LED life vs. ?

All answers, links & opinions are welcome.

I think this is led model specific, and theres no magic number but an average, and it would follow a normal distribution.

I’d love see the ramifications for emitter life to 70% output as a function of drive current.

It will have far less to do with drive current than it will with junction temperature.


That’s what I figured. It’s more precise for me to say that due to match’s latest emitter testing roundup I want to analyze the extent to which high (read: not so extreme as to cause immediate, catastrophic failure) drive currents contribute to metal migration, and their impact on bond wire integrity and output over time.

i look forward to seeing all this data with great anticipation!

Alas, I’m just a student today :nerd_face:
But one day, I tell you! One day! H)

Don’t need a degree to follow the scientific method for testing theories. Get some samples, a start them cooking…


but the lifetime of LEDs is so high it would take years, as i understand it Cree releases their LEDs with about 3000 hours of testing even though they are rated for 25-100 thousand hours.
Mind you, if i had the money and equipment, i would like to test some CFLs to see how many on/off cycles a particular model can take so i can determine how many hours of life (equivalent) i lose every time i switch it on and off.

Typically the failure rate of electronics doubles for every 10 degree C increase in temperature…


interesting results, some of the numbers don’t seem to make a lot of sense though, and these are extrapolations as well, but lets hope conservative ones.

Some info on life and accelerated life testing

The 8C to 10C applies to many chemical reactions but I’ve seen it mostly applied to motor insulation life from testing done a century ago, probably at GE.

Since case and ambient temps are the same, the table data used an infinite heatsink, but theta J-C is necessary to figure junction temp.
A silicon diode junction Vf drops 2.2 mV per degree C with good linearity but LEDs might have a different coefficient.

Thanks for all your inputs :slight_smile:

As far as I can tell, by looking at long-dead posts on this forum and others; and with a little evidentiary support from my own junk box, I would like to offer this as a general theory:

The LED in question will last about 3 times longer than it will remain useful in real life.

A corollary assumption would be that, therefore, it will NEVER die unless you or I take action to kill it.

But I’m just


up to me to scrounge up some data sheets and do some homework.