100w LED chip temp

[sportcoupe]

What color was the heatsink? If it was black (anodized), then that measurement could have been somewhat accurate though perhaps 10% lower than the real number. If it was bare aluminum, or even heavily oxidized aluminum then the actually temperature was a lot higher than your reading. If it was clear anodized then count on it being 20% higher. Those thermometers are usually calibrated for 0.95 emissivity ... but aluminum is almost always lower than that. You can cover it with a lot of black marker and get a better reading, even if black anodized. Even better would be black paint.

Since I have a fan on the heatsink, the surface finish doesn't matter very much at all. Otherwise, the debate is still out on silver (bare) verses black heatsinks.

 

[ssanasisredna]

Since I have a fan on the heatsink, the surface finish doesn't matter very much at all. Otherwise, the debate is still out on silver (bare) verses black heatsinks.

It matters when you are measuring the temp.

Jury not out on color. Black is better versus bare.

 

[bwinzey]

Something doesn't add up. Like enderman, I've run at less than 100W on a big old Pentium 4 heatsink with a fan and it gets too hot to touch (50 - 60C). Only takes a few minutes.

I have a 100w LED running at 34v @ 3A(full draw) on an OEM FX-8120 heatsink. With the fan off, it will probably burn after a few minutes, but with the fan on, it's only warm to the touch and can run indefinitely (I left it on for 2 hours at full brightness and it was as warm as when I left it). I made sure to put plenty of thermal paste on it and it works fine, so I'm not sure what you're doing wrong. Is the LED you're running designed to be run at 100W? You gotta remember that LED's become EXTREMELY inefficient when driven above their specifications, and instead of creating more light, they will just create more heat.