LED life

[williamv0123]

Whats with the huge disparity lately with advertised LED life?
I am seeing manufacturers tweaking useable life waaay down from 100000 hours.
Im not real knowledgable about what goes into testing, but is this a result of how these chips are being driven or overdriven..
What can REALLY be expected of say Nichia, Luxeon or Cree ??
Any info appreciated.

 

[soffiler]

100,000 hours has never been a real number. Grab a calculator - it translates to 11.4 years. Obviously, when the first 100,000 hour claims began, no white LED had ever been tested for 11+ years because they weren't invented that long ago.

Accelerated testing - overdriving - is done to make the time go by faster. But you can argue all day long about how the accelerated testing correlates with reality.

Then you get into the definition of "life". As you know, LED's don't generally fail catastrophically, they just gradually get dimmer. The earlier 100,000 hour claims were based on a dimming down to 50% of original output. These days, the industry is tending to use 70% instead so right off the bat, if nothing else were different, the claims should be shorter and they are.

But things are different - manufacturers are getting smarter about the aging mechanisms and how they relate to drive/overdrive currents and temperatures, and they are getting smarter about how the accelerated testing translates to the real world. So these days, if the engineers get their way, the life claims are getting more and more realistic. Just don't let the marketing guys at those data sheets.

 

[greg_in_canada]

From what I've read on CPF, some white LED ratings are based on the life of the blue die alone, but in real life the phosphor degrades over time and the epoxy becomes cloudy. So the actual useful life is much lower than claimed.

Greg

 

[williamv0123]

Thanks. Good information. Still...10000 hours is a long time. And from what I've read thats a conservative figure. As long as real world data doesnt continue to cause this figure to drop even more.

 

[elgarak]

Most of the data is projected; i.e. extrapolation of the dimming curve in the first running time... That's why Luxeon Vs were initially rated at 500..1000 hrs. No one knew how long they would really run, numbers went up after the LED hit the market.

It's fairly difficult to model the failure of LEDs. They're complex structures, and a lot of things can go wrong that change the output. However, total failure (= no light) is pretty difficult to get without physical destruction. Might only be a tiny fraction of initial brightness, though .

It's seldom stated to which dimming level the 100,000 hrs go... probably not the CPF standard of 50%.

 

[Nake]

And if one gives up it's life, just solder another one in. There's enough sources for them around.

 

[williamv0123]

Excellent point. Just wish I had some modification experience.

 

[Nake]

Excellent point. Just wish I had some modification experience.

The only way to get experience is....do it. Until a few months ago, the only thing I had ever done to a light was change the batteries. I started playing around with cheap lights, changing leds, modifying reflectors to fit, etc. Most of it is common sense. Once you gain some confidence, you'll be willing to tackle allmost anything. You'll come to realize your limits. There's always help on CPF.

 

[williamv0123]

That's about all I've done...change batteries. I think I'll have to try my hand with some cheap units. Nothing like experience.
Thanks again to all you flashaholics who sent the excellent info..way more than I ever knew.

 

[Patriot]

The only way to get experience is....do it. Until a few months ago, the only thing I had ever done to a light was change the batteries. I started playing around with cheap lights, changing leds, modifying reflectors to fit, etc. Most of it is common sense. Once you gain some confidence, you'll be willing to tackle allmost anything. You'll come to realize your limits. There's always help on CPF.

Hey thanks Nake. You've given me a little confidence to try some fiddling around with some cheapies.

 

[reltor]

Whats with the huge disparity lately with advertised LED life?
I am seeing manufacturers tweaking useable life waaay down from 100000 hours.
Im not real knowledgable about what goes into testing, but is this a result of how these chips are being driven or overdriven..
What can REALLY be expected of say Nichia, Luxeon or Cree ??
Any info appreciated.

It is a matter of time until the manufacturers are building a 750 hour led. Wouldn't you want to sell more? How much engineering and testing goes into building a 60 month automobile battery? Whatever they spend must be worth it.

 

[williamv0123]

That is definitely food for thought but wouldn't that be cutting into one of the biggest selling points over incandescent? I guess there are only so many applications you can replace with LED's. Then you have to start selling more LED's

 

[VidPro]

just from general use.

the 5mm i would put at 20k hrs
luxeon 1W at 50k max
luxeon 3W at 50k (meaning they are a bit brighter after many years)

Overdriven
5mm about 5 min
Lux 3W 100hrs

if i was putting money on survival of the LED it would NOT be the cheap 5mm leds that they CLAIM will do the 100,000 most. the people with the most realistic specs , have provided light for the longest.
its the phosphors.

Single color NON-WHITE leds, go on forever it seems like, so your average totally underdriven indicator light rarely fails to do its job.

 

[williamv0123]

Thanks VidPro. Excellent info.
Too bad white is the most usable color for flashlights. It seems, without the need for phosphors, LED life becomes far less of an issue.

 

[wjb3]

delete

Sorry. Dead or Broken link.

 

[EngrPaul]

Because the MTBF of a discrete component (LED emitter) has no direct relationship to the performance life of the flashlight in service. As in, flashlight "ON".

Traditionally, manufacturers put on the package "100,000 hour LED" and let you assume whatever you want... which is good for the seller because the general consumer doesn't know what the reliability value really means.

Probably more recently consumer groups are forcing them to reconsider their vague claim on the package. Information printed on the package should be referenced to general consumer knowledge.

Less deception, which is a good thing.

 

[williamv0123]

WOW! This thread is gonna keep me busy for a while. Thanks for the data. Great stuff.

 

[nine]

Should start a magazine called LED Life: An illumination aficionado's guide to the LED lifestyle.

The cover picture should be portraits of B-Movie stars holding flashlights in ridiculous settings like the Amazon, dark alleys, etc. Sort of like the Surefire catalog and Blackhawk's pseudo magazines but tongue-in-cheek.

 

[chadd_from_pa]

I found a good article online about LED life here: http://www.lunaraccents.com/educational-white-LED-life.html

 

[soffiler]

I found a good article online about LED life here: http://www.lunaraccents.com/educational-white-LED-life.html

There's a problem with that article in the very first paragraph, where it asserts the "industry standard" is 50% lumen depreciation. Not true. While I suppose there is still no industry consensus, 70% seems to be emerging as a much more accepted number.