LED Lifetime measurement

[kulasekaran]

Hi,

Is there any instrument for standard method of measuring LED lifetime?

Is it possible to predict LED lifetime by observing its output with overdriven current supply?

Thanks in advance

 

[IMSabbel]

Well, there is the obvious one: Just run the LED under specified usage conditions (i.e. temperatur&current according to datasheet) and sample the output in regular intervals.
You should get enough data for a reasonable fit at 20-30% of the lifetime (in the 1/2 Output definition).

While overdriving the led does indeed cause a VERY strong increase in aging, i dont think that its useful in absolute gauging. Its just very hard to correctly seperate the different non-linear factors (electromigration at the contacts, point-fault creation in the semiconductor, phosphor aging, encapsulation breakdown).
It might very well allow educated guesses, but for the real thing you just have to have patience.

 

[Calina]

While overdriving the led does indeed cause a VERY strong increase in aging, i dont think that its useful in absolute gauging. Its just very hard to correctly seperate the different non-linear factors (electromigration at the contacts, point-fault creation in the semiconductor, phosphor aging, encapsulation breakdown).
It might very well allow educated guesses, but for the real thing you just have to have patience.

Yet it is the way the manufacturers do it!

 

[David_Web]

Don't trust lifetime figures ether.
As they are for when 50% of the leds are dead.
If it emits light it's working.
This is the normal way to show the runtime for leds.

The good thing is that some start to use service life to show lifetime. That means that when the total output has fallen to 20% that's their life. Ether by decrease in brightness or by burning out.

But that's not always used ether. Some mfgrs make ut their own ways to specify lifetime. Some better, some not. Most is not.

Lumileds does make it clear that it's only pridictions.

 

[evan9162]

Don't trust lifetime figures ether.
As they are for when 50% of the leds are dead.

No, that is not how it's done.

Lifetime figures vary, but the vast majority of them are a projection to when the LED will lose 30% of its initial output.

Lumileds goes one step further with it's lifetime analysis. For the Rebel and K2, they provide charts that show estimated lifetime based on junction temperature, current, and acceptable failure rates (both 50% of devices losing 30% of output, and 10% of devices losing 30% of output).

LEDs don't catastrophically fail due to aging like light bulbs do. Under regular usage scenarios (driving under normal paramters, with Tj less than maximum), the output slowly decreases over the lifespan.

http://www.luxeon.com/pdfs/RD06.pdf

 

[RODALCO]

From field experience i know of LED's in utility service which have gone well over 100,000 hours. 140,000 hours ( 16 years) with no faillures.

Ok I have fitted many thousands of them in POCO kWh meters for potential fuse indicators on CT meters.
LED on , fuse Ok, meter registering correctly.

I started fitting these 1991 onwards running directly on 240 Vac via a 47 kOhm 4 Watt series resistor per phase.
LED current about 5 mA which yields very good light with high efficiency LED's anyway.

140,000 hours is not bad and these are still going strong.

Today i fit a lot of LED indicator lights on overhead pilot relays and these have to withstand spikes from lightning etc.
The simplest option usually works the best. 2 x 33 or 39 kOhm 1 Watt resistor in series, 1 N4007 ( to reduce power in R's) and 1 N914 blocking diode for negative spikes.

Cheers, Raymond

 

[David_Web]

evan9162 No that is exactly how it's done unless otherwise stated.
As you noted out Lumileds (Philips) has kindly stated the service life of their leds down to 30% loss of total light in an installation. It is nice of them to do so and it does show that they care about quality.

Sadly the standard way to measure life is to 50% total failure of the light source. No matter how much we wish otherwise.

It is nice to see that most of the big guys thinks that this is wrong and want to set higher standards when it comes to LED.

The 50% figure of popped lightsources to measure their lifetime is the international standard of measuring the lifetime for most light sources. It is done with 3h on and 20 min off and so on.

Most serious mfgrs state service life for most "long life" sources like fluorescent tubes, metalhalide etc.
It is down 20% of the total light output. By failure of the lightsource or by gradual loss of light. Both are taken into account. Measured with the same 3h on 20min off.

 

[laserblue]

Also the lifetime of the LED is dependent on the amount of overvoltage or overcurrent it received.