Calculating heat sink for power leds

[pier]

Hello Everybody,

I am new to this forum and this forum looks great for led lighting.
I am into bilding some power led based lighting system in my house.
I want to know the calculations for designing heat sink for power led systems.
I have worked with low power leds which did not require heat sinks.
I also was wondering if some sort of software was available for this.

Kindly reply back if any of you have some info regarding this.

Thanks

Regards
Pier.

 

[COAST]

I'm not sure there actually is a.... "formula" to "calculate" heatsink size/type..... But maybe I'm wrong, which I very well could be....

 

[HarryN]

Hi - useful question.

Lumileds has some nice example calculations and models in their data sheets. Heat transfer is a well known and studied area and it can be software modeled, or just worked out by hand.

As a very rough "rule of thumb" you need 1 sq inch of surface area exposed to the open air for every watt of LEDs on the heat spreader. Keep them at least an inch apart and a minimum 1/8 in thick Al under them and you shouldn't be too far off.

Most home lighting benefits from running them at about 1/2 max power levels.

 

[blasterman]

Most commercial heat sinks are rated in thermal resistance, which is a basic number telling you how much heat it can handle. For ad hoc heat sinks, it's a lot of trial and error (and how long you want the LED to last).

For 3-watt LEDs running at a full 700mA I try to give them at least 12 sq inches of radiating area. If I have a piece of typical 2" alu bar three feet long, then it equals 144sq inches front and back. That means 12 LEDs spread out - at most.

Note that dropping down to 500mA really cuts the heat a lot but doesn't change the light intensity so much.

 

[pier]

Hi - useful question.

As a very rough "rule of thumb" you need 1 sq inch of surface area exposed to the open air for every watt of LEDs on the heat spreader. Keep them at least an inch apart and a minimum 1/8 in thick Al under them and you shouldn't be too far off.

Most home lighting benefits from running them at about 1/2 max power levels.

So if i get it right,
for eg: if i use 5 x 1watt leds, I use a rectangular aluminium plate of 1/8" thickness and 1" width and 10 to 12" of length ? right ?

If this is a rough thumb rule, How do you take the consideration of fins on a heatsink ?

I tried a lot of googling but coudnt find any for the day. I am still trying...

 

[HarryN]

Please consider blasterman's comments carefully, as he is right on this.

If you want the LEDs to run a long time and be the most efficient, then 12 sq in per watt is a good "conservative" place to work. This will work for pretty much any LEd.

The other end of the thermal area needed - about 1 in sq per watt - is on the aggressive side, and will result in higher die temperatures, somewhat shorter life, and slightly less efficient. To use this much smaller approach, you need to be willing to acept some limitations, such as:

a) Your wife saying "wow - that is pretty warm". It won't be burning hot, but definitely warm.

b) Probably should limit yourself to Lumileds LEDs like a K2 or Rebel, as they are more heat tolerant.

Fins may or may not help you much with general lighting, depending a lot on how it is mounted. The way fins help, is that that the hot fin warms the air, and the air rises away from it, carrying away the heat. (or a fan blows the air through it) If you design limits air flow around the fins in any way, they are not very helpful at all.

Once again, as a quick and dirty "rule of thumb", you can assume the surface area of a fin contributes about 1/3 to the total area. Example - 3 in sq of fin is about like 1 in sq or flat area attached to the LED.

BTW - I was trained as an engineer and did plenty of thermal calculations with greater accuracy than this, but getting more detailed than this will not really help you that much more.

There is a certain amount of trial and error involved unless you are willing to hire a thermal design engineer to do the calculations. With modern software, these calcs are just amazingly accurate. Certainly I would do this if it were more than just a goof around the house hobby project.

 

[JohnR66]

+1 on running the LEDs on lower current (500ma).

If I were a luminary designer I would practice this. For example, you have limited control of what customers do with lighting. Have you ever noticed how some business leave their lot lighting on during the day? I see it all the time and it annoys me as it is a complete waste of energy.

Anyhow, the hot sun can severely limit the cooling latitude of the heatsink even if it were oversized, so lowering the current and allowing extra capacity in the cooling can give the LEDs a fighting chance at long life.

Of course, in your application, you have more control, but even so, I tend to run the LEDs at around 75% of current used to rate the lifetime.

On another note, there was an excellent test of how much better painted or anodized aluminum was at radiating away heat than bear aluminum. Consider this if you can't have a lot of convection fins in your design. Anyone have that link? I can't find it.

 

[kan3]

Always thought this site was fun

http://www.frigprim.com/online/natconv_heatsink.html

 

[degarb]

+1 on running the LEDs on lower current (500ma).

Anyhow, the hot sun can severely limit the cooling latitude of the heatsink even if it were oversized, so lowering the current and allowing extra capacity in the cooling can give the LEDs a fighting chance at long life.

On another note, there was an excellent test of how much better painted or anodized aluminum was at radiating away heat than bear aluminum. Consider this if you can't have a lot of convection fins in your design. Anyone have that link? I can't find it.

The people in this forum that run leds at 3 amps, kinda bothers me, since my constant eye rolling distracts me from their insanely bright beamshot photos that they work so hard to make... Firstly, most people are not flashlight fanatics because they 1. see them as penlights. 2. see them as having only 1 to 2 hour max runtime. And so, they don't fit in with daily usage patterns. I do notice a huge amount of extra heat above 500 ma even with today's leds.

Does it matter the kind of paint? I want a finned 1inx1in heat sink that is black for my xml wristlight driven (with dimmable buckpuck) at a max of 700 milliamps. (I will probably dime to about 560 milliamps, since my eye sees little meaningful diff in other tests between 700 and 560 ma.)

 

[Steve K]

this would be a place to start to understand thermal resistance...
http://en.wikipedia.org/wiki/Thermal_resistance

If you are buying a commercial heatsink, it should have a thermal resistance value for still air. Knowing how warm you want to let your LED get, and knowing the temperature of the air, you should be able to calculate how much power the heatsink can handle.

You can visit Digi-key or Mouser to find a commercial heatsink and links to the manufacturer's website.
I checked the Wakefield site, expecting to find some sort of app note on thermal resistance, but found nada. Kinda surprising, really.

 

[AnAppleSnail]

The big split is: Fanned or fanless heatsink

A fanned LED lamp can get you stupendously high brightness in an attractively small package. I have a little 8cm cube with a slow fan and 24W of LED on it over my lettuce plants inside. The fanned heatsink feels the same as ambient temperature. Without the fan, it would need a heatsink about the size (And weight) of a full dinner plate. This would take the cost from $60 (LEDs + spare computer heatsink for free) to about $150, just due to the heatsink cost. However, with just one fan (And no thermal protection...yet), this lamp will destructively fail if the fan jams. A stray ladybug could damage this light. Multiple fans, thermal cutoff fuses, etc are nice to have. A driver that detects a Vf change would be the best way to measure the LED temperature, but that is complex customized behavior. High-intensity lighting, small/compact fixtures, and tech demos are best done with fanned heatsinks.

An unfanned heatsink is fairly simple. You get a great honkin' chunk of metal and stick LEDs to it. Relying on convection or stray breezes to distribute air, the heatsink has a higher temperature and is vulnerable to dust coating the heatsink over time. This works well for bar lights (A 2m / 6ft or longer metal bar, with LEDs along its length), large plates, and so on. Lower light density, more diffusion, softer shadows. Larger fixtures, more metal cost, and completely passive heat control.

If I were to try creating a line of LED fixtures for hobbyists, I would design around 12v power with standard computer fan sizes supported. I would include a thermal snap fuse that opens at 70C (Dangerously warm; burns skin within 5-10 seconds) in the center of the heatsink, a red LED in series with the fan (To indicate power consumed at the fan), and a hanging-pot balanced chain system with rail mounts on the heatsink face for 20mm stars. I'll have to see what is needed to modify computer heatsinks for this architecture. That's my thoughts!

 

[Steve K]

that's a good point about fans. Some heatsinks come with thermal resistance data based on x cubic feet of air per minute (is there a metric equivalent?). I'm not sure how you interpolate that data if you have a different size of fan.

Dust is the enemy of heatsinks everywhere, forced air or not. I've got a house fan (desk fan?) that collects dust on its blades. My desktop computer collects not just dust, but cat hair and anything else floating around. Yuck. Anything with air flowing past it will collect stuff floating in the air. My guess is that the lower air speeds of passive cooling won't suck up the big stuff such as cat hair.

For lights where there's no reasonable way for convection to work, such as those can lights recessed in a ceiling, fans are certainly attractive. Same for any place with space restrictions.

Each projects will have its own constraints on the solution. Some projects are more cost sensitive, some are more space limited, some are placing a priority on reliability, and some may need a sealed enclosure. The important thing is to know what all of the options are, and how to do the required calculations and determine which best meets your requirements.