100,000 Hour LED Life??

SATURN

Newly Enlightened
Joined
Sep 7, 2003
Messages
17
With respect to the actual lifetimes of LEDs, the manufacturers could not have possibly tested them for 100,000 hours. It would require more than 11 years of continuous testing to accumulate 100,000 hours on an LED (at approx. 8,800 hours per year). By then the tested LED would have been obsolete for a number of years.

It is unlikely that the a manufaturer would have time to perform more than a 10,000 hour life test before bringing a new LED to market.
 

Navck

Enlightened
Joined
Oct 15, 2005
Messages
728
Location
Southern California
They usually "simulate" those hours (Instead of driving the LED at 50ma, they'll slam it with 250ma or something really high.) Or drown it in a ultra humid enviorment, or base it off some other LEDs
Old LEDs that are driven lightly, like in computer lights for your keyboard, can technically last forever, the lights you have won't last forever, but they'll last damn long.
 

chimo

Flashlight Enthusiast
Joined
Sep 16, 2004
Messages
1,905
Location
Ottawa, Canada
Very often it is the manufacturer/repackager of the light product that slaps the 100,000 hr claim on the LED life and not the actual LED manufacturer. It is just another example misleading marketing.

Look for the 100,000 hr claim on Nichia data sheets, I don't think you will find it.
 

Moat

Enlightened
Joined
Sep 24, 2001
Messages
389
Location
Mid Mitten
From what I understand (please, anyone correct me if I'm wrong), even high quality LED's - driven at spec - will lose output over time. Also, the lumen rate of loss is fairly linear. So, for example, if a test shows a 5% loss over 1000 hours, it's pretty accurate to assume a 50% loss over 10,000 hours. They can make a fairly accurate projection of where it's output will be years down the road, in just a few months of testing.

AFAIK, anywhoo.

I'm still not sure where they draw the "lifetime" line, tho... 50% of the initial output? 80%? 10%? 0%?
 
Last edited:

voodoogreg

Flashlight Enthusiast
Joined
Apr 27, 2003
Messages
1,829
Location
Global
Good read.chimo. I posted a thread about this myself after running an informal
test of longevity on a CMG ultra "G" without me knowing it :huh2:

I had the light hanging over my PC keyboard to work at night when my daugther crashed on the couch so i didn't need to keep light's on.

I ran it on a ROV IC3 nimh AA and 9 times outta 10 I just left it on to run down. In about six mth's i took it down when it seemed to be a little dim.
I Put two E92's in it and another "G" one that new, was a little dimmer then my
"overhead" "G". It is as now noticeably brighter by a big margin, then the one i had been using
all the time over the PC. so these do go dim and down hill.

Side note, someone has to have said to the manufactor's of the light that it does work for 100.000 Hr's at least back a few yr's, because i keep the inserts, instructions, everything but the plastic package in a folder
most all 5mm LED are rated at 100,000 or 50,000 hr's. Nichia may not post
those runtimes now but i have to many brands and multiples of 5mm light's
so i just don't buy so many light makers deciding on that life span,i think
what is published now on there site wouldn't be the same when they came out with 5mm white LEDS. (imho ymmv never pet a burning dog :p VDG
 

darkknight

Newly Enlightened
Joined
Feb 5, 2006
Messages
25
In general the Luxeon I and III emitters are expected to provide an average of 70% lumen maintainance after 50,000 hours provided the Luxeon is driven at or below 350mA for Luxeon I and 700mA for Luxeon III. Also, provided the junction temperature is kept at or below 90 degrees celcius. In other words a Luxeon I or III emitter, when driven at specified levels and temperatures, will still be 70% as bright after 50,000 hours of service. If you overdrive them or do not take adaquate cooling measures this number goes down. For instance, A Luxeon III emitter driven at 1000mA, but still with a junction temperature at or below 90 degrees celcius, will only be 50% as bright after 20,000 hours of use. Which is still very good. Keep in mind your eyes are not very sensative to light changes. So you probably wont notice any difference even after extended use. Also keep in mind the Luxeon 5 emitter has a much shorter life, so I would recommend if you want a long lasting light, stay away from the Luxeon 5 and just get a xenon based light. This is just my opinion of course. If you would like to read more about this kind of stuff and more just go to www.lumileds.com and read the pdfs for your emitter of choice.

Dark Knight
 

jtr1962

Flashaholic
Joined
Nov 22, 2003
Messages
7,506
Location
Flushing, NY
Red, amber, orange, and yellow-green LEDs probably do indeed last 100,000 hours or more. In the mid-1980s we started using LED backlights for LCD taximeters. The first ones were mounted on a board to replace the incandescent lamps which were always burning out. Later on the company started using 8 yellow-green LEDs mounted on the board, driven at roughly 30 mA in a 2 series x 4 parallel arrangement with resistors. This was around 1989. We're starting to see those same meters being sent in for repair now because the backlights are getting too dim, probably half the original brightness. However, note that this took 17 years to occur, and on average the backlights are on 14 hours a day. That's ~90,000 hours accumulated run time at 50% overdrive.

What I'm finding fascinating is how much LED technology has advanced in the intervening neary two decades. We're now replacing these LEDs with true green ones. The higher forward voltage with the resistor setup means they're running at a safer 20 mA. The brightness increase is phenomenal, easily 10 times what the originals did. Lifetime shouldn't be much of an issue for two reasons. First, it seems the white 5mm LEDs and to a lesser extent the blue ones suffer from reduced lifetime. The true green should last 50,000+ hours before dimming to half brightness. Second, the new LEDs will actually need to dim to 1/10 brightness before they would approach the level where the customer would complain. The meter will be obsolete long before that happens.
 

DCFluX

Newly Enlightened
Joined
Jan 4, 2006
Messages
145
Location
Beyond The Sun
It's simple, you take 1000 LEDs and run them for 100 hrs in a enviromental chamber, if none fail you have your 100,000 hr MTBF. This is done all the time with consumer electronics. Ever wonder why a DVD drive with a MTBF of 50,000 hrs fails in 3 months?
 

jtr1962

Flashaholic
Joined
Nov 22, 2003
Messages
7,506
Location
Flushing, NY
DCFluX said:
It's simple, you take 1000 LEDs and run them for 100 hrs in a enviromental chamber, if none fail you have your 100,000 hr MTBF. This is done all the time with consumer electronics. Ever wonder why a DVD drive with a MTBF of 50,000 hrs fails in 3 months?
Yes, except that MTBF is entirely different from lifetime. For example, I've seen some hard drives with 500,000 hours MTBF. This does not mean that they'll last 500,000 hours on average before failing (that would be 57 years continuous running). Rather, it means that if you have a large sample of drives you can expect one failure for every 500,000 hours of combined running. For example, if you have 100 drives, then in one year these drives run a combined 876,000 hours. You can expect on average 876,000/500,000, or about 1.75 drives to fail each year. Or put another way, you should have one drive failure on average every 7 months or so. In essence MTBF is simply a statistical parameter.

Now let's take lifetime. The same drive might have components with a designed lifetime of 5 years. This doesn't of course mean that every drive will self destruct when it is 5 years and one day old. Rather, after 5 years the MTBF of these same drives will gradually start to decrease. By the time the drives are all ten years old MTBF might well be down to only 5000 hours for those few which are still running, meaning that if you have 100 of these drives then you can expect one failure on average every 50 hours.

Now as far as LEDs go it has been well documented that many types of indicator LEDs, 7-segment displays, etc. have lifetimes of 100,000 hours or more. In many cases lifetime is arbitrarily chosen as the point at which the LED has dimmed to 50% of original brightness. This makes sense because half-brightness is the point where most people will start to notice the dimming, and this will mark the end of the useful life of these LEDs. LEDs are rarely speced by MTBF. In theory, they could operate practically forever, but past a point they won't be giving any useful light. My guess is if you were to test LEDs for MTBF you would get figures on the order of many millions of hours just like any other electronic component. In other words, an LED display screen with 1 million LEDs might only have a few actually fail every year, although after maybe 100,000 hours the entire screen would need to be replaced because it had gotten too dim. This would be despite the fact that probably upwards of 99% of the LEDs in the display would still be functional.
 

DCFluX

Newly Enlightened
Joined
Jan 4, 2006
Messages
145
Location
Beyond The Sun
I've seen some LEDs that have lasted 25 years and are still going, I have seen others last only 10 minutes (without being overdriven). And then there are the LEDs that don't survive shipping from the factory, those are my favorite.
 

photo2000a

Newly Enlightened
Joined
Jan 25, 2005
Messages
138
Location
ny
jtr1962 said:
Red, amber, orange, and yellow-green LEDs probably do indeed last 100,000 hours or more. In the mid-1980s we started using LED backlights for LCD taximeters. The first ones were mounted on a board to replace the incandescent lamps which were always burning out. Later on the company started using 8 yellow-green LEDs mounted on the board, driven at roughly 30 mA in a 2 series x 4 parallel arrangement with resistors. This was around 1989. We're starting to see those same meters being sent in for repair now because the backlights are getting too dim, probably half the original brightness. However, note that this took 17 years to occur, and on average the backlights are on 14 hours a day. That's ~90,000 hours accumulated run time at 50% overdrive.

<<you know one thing not mentioned in addition to all the above those taxi metere are prob being baked by the sun ,heat and or perhaps cold and vibration to have that last 17yrs in that harsh environment not bad!!
 

NewBie

*Retired*
Joined
Feb 18, 2004
Messages
4,944
Location
Oregon- United States of America
Many folks are prone to over driving the LEDs and also don't heatsink them properly. This causes the LEDs to degrade much more rapidly than normal.

Now, if you want to extend the life, you can underdrive the LED, which will greatly increase it's life.

A good example would be this chart:

nichlife.jpg
 

SemiMan

Banned
Joined
Jan 13, 2005
Messages
3,899
DCFlux - What you have defined is not so much the "life" numbers that are used for LEDs, but MTBF, which is very rarely quoted for LEDS. MTBF for LEDS will be on the order of millions of hours if the LED is well designed and treated well.

Life for LEDs, at least reputable power LEDS is actually a light output specification or ususable light specification. Supplier dependant either at 50% down or 30% down. I think some suppliers just guess (especially since I have seen some of their equations and they are just wrong). Others take the output they have achieved over a period of time, say 10-20,000 hours, and extrapolate based on semiconductor and LED knowledge the figures for say 50K or 100K hours as they would never been able to have real number for their products or they would never be able to release a new product.

Semiman
 
Top