my window for 20 years, and it works, I've even overheard discussions outside once, and the general tenor was one of nervous defiance, but the nervousness won out.

If it looks like a duck, how long do you wait to see if it quacks like a duck? The longer you wait, the more uncertain you get, and uncertainty undermines resolve, especially of criminals.

And that, ladies and gentlemen, ends my self-amusing rant for the day. >:)

Did you read the answers? Which one(s) did you find surprising in view of the fact that the op did not tell us the current requirement?

I saw only 1 answer that would have required knowing the op's current requirement to support the respondent's conclusion. So I wonder what I am missing?

Ed

Yes.

All of them - because the OP did not tell us the current.

The answers that didn't require knowing the OP's current requirements were probably correct for the assumptions made by the responder, but not specific to the OP's question.

Doesn't matter though, there are two ways to go, either demand exact specs from someone who might not have a clear idea yet, or post a response based on your own carefully judged assumption. The latter is usually easier that waiting and second-guessing your response before you've made one, and it gives the OP something to think about. If the result is confusion and an apparent grasping at each offered straw in turn, then it is probably wise to demand some kind of focussed assessment of the original situation. At least this way the OP is not considered stupid until proven intelligent. :)

My own thoughts on the topic are: you can have a 'super-bright' LED and long life, but a few things should be managed to do it. Use a lithium battery preferably, then alkaline, and avoid Nimh types. Use a power converter to efficiently make use of the entire battery charge. Use a narrow viewing angle on the LED, so that it projects strongly for only a limited point of view. Use a clear LED not a coloured plastic. Use a short duty cycle, less than 1% and pulse length no longer than 50 milliseconds even when the flash period is longer than 5 seconds, with that period never being longer than 10 seconds unless you're really going to be staring at that LED for 30 seconds or more regardless of what it might be doing. These are not hard and fast rules. :)

Lastly, consider a low power laser diode, UNLENSED. These are eye safe at distances beyond a few inches, (take care NOT to use a microlensed version though) as they diverge strongly. Their output is highly monochromatic, specular, very eye-catching, and their conversion of electricity to light is second to none, so they can make ideal beacons. They're much harder to use though, easily damaged, and initial power requirements are higher. Also, while you can make an LED stand much higher currents than their nominal rating, often tenfold increase, with a laser diode, the pulse rating for a CW diode is usually less than twofold higher.

I have some ideas...

1. If this is to be used in the home, power the LED from a wall wart. And maybe power the LED through two steering diodes coming together - one from the wall wart and the other from a backup battery. For that matter, get the cheapest actual security camera you can find from a hobbyist/surplus catalog - even if it's incompatible with anything practical. Or further for that matter, find some little professional looking case and fit some cheap webcam into it and you have a fake security camera that may actually be made to work in some way as a real one! Put a red LED on the case...
2. I have an idea for the red LED - make it glow continuously like those on real cameras that have LEDs do. Get a GaAlAsP red LED, peak wavelength 660 nm, 3000 mcd (with a usual beam width of 15 degrees but sometimes claimed more), such as Radio Shack 276-307. Red LEDs of same/similar chemistry and efficiency are Agilent HLMP-8103 and HLMP-C124. Get some fine sandpaper and sand down the tip by about a millimeter, then restore the "bullet" shape but about a millimeter shorter and with the tip very slightly more blunt, then get some really fine sandpaper and get the LED good and evenly frosted. After that, you have a super high efficiency diffused wide angle red LED that works well and reliably at 1 milliamp, and may be bright enough at half a milliamp. It will glow with a color close enough to that of lower efficiency red LEDs.

If you really want to go all out, get an InGaN green LED (nominal wavelength usually 525 to 530 nm) of the common 5 mm "bullet" style and give it the sandpaper treatment described above. Then give it half a milliamp or maybe a quarter of a milliamp. Most Nichia 5 mm green ones and those trying to compete with them (ETG, and others) should at half a milliamp have a brightness that I consider "fully that of a usual green LED indicator light". Furthermore, the wavelength of these tends to shift a little inversely with current, and at half a milliamp the color is likely to be about that of LEDs with nominal wavelength in the 550's nm, maybe close to 560 - a yellowish shade of green likely to look enough like a usual 565 nm green LED indicator lamp to look like a usual indicator LED to most criminals.

A pack of (4) AA alkaline cells with a 6.8K resistor should power such a green LED reasonably well 24/7 for at least half a year.

If you want an LED to falsely indicate presence of a security system in an automobile, don't worry about conserving every milliamp since the battery has self-discharge in the 10's of milliamps. You can afford a few milliamps to protect a car.

- Don Klipstein ( snipped-for-privacy@misty.com)