Tuesday, April 16, 2013

Protecting LED lights

Lighting vendors have dialed-back their claims for LED life. A decade ago they were all telling me I’d get up to 100,000 hours but now 50,000 to 60,000 is a more typical claim. In my experience though, even that is rather bold.

The issue is heat. LED’s generate quite a bit of it, and life falls as the LED junction temperature rises. (There’s a chart on page 8 of this Phillips Lumiled paper showing how a rise from 120 to 150oC can cut life by 50,000 hours. Noribachi also publish an excellent paper: “LED Junction Temperature and Lifetime.”) So the key to long life is good thermal management.

I have to leave the internal design to the manufacturers, but I do try to ensure LED lights are mounted on aluminum heat-sinks. However, I have tended to overlook the interface between the two. Recently, I’ve realized that a thermal interface material could dramatically improve heat transfer, so that’s what I shall be including from this point on.

If you’re interested in learning more about thermal interface materials, I found a good tutorial on the Dow Corning website, and a selection of “cool” products at Indium Corporation. And yes, that was a very feeble pun.

1 comment:

David Dechow said...

Good point regarding the LED killer: heat. For most machine vision applications I find it best to strobe or at least turn the LED source on only during acquisition. LED's are not (or minimally) affected by on-off cycling, and as long as the duty cycle limits of the light at specified power consumption are not exceeded, any periods of being off should both reduce heat and increase life relative to a constant on state.