Best Low-Drain, Long-Life Battery

November 14, 2012 9:10 AM

How can I continuously run a 3W - or less - LED light with +/- 350 lumens as long as possible (1 year minimum) with a relatively compact battery source?

Objective: To continuously run a 3W - or less - LED lamp with +/- 350 lumens as long as possible (1 year minimum) with a relatively compact battery source.

Restrictions: Protected, enclosed, exterior application. 8" x 8" x 10" space. Battery should be relatively easy to purchase and replace. No exterior or remote parts or sources allowed. Therefore no photovoltaic panels, et cetera.

Latitude: Does not need to be pretty.

Proposed light fixture or similar.

Alternate (with simple screw mount fixture).

posted by xod to technology (18 answers total)

You can't.
posted by ryanrs at 9:17 AM on November 14, 2012

Well, not at 3 watts anyway. 3W for 1 year is 27 kW hrs. With dozens of D-cells, you can maybe do 1% of that. You'll have self-discharge issues if you try to use rechargable batteries.
posted by ryanrs at 9:25 AM on November 14, 2012

A year is 8,760 hours, times 3W is 26.28 kWh. I am no expert on batteries or LEDs, but my instinct is that an 8*8*10: battery that can hold 26.28 kWh is improbable.
posted by jon1270 at 9:25 AM on November 14, 2012

3w * 24h * 365 = 26.6kwh. By way of comparison an iPad 3 has a 42.5wh battery, which are pretty near the top of the game of energy density. You'd need 500 of them. Which aren't going to fit in the space you have set aside. And would cost tens of thousands of dollars.
posted by seanmpuckett at 9:25 AM on November 14, 2012

And would self-discharge before a year even with no load.

Reconsider solar or running a wire. It is the only reasonable way to make this work.
posted by ryanrs at 9:28 AM on November 14, 2012

3 watts at 12 volts is 0.25A. Over a year, that's 2200 amp-hours. A motorcycle-sized deep cycle battery only holds 33 amp-hours. A couple 6V lantern batteries is about 40 amp-hours. A couple of these monsters is about 100 amp-hours.
posted by zsazsa at 9:28 AM on November 14, 2012

So, to clarify, y'all are saying, "No." Do I have that right?

Seriously, thank you. I needed that.
posted by xod at 9:36 AM on November 14, 2012

Your volume is about 10.5L, and your energy required is about 97MJ (or about 27kWh, as above). That's a volumetric energy density of over 9MJ/L. The only practical energy sources that can achieve this are combustible or nuclear fuels, and the size of the components required to turn those fuels into electricity will not work with your constraints.
posted by Behemoth at 9:38 AM on November 14, 2012 [6 favorites]

So the Cree XLamp XM-L LED assembly looks like it'll give you 160 lumens per watt. I'd have to look up the efficiency curve, but that suggests you can get 320 lumens for 2 watts by just doubling up. So that gets you down to under 20 kW hours, but all the above issues about battery density apply.
posted by straw at 9:39 AM on November 14, 2012

Just to give you an idea, if you could somehow get ahold of a nuclear-powered battery, you'd need about seven grams of plutonium inside the battery to provide the power you desire.

The few remaining folks with plutonium pacemakers — under a hundred — have their pacemakers removed and shipped back upon death, and those involve tiny specks of plutonium, so, yeah, they don't let that stuff wander around.
posted by adipocere at 9:43 AM on November 14, 2012

Yeah it's too bad, even a hydrogen fuel cell won't get you what you need in the volume you want at 5MJ/L. Still, these guys have some seriously nifty stuff. Maybe you'll find another approach you can use.
posted by seanmpuckett at 9:50 AM on November 14, 2012 [1 favorite]

As everyone has written, you simply can't. But, depending on your application, there may be some way to use an energy input into your box. Is it possible to use air currents, temperature differences, or light (would a PV panel inside glass count as enclosed)?
posted by ssg at 9:50 AM on November 14, 2012

I'm no expert, but one thing neglected above; you don't run an LED at full power continuously. You pulse it rapidly, both to lower power consumption and to extend the length of the LED. Apparent brightness is affected but not as much as you might think. I don't think you'll get the 500x power savings the estimates above require, but it might help.
posted by Nelson at 11:54 AM on November 14, 2012

Can you futz with the duty-cycle any? I vaguely remember when the LED throwies were a big deal that one could extend the lifetime by throwing a little circuitry in the mix to give the LED something like a 10/90 on/off cycle at a high enough frequency to not cause obvious flicker. I came away with the impression that full blown LED installations weren't simply LED+resistor+battery going full blast all the time.
posted by zengargoyle at 11:58 AM on November 14, 2012

You might want to pose this question, particularly the duty-cycling angle, on the Candlepower Forums. This is a discussion forum for hardcore flashlight and LED nerds. Highly recommended .
posted by werkzeuger at 12:16 PM on November 14, 2012 [1 favorite]

Running an LED pulsed is less efficient than running it at constant current. Pulsed operation also does not increase the lifetime of an LED.
posted by ryanrs at 12:21 PM on November 14, 2012

Your volume is about 10.5L, and your energy required is about 97MJ (or about 27kWh, as above). That's a volumetric energy density of over 9MJ/L. The only practical energy sources that can achieve this are combustible or nuclear fuels, and the size of the components required to turn those fuels into electricity will not work with your constraints.

tl;dr: When this baby hits 88 miles an hour, you're gonna see some serious shit.
posted by thewalrus at 12:58 PM on November 14, 2012 [1 favorite]

You seem to be starting your quest with a solution, rather than a problem.

From what you propose in your solution, it seems your technical skills are somewhat limited.

I wonder if this thread may help you reach your goal more directly if you rephrased it with a description of what you are trying to achieve and getting some input based on that? There is a lot of expertise floating around here waiting to pounce on interesting problems.

Gotta say though, for anyone who glanced at, read your question, and read the responses, you've put out a puzzle that is likely aggravating the hell out of several curious folks, i.e., "What possible application would need this under these constraints?" So we're all out here dealing with that. You really owe it to your fellow mefis to clear this up! (Said with a smile!)
posted by FauxScot at 2:20 AM on November 15, 2012 [2 favorites]