Drive 1W LEDs from batteries

I have a project where I'm using a 1W, 350mA LED and the entire project needs to be powered from AA batteries (alkaline or NiMH). Is there a good way to drive the LED?

It's working right now using a transistor to switch the power and resistor to limit current to the LED, but burning half a watt on a battery powered project is pretty bad.

I've been looking at the little power supply boards online but they all seem to want 12V in and power 1-3 LEDs in series. I'd rather use as few batteries as possible. Also, ideally I'd like to switch the LED as fast as possible (clean square wave in the 10's of kHz range), which the transistor solution lets me do.

Thanks

Look into constant current drivers vs straight voltage with current limit resistor.

You can use an LM317 Adjustable Voltage Regulator as a constant current driver. Google: LM317 constant current

cheapest solution would be a 555 chip as PWM ..... that way a small potentiometer could be used to vary the brightness.

But that wouldn't be constant current, would it?

CrossRoads:
Look into constant current drivers vs straight voltage with current limit resistor.

That's what I'm looking for but all the modules I can find are either 12V or 85-240V inputs. What I want must exist judging by the number of cheap flashlights that use a single cell to power a 1W LED. I've spent a while with google, deal extreme, and aliexpress, but if you can point me in the right direction it would be helpful.

johnwasser:
You can use an LM317 Adjustable Voltage Regulator as a constant current driver. Google: LM317 constant current

I understand these require at least 3 volts higher in than out even as a constant currant driver and as linear regulators just dissipate the extra power, so they're not very good for battery powered circuits.

cjdelphi:
cheapest solution would be a 555 chip as PWM ..... that way a small potentiometer could be used to vary the brightness.

I'm not sure what you're suggesting. I don't need to vary the brightness, I'm looking for a power driver. A 555 can not supply 350mA (a quick search seems to show 200mA, 600mW), and it's not cheaper than using a PWM pin on the arduino that's already controlling the project.

If you're not changing the brightness, then why are you looking for this?

"ideally I'd like to switch the LED as fast as possible (clean square wave in the 10's of kHz range)"

I think that's adding to the confusion.
Digikey has many options for PMIC LED drivers.
http://www.digikey.com/product-search/en/integrated-circuits-ics/pmic-led-drivers/2556628?k=led%20driver
Filtering down for 1 output, constant current, in-stock, low voltage supply (1V to 3.2V) yields a part like this

http://cds.linear.com/docs/en/datasheet/3490fa.pdf

Oracle:
It's working right now using a transistor to switch the power and resistor to limit current to the LED, but burning half a watt on a battery powered project is pretty bad.

I've been looking at the little power supply boards online but they all seem to want 12V in and power 1-3 LEDs in series. I'd rather use as few batteries as possible. Also, ideally I'd like to switch the LED as fast as possible (clean square wave in the 10's of kHz range), which the transistor solution lets me do.

Thanks

Unless I'm mistaken, I think he asked for a PWM solution?... there are plenty of other ways including a switching regulator for a couple of bucks off of ebay (this is the most efficient way though for the price) but feel free to play with the 555 if you want a cheap solution to make your circuit more efficient.

" Also, ideally I'd like to switch the LED as fast as possible (clean square wave in the 10's of kHz range), which the transistor solution lets me do."

Not really sure what intended here - unless it was the switching speed of a boost converter:

"The LTC®3490 provides a constant current drive for 1WLED applications. It is a high efficiency boost converter
that operates from 1 or 2 NiMH or alkaline cells and generates 350mA of constant current with up to 4V of
compliance. It contains a 100m? NFET switch and a 130m? PFET synchronous rectifier. The fixed switching
frequency is internally set to 1.3MHz."

Oracle:
I have a project where I'm using a 1W, 350mA LED and the entire project needs to be powered from AA batteries (alkaline or NiMH). Is there a good way to drive the LED?

It's working right now using a transistor to switch the power and resistor to limit current to the LED, but burning half a watt on a battery powered project is pretty bad.

Lower the voltage to a point where you're not wasting much.

Oracle:
I've been looking at the little power supply boards online but they all seem to want 12V in and power 1-3 LEDs in series.

Adjustable booster: http://www.ebay.com/itm/261110225710

That booster has:
Input voltage: 3V-30V

OP indicated "I'd rather use as few batteries as possible." To me, that is 1.

There are boards that will do that if building up a circuit using a part like LTC3490 is not desired. Example:

In stock without waiting 2.5 to 4.5 weeks for China delivery.

Have to look at input current vs output voltage & current.
For 12V out & 680mA, 2 cells & this board would probably be a better example.

Altho from the graphs that much current doesn't seem possible.

The e-bay part has no graphs, can't tell what chip is used, and "(There is no User's Manual in the package.)"
So buyer beware, eh?

Reading the LTC3490 datasheet some more:
The LTC3490 is a high efficiency, constant current source
for 1W high intensity white LEDs. These high intensity
LEDs require a fixed current of 350mA with a voltage
compliance of 2.8V to 4V.

So this may not the right chip for a single-chip solution, maybe a few circuits in parallel are needed.
Some more research for parts at digikey may find a better solution.

CrossRoads:
That booster has:
Input voltage: 3V-30V

OP indicated "I'd rather use as few batteries as possible." To me, that is 1.

There are boards that will do that if building up a circuit using a part like LTC3490 is not desired. Example:
http://www.pololu.com/catalog/product/2564
In stock without waiting 2.5 to 4.5 weeks for China delivery.

eBay has stuff like that as well, eg. http://www.ebay.com/itm/370880414416

The reason I didn't post it is because it outputs 5V. 1W @ 350mA suggests the LED needs about 3V. If the stated goal is to save power you should run at less than 5V, which the adjustable regulator can do

(although the voltage won't be much less than 5V once you add in a transistor and resistor...)

CrossRoads:
The e-bay part has no graphs, can't tell what chip is used, and "(There is no User's Manual in the package.)"
So buyer beware, eh?

I've used loads of them, no problems.

PS: I get stuff from China in about 10 days, probably not much different than Pololu for international shipping. YMMV depending on where you live.

Thanks for your replies.

CrossRoads:
" Also, ideally I'd like to switch the LED as fast as possible (clean square wave in the 10's of kHz range), which the transistor solution lets me do."

Not really sure what intended here - unless it was the switching speed of a boost converter:

"The LTC®3490 provides a constant current drive for 1WLED applications. It is a high efficiency boost converter
that operates from 1 or 2 NiMH or alkaline cells and generates 350mA of constant current with up to 4V of
compliance. It contains a 100m? NFET switch and a 130m? PFET synchronous rectifier. The fixed switching
frequency is internally set to 1.3MHz."

the LTC3490 looks quite nice, its a bit expensive in small amounts and of course a pre-assembled "typical application" module would make things easier, but it's the best option so far. I think I'd have to use 2 cells though. With one, the LTC3490 efficiency drops to 50% and drawing 1.5 amps from an alkaline AA will drastically reduce its capacity. That's not a huge problem though, and as far as your later post, the LEDs I have in mind require 350mA in the 3-3.4V range, so it appears this is the right chip for what I'm doing.

I'm actually thinking of a few projects that involve switching the LED, the more complex involves carrying data. Imagine people carrying flashlights and I can identify them by a signal carried on the beam. Like an IR remote but visible with a high-power LED. The simpler project is a stroboscope, an external trigger turns the LED on for 50 to 100 uS.

I'm not familiar enough with these power supplies, can I just use a transistor to switch the output and when I want to turn the LED on the power is there and stable fast enough?

cjdelphi:
Unless I'm mistaken, I think he asked for a PWM solution?... there are plenty of other ways including a switching regulator for a couple of bucks off of ebay (this is the most efficient way though for the price) but feel free to play with the 555 if you want a cheap solution to make your circuit more efficient.

I still don't understand how PWM helps me get a constant current to the LED.

fungus:
Lower the voltage to a point where you're not wasting much.

...

Adjustable booster: http://www.ebay.com/itm/261110225710

You mean use the power supply you linked and a series resistor so the brightness is more stable than just directly from batteries with a resistor?

I'll show you PWM in action, it's a constant 5v, what you vary is the overall current the 555 will chop up the load giving the LED a break and you use less current, 555's not going to be that efficient.

I was testing this cheapy oscilloscope i bought off of ebay so i tried it with PWM (which can be done with a 555 or an Arduino)

I was watching a video at adafruit discussing a 4-AA battery pack for charging an iphone. Bottom line - you can only get 500mA from AA battery.
Maybe if you use C cell or D cell - you're likely to have plenty of room behind your LED array. Maybe one of those 6V lantern batteries even.

cjdelphi:
I'll show you PWM in action, it's a constant 5v, what you vary is the overall current the 555 will chop up the load giving the LED a break and you use less current, 555's not going to be that efficient.

I was testing this cheapy oscilloscope i bought off of ebay so i tried it with PWM (which can be done with a 555 or an Arduino)

my oscilliscope :) - YouTube

That looks like a normal 5mm LED. How will that circuit help me drive a 350mA LED? Also, I don't need to dim the LED.

I was watching a video at adafruit discussing a 4-AA battery pack for charging an iphone. Bottom line - you can only get 500mA from AA battery.

I suppose if I use the boost regulator off 2-AA's that won't be a problem as long as efficiency stays above 70%. 3V 500mA is 1.5 Watts. But I'm okay with 2 batteries, so not a problem.

CrossRoads:
I was watching a video at adafruit discussing a 4-AA battery pack for charging an iphone. Bottom line - you can only get 500mA from AA battery.

That would be an Alkaline battery.

Rechargeables can do a lot more than 500mA.

Oracle:
I've spent a while with google, deal extreme, and aliexpress, but if you can point me in the right direction it would be helpful.

The prebuilt step-up LED drivers on Deal Extreme won't work? Have you tried FastTech or intl-outdoor? I know they have a few 1 or 2 AA 1W LED drivers which use the PAM2803 IC.

If you're in the US, I can give you a link a to US vendor that stocks them.

Mike_S:
The prebuilt step-up LED drivers on Deal Extreme won't work? Have you tried FastTech or intl-outdoor? I know they have a few 1 or 2 AA 1W LED drivers which use the PAM2803 IC.

If you're in the US, I can give you a link a to US vendor that stocks them.

Excellent question. Deal Extreme has a bunch of items like 6mm 5mW Red Laser Module (3.54.5V) - Free shipping - DealExtreme which may well be exactly what I want, but the description is so poor I can't tell. "Voltage: 1.5V" do I just assume they mean input voltage and it outputs a higher voltage? Even if so, what is the maximum output voltage, the maximum input current, will an Alkaline AA even be able to power it? Does it have to be 1.5 volts, can I use 2 cells to lower the current needed? It really is frustrating. Maybe I should just order something that looks promising.

I took a look at the datasheet, PAM2803 based modules could be good. Maybe that's what a lot of these modules use, they seem cheap enough. I'm located in Canada.

Oracle:
I took a look at the datasheet, PAM2803 based modules could be good. Maybe that's what a lot of these modules use, they seem cheap enough. I'm located in Canada.

I'd use two batteries (or more).

The bigger the difference between input/output voltages, the less efficient those boards are and the more amps you have to suck out of the battery to make it work.

eg. If you're only running at 50% efficiency then boosting a single battery to 5V with a 350mA output means you have to draw about 2.5 amps from the battery! Not good...