100K hour life vaporware

[tvodrd]

Was reading Machine Design today and found this:

The technology for 100,000-hr, high-brightness LEDs doesn't exist," says Philips Application Engineer Pat Goodman. "We have published what is probably the longest database on lifetime data and it only goes out to 9,000 hours. To extrapolate from 9,000 to 100,000 hours is quite a stretch.

Larry

 

[jtr1962]

Well, I already have a data point that goes out to over 37,000 hours ( 4 years, 3+ months). That's how long my Q-bin test Luxeon has been running at 350 mA. For about the last three years, it has measured 10.3 lux at 1 meter. No idea what it started at because I didn't have a light meter then, but it didn't really fade noticeably yet.

 

[tvodrd]

That's encouraging news, as I'll be long dead before I can put 37K hours on my avatar! :thumbsup: (Oh, it's relatively well heatsunk!)

Larry

 

[TigerhawkT3]

Goodman may be referring to 5mm LEDs, which can degrade quite noticeably over a relatively short period of time, especially if poorly heatsinked. Power LEDs (Lux, Cree, etc.) do much better.

http://www.molalla.net/~leeper/5mmdeg.htm

 

[nerdgineer]

Can't find the article, but a 5 mm white LED test a few years back indicated that, driven at 20 mA with proper heatsinking, 5 mm white LEDs (including Nichias IIRC) go about 6000 hours to 50%. Upping the current reduces their lifetime drasticallly. I think 60 mA dropped the 50% life to 200 hours. At 100+ mA, it was less than 10 hours. My memory for numbers may be a little off, but that was certainly the gist of it.

Nowadays, we can get more light at lower currents, so the lifetime can be extended.
Luxeons were tested elsewhere and (at 350 mA with good heatsinking), did something like 70,000 hours to 50%.

 

[matrixshaman]

Phillips is just in a panic that they are going to get stuck with a gazillion flourescent and incandescent bulbs if LED's take over as fast as it looks like. :laughing:

 

[chocho]

Goodman may be referring to 5mm LEDs, which can degrade quite noticeably over a relatively short period of time, especially if poorly heatsinked. Power LEDs (Lux, Cree, etc.) do much better.

http://www.molalla.net/~leeper/5mmdeg.htm

+1 to that. I have LED candles sitting in our windows (always on) that I make with 2 D batteries and 10mm LEDs. The 10mm LEDs have a 3.3V forward voltage, thus, they are being under driven. The only heat-sinking is from contact with the (+) battery nubbin to which it is firmly taped. They lose about half their brightness every 3 months. I replace them every year (the LEDs, not the batteries).

 

[2xTrinity]

Can't find the article, but a 5 mm white LED test a few years back indicated that, driven at 20 mA with proper heatsinking, 5 mm white LEDs (including Nichias IIRC) go about 6000 hours to 50%. Upping the current reduces their lifetime drasticallly. I think 60 mA dropped the 50% life to 200 hours. At 100+ mA, it was less than 10 hours. My memory for numbers may be a little off, but that was certainly the gist of it.

I'd be highly surprised if the figures are actually that high. Especailly for 60mA compared to 20mA. That's more than a 3x increase in heat production, in a device with extremely high thermal impedance, which means skyrocketing temperatures. Life expectancy of LEDs decreases exponentially based on temperature, which should mean a much more dramatic falloff in life due to overdrive like that.

I have no doubt 100,000 could be done -- consider that high power emitters have thermal impedances on the order of 10 Celcius-Degrees/Watt (as opposed to hundreds for 5mm LEDs), if you were to drive such an emitter with half a watt, junction temp will only be a few degrees higher than the heatsink, which if it's big enough could easily be kept only a few degrees above ambient.

 

[evan9162]

Phillips is just in a panic that they are going to get stuck with a gazillion flourescent and incandescent bulbs if LED's take over as fast as it looks like. :laughing:

Just who do you think owns Lumileds?

 

[SemiMan]

This comment just does not make sense. I had seen presentations years ago from Lumileds with this much life test data. I wonder if the comments, etc. were taken out context or this was a mix of old and new stuff?

Semiman

 

[evan9162]

He's probably referencing this data:
http://www.luxeon.com/pdfs/AB07.PDF ... page 3.

But if you look here:
http://www.luxeon.com/pdfs/RD25.pdf
On page 4, they have a graph that clearly shows test data out to almost 20K hours, and this was data from 2003.

So he's either referring to data on 5mm LEDs (which seems to be true), or the guy is completely out of the loop and talking out of his ***.

 

[TigerhawkT3]

Anyone remember Philips' other gem of "there's no heat involved [with LEDs]" from the NGC "Manmade" flashlight episode? :laughing:

 

[evan9162]

That show was bad. Really bad.

 

[tvodrd]

That show was bad. Really bad.

Amen! I think I was in it.

Larry

 

[Trashman]

That show was bad. Really bad.

Except for, of course, the showing of our fellow CPF members in the "flashaholic" sections of the show!

 

[greenLED]

Except for, of course, the showing of our fellow CPF members in the "flashaholic" sections of the show!

If you ignore the part where Beanhaed shows up. :nana:

 

[chimo]

The life of an LED depends on how hard your drive and cool them.

"The candle that burns brightest........."

Here is a thread from a while back to see the effects of testing on some common 5mm LEDs.

Paul

 

[nerdgineer]

I'd be highly surprised if the figures are actually that high. Especailly for 60mA compared to 20mA...

I finally found the article. My memory was in fact off. The tests they documented (see page 9) show (for good quality white 5 mm LEDs) a 50% degradation life of 5000 hours at 20 mA, 2000 hours at 50 mA, 500 hrs at 90 mA, and 200 hours at 110 mA.

I'm guessing that these current 5 mm white LEDs are brighter but may be built to the same overdrive lifetime degradation performance.