simple current source for LED and 3 cells?

Go to maxim-ic.com and find their white LED drivers. A bunch.

Go to maxim-ic.com and find their white LED drivers. A bunch.

In which case all you need is a simple switch to change current sources momentarily for checking.

You could use the LM134 (current source, 10mA) with a bipolar to get your 20mA. See the datasheet for an example.

- Cesar

Why do you want to do that? The slow decrease in brightness means you wont likely get stuck in a bat and bat guano filled cave with a dead flashlight, as I once did in search of the source of bird-spit soup in Sarawak.

Best regards, Spehro Pefhany

Just keep in mind that in addition to the need for 2 LM134s per LED (unless you meant something else by "a bipolar") the minimum operating voltage for the LM134 series-element is 1V. Make sure your LED voltages at 20mA and the nearly-deplated cells give you that extra volt.

As a former optics design consultant to a LED flashlight company, may I make two suggestions? Keep the factory resistor value unless your leds are showing signs of turning blue (assuming low cost white leds using phosphor) . Two throw away the alkaline cells and use good old carbon-zinc cells. We ran all sorts of tests, and alkaline cells did not add enough extra lifetime to bother with in low current draw applications.

We also ran two temperature rated tests, with rather creative marketing department names. One, the "Arizona Glovebox", and two, "Artic freeze". These were commercial flashlight bodies retrofitted with leds in standard lamp bases. The goal of the glovebox test was how long can a simple resistor based flashlight sit at 120'F in a car with periodic use and still start up. The answer, quite a long time, on the order of a year or more, as LED usually doesnt draw enough current to start the batteries outgassing (yes, many battery makers put in vents) and thus corroding the copper inside.

The results of Artic Freeze, with Nichia white and green leds, were quite interesting. The extra ~ two volts consumed by the classic 2 transistor current source really cuts into the overall maximized light output of a set of batteries, assuming you bought a led source for those applications where you need maximal battery lifetime, ie camping, caving,security,emergency kits and automobile glovebox. In deep freezing temperatures, a current source may cost you a working light, even if just a dim one. We ran some tests down to dry ice temperatures , even at the point of freezing the water in the batteries, with just resistors you still got enough light to get you out of a cave, especially with green leds. In fact the led does quite good in a cave situation with dry ice junction temperatures and very low currents of .5 to 2 mA . Claims of useful light @ 100 hours are really not unreasonable, and its much better if the light is turned off and the batteries given periodic recovery time.

So why would you want to trade off lifetime for constant I?

Steve Roberts

PS If I HAD to do it, i'd consider putting 1.5V 25 mA lamps in series with each led in place of the resistors, Radio Shack part number 272-1139.

Steve Roberts

Google for "led power supply",there is a small switching device,which puts very efficiently a wanted current into one or more serial LEDS. Has been designed for battery efficiency.

Build a current source, and use it to drive a mirror. That way, the compliance of the source won't be as much of a problem.

batt ---o--------------------------------o--------o--------. | | | | | | | | | --- --- --- | \\ /==> \\ /==> \\ /==>

| --- --- --- | .------. | | | | | | | | | '---| 317 |---[62]---o--o---. | | | | | | | | | | | | | | c | c c c '------' | b--o--b--------b--------b | | e e e e '--------------' | | | | | | | | batt ----------------------------o-------o--------o--------'

You can also hack up your own current source, but it won't be as temperature-friendly...

The 317 may have too much overhead for this application, so an adjustable LDO might be a better choice. Using a transistor array would make it more consistent; or, you can match beta by hand.

If the current overhead is too much, you can decrease the wasted current by increasing the 62 ohm resistor to 1k, and putting a 62 ohm resistor between the emitter and ground of the leftmost NPN. The problem with this is that the temperature coefficient of the current through the LEDs will suffer.

-- Regards, Bob Monsen

"we can allow satellites, planets, suns, universe, nay whole systems of universe[s,] to be governed by laws, but the smallest insect, we wish to be created at once by special act" -- Charles Darwin

Or use a resistor instead of the LM317. The thing will stop working at

Thomas

Or use one chip to do the lion's share of the work without regulator dropout issues:

Well, if you can drive the base of a resistor to around, say, 1.1V, then the emitter will be at around 0.5V, if it is loaded. So put, let's say, a

Connect the LED in between the battery's positive terminal and the collector.

So how do you drive the base to 1.1V? maybe use a reference, or a couple of diodes or whatever.

Just an off-the-cuff idea, but I think it would work OK. Here's a quick ascii-art schematic. Use courier or similar:

V+ | ____ \\ / LED \\/ ---- | / c Vb---b | NPN \\ e | | \\ / Re (depends on Vb) \\ / | | GND

Maybe use an LM385 + a resistor to generate Vb.

--Mac

:> Build a current source, and use it to drive a mirror. That way, the :> compliance of the source won't be as much of a problem.

: Or use a resistor instead of the LM317. The thing will stop working at : 3.5 volts anyway, from 4.8 or so volts with fresh batteries.

: 4.1/2.8 current ratio is not constant but might be perfectly acceptable : for a lighting task. That was in deed my first idea. But if I choose a resistor so that I get a maximum of 20mA/led @4.5V, I get less than 10mA @3.5V. The simplest solution is to use NiCd or NiMh instead of alcaline batteries. I do not use the lamps in security critical situations, so it wont harm if battery voltage rapidly falls down.

: Thomas

Just keep in mind that nimh batteries - while carrying nice mAh specs - are 1.2V cells, not 1.5V. If you increase to 4 cells, the discharge profile might be a lot flatter than the alkalines.

I really think that MAX1916 I pointed out would give you what you need with very little trouble.

I just posted (I think) a very simple circuit that I've used on 3-cell flashlights. It uses one npn, one MOSFET and a couple of resistors. Sorry, I lack the patience for ASCII art so I posed it on alt.binaries.schematics.electronic. (I think)

I used it for driving a 1-watt Luxeon at 350 mA. You could scale it to 20 mA or whatever by changing resistor values. You could probably replace the MOSFET with a garden-variety PNP transistor.

This circuit uses the base-emitter drop of an NPN as a reference so current changes a bit with temperature, but it works quite well in terms of constant current from 4.5 volts down to Vled + Vbe+ 0.2 or so.

Brainfart! A PNP won't work here and an NPN doesn't work well. It really likes a MOSFET. A small one in TO-92 would work fine. One good choice might be the Zetex ZVN4306, but whatever Radio Shack has would probably work too. Make the current sense resistor about 22 ohms for about 20 mA LED current. Make the other resistor 220K or so. The current stays constant within a mA or so until it's about ready to quit altogether.