LED Efficiency

[twhitehouse]

I cannot find this info anywhere, all my google searches leave me with results that don't answer my exact question. I'm sure someone here will know...

My question is, what is the ratio of light-to-heat produced by LEDs? (I know this will vary, of course.)

For instance, an XR-E gives off a certain amount of lumens per watt, measured at 350mA. And an XP-G gives off more lumens per watt, also measured at 350mA, because the light output to watts ratio is more efficient.

My question is, what percentage of that 1 watt is light, and what percentage of that watt is wasted heat?

This would help me in building heat sinks for a bike light I'd like to build. I know the rules-of-thumb for square inches of HS to watts of power... for example, if I had one XR-E, 3.7Vf @1000mA, that would be 3.7 watts of power, but since some percentage of those 3.7watts is light, and the rest is heat, I don't need to heatsink for the entire 3.7 watts. (Or so I believe).

Also, if anyone has any simple answer to this one: what is the thermal resistance in degrees celsius/watt of aluminum and copper? (I know that is a much more complicated answer than can probably be given, involving the size and thickness of the heatsink and such...)

This: http://www.candlepowerforums.com/vb/showthread.php?t=289441 gives me a partial answer, but doesn't tell me the thermal resistance of materials such as copper or aluminum. a reply also touches on the heat/light ratio, but it seems abstract.

Thanks for any info, this was sort of a rambling post...

 

[AnAppleSnail]

I cannot find this info anywhere, all my google searches leave me with results that don't answer my exact question. I'm sure someone here will know...

My question is, what is the ratio of light-to-heat produced by LEDs? (I know this will vary, of course.)

Also, if anyone has any simple answer to this one: what is the thermal resistance in degrees celsius/watt of aluminum and copper? (I know that is a much more complicated answer than can probably be given, involving the size and thickness of the heatsink and such...)

A broad rule of thumb is that LED output is approximately 350 lumens per watt. So a light emitting 100 lumens per watt emits (100/350) W as light and (250/350) W as heat.

The C/W depends entirely on the heatsink in question. Broadly speaking, it depends on the exact alloy, air temperature, air circulation, and fin design (size, spacing, thickness). You can find this information published on a very few heatsinks, but extra heatsinking is cheaper than not enough in the long run.

 

[twhitehouse]

A broad rule of thumb is that LED output is approximately 350 lumens per watt. So a light emitting 100 lumens per watt emits (100/350) W as light and (250/350) W as heat.

The C/W depends entirely on the heatsink in question. Broadly speaking, it depends on the exact alloy, air temperature, air circulation, and fin design (size, spacing, thickness). You can find this information published on a very few heatsinks, but extra heatsinking is cheaper than not enough in the long run.

Ah, very good. One quick question, though: with the rule-of-thumb of LED output being 350 lumens per watt, is that a theoretical figure, used to make basic calculations? I only ask because I've read the Cree XM-L news releases about it having 160lumens per watt... and it being the new high mark. (If so, I wonder what will happen if they ever get to an LED that does 500 lumens per watt!)oo:

I kinda figured heatsink info would be complicated in those ways mentioned. Best way to go is build something on paper, give it a margin of error, and test it in the real world, I suppose. (That's what I did with my first bikelight... 12XP-Es on a big ol' aluminum rig... on paper, it was cutting it close, but in real life, it hardly gets warm)

Thanks!:twothumbs

 

[AnAppleSnail]

Ah, very good. One quick question, though: with the rule-of-thumb of LED output being 350 lumens per watt, is that a theoretical figure, used to make basic calculations? I only ask because I've read the Cree XM-L news releases about it having 160lumens per watt... and it being the new high mark. (If so, I wonder what will happen if they ever get to an LED that does 500 lumens per watt!)oo:


Thanks!:twothumbs

Well, that'd probably be impossible. Of course, calculating the exact lumens per watt contained in white light is tricky, as you have to sum up the power at every wavelength and sum them up. But at some theoretical maximum, all the electrical power is becoming light, and no more light could be made from that power.

And yeah, you kinda have to build it and see if it works. You can measure the C/W for a heatsink, assuming you have a known heat source and a good thermometer. 10 watts in, measure temperature of the air and temperature of the heatsink, and you're good.

 

[DIWdiver]

The thermal resistivity of aluminum is about 0.004 meter-kelvin/watt, copper is about 0.0025.

As you suspected, that really doesn't help. You want the thermal resistance of the heatsink, which probably has far more to do with it's size and shape than with the resistivity of the material it's made from. At least that's true if you're comparing aluminum to copper. Iron and steel are quite a bit worse.

You can get an idea of whether thermal conductivity (the more commonly used term, which is the reciprocal or resistivity) is a factor by checking the temp of various parts of the heatsink. If it's hot near the source and cool near the edges, then the conductivity is a factor, and raising the conductivity (say, by changing from aluminum to copper) would help. If the temperature is fairly uniform, then conductivity isn't a factor and you must increase the surface area to make improvements.

 

[wyager]

LEDs are less efficient than lasers. Lasers are about 25% efficient (except for red or IR diodes). It would be easy to measure if you have a laser power meter.

 

[twhitehouse]

Thank you to all who have replied! I do so love CPF... it's nice having a group of supportive, enthusiastic people who enjoy LEDs and light the same way I do. As an auto tech by trade, I understand electricity well, but the ins and outs of LEDs sometimes escape me due to lack of general, well-explained info online. When that happens, I turn to CPF and find a welcoming treasure trove of knowledge!

This info will definitely help me build a bike light for my wife (who tends not to understand my obsession, hehehe) as I need to build a bright, compact light for her. My bikelight is a monstrosity of cobbled-together parts, which looks like crud but works for me... hers must be sleek and unobtrusive. And bright!:devil: She might not understand lumens, but dangit, I will make her an awesome light!

Thank you, again.

-Whitehouse

 

[blasterman]

If you dig around Cree's tech notes you can find the general thermal resistance requirements for their XP-Gs at designated wattages. Some commercial heat-sinks will advertise a thermal resistance, or sites like Mouser or Newark will allow you to search for them specifically. Then it's easy because you can match the two up.

However, heat-sink specs are rarely as accurate as they should be. Or, if you're making your own sink you need a starting point. In which case you can look up similiarly sized sinks on those sites and go by a rough average. At worst it will keep you from toasting an emitter and give you a general design starting point.