White Leds degrade a lot faster than we think...

I found a paper describing the results of 5 mm white LED life (and other) testing done by Lighting Research Center for a conference in 2000. A link is here

What the researchers found is that the 50,000 to 100,000 hour lifetimes most manufacturers claim and most people believe for leds is not true, especially for white leds.

The number of hours for a typical 5 mm white led (circa 2000) runs before degrading to 50% initial efficiency varies with the current level (of course), but even at the recommended drive current of 20 mA, it was less than 6000 hours. When overdriven, the led life (to 50% loss) went down fast: to about 500 (hundred, not thousand) hours at 90 mA and about 200 hours at 110 mA.

Maybe 200 hours is enough (20 AA batteries through an Ultra G), or maybe Leds have gotten better since 2000, but this is something the CPF hotrodders should be aware of.

The article mentions Nichia by name, although it doesn't state what make led they used for the test. They do state that they tried another brand of 5 mm white leds and got pretty much the same result.

The authors note that this may be a unique characteristic of white leds due to deterioriation of the yellow phosphor layer used more than due to breakdown of the led solid state layer itself.

Anyway, my 2 cents.

I'm hoping that since it is over three years ahead of 2000, that these stats change.


IIRC, the Luxeon class LEDs are longer lived than the 5mm LEDs?

Re: White Leds degrade a lot faster than we think.

Interesting. When overdriven, the white leds take on a blue cast, which means that the yellow layer already isn't operating so well. I wonder if the yellow comes back if you lower the overdrive.

Nope. I have the life testing data from Nichia themselves and see very much the same issue, especially for the NPSW500BS. At 50C they go to south even faster. Under certain conditions, with currents as low as 20mA, you'll see them loose over 35% of their light output by 2000 hours. In fact, they loose 10% of their light output in only 100 hours, at 20mA if it is at all warm around the LED. They will also loose 50% of their light output in only 3,000 hours if in a warm environment.

The degredation of these Nichia LEDs will be rapidly accelerated if you shorten the leads and do not have a good area to sink the heat output from.

Also, if you drive them over 20mA, their life will be greatly accelrated, resulting in a quite rapid degredation.

FYI, this is why I figure some folks have an older version of a flashlight, and get a new one, and say, "wow, this one is so much brighter!"

Ahem.

Re: White Leds degrade a lot faster than we think.

Then where do manufacturers get the 100,000 hour number they all advertise? Are they down right lying?

Re: White Leds degrade a lot faster than we think.

Can good heatsinking help?

Certainly puts the Arc AAA and AA in perspective. Since they are very much over driven (IIRC), they will lose over 10% light output in just 100 hours. /ubbthreads/images/graemlins/mecry.gif

What about the life of Luxeon class white LEDs?? 1 watt, 3 watt?

Re: White Leds degrade a lot faster than we think.

I've been driving a single 5mm led from ebay with several lithium chloride cells for more than 3 weeks straight and it still is as bright as a brand new one driven by the same powerpack.
That's more than 500 hours with no problem but I guess I'm driving it more or less at specs as the batteries only produce 3.6 volts.
I wonder how long will an arc aaa last.

Re: White Leds degrade a lot faster than we think.

Luxeons are a whole different ball of wax.

The Luxeons need to be heatsinked, and it is very important for their life.

A good place to start is here:
http://www.lumileds.com/pdfs/protected/AB07.PDF

Then move on to here:
http://www.lumileds.com/pdfs/protected/AB05.PDF

For the LuxIII
http://www.lumileds.com/pdfs/protected/DS25.PDF

"Lumileds projects that Luxeon products will deliver on
average 70%lumen maintenance at 50,000 hours of operation.This performance is based on independent test data, Lumileds historical data from tests run on similar material systems,and internal Luxeon reliability testing.This projection is based on constant current 350 mA operation with junction
temperature maintained at or below 90 °C."

Keep in mind, this junction temperature is the little die inside the luxeon, and not the slug. There is a thermal resistance between the two. As you push the Luxeon harder, it is very important to keep it cool. Then comes the resistance from the LED slug to the thermal compound, then from the slug to the mounting surface, then the resistance from this to the flashlight body, then from the flashlight body to the ambient air. This all adds up.

http://www.lumileds.com/pdfs/RD25.PDF

"In addition,Luxeon III is expected to provide an
average of 50%lumen maintenance after 20,000 hours
when driven at 1000 mA and the junction temperature is
maintained at or below 90 °C."

As you go up in power, to the Luxeon V (5W), it gets harder and harder to keep the device cool. When the white 5W came out, many customers saw a rapid degredation of the device, which lead them to re-designate the part as a "Portable".

Note:
Luxeon V Portable is designed primarily for
portable lighting and other applications
requiring operating lifetimes of 500 hours
and less. While the device will operate past
500 hours, its lumen maintenance cannot
yet be characterized. Longer life versions
of white Luxeon V will become available in
the future.

(Some of us have been waiting for two years for the improved device to be announced...)

LumiLEDs usually considers lifetime to be degredation to 70% output. Keep in mind, this is with a emitter that is very well heatsinked.

The effects of what happens when an emitter is not properly heatsinked are in this document, look for the graphs:

http://www.lumileds.com/pdfs/RD25.PDF

[ QUOTE ]
357 said:
IIRC, the Luxeon class LEDs are longer lived than the 5mm LEDs?

[/ QUOTE ]

IIRC, the 5mm's output loss is partially a result of UV degredation/darkening/clouding of the epoxy encapsulant, nearest the die - as apparently the blue chips that white LED's are based on, emit some UV. And epoxy - like many plastic polymers - can break down when exposed to UV.

Luxeons, on the other hand, use a UV resistant gel (silicone based?) to immediately encase the die - thus decreasing/eliminating that source of lumen degredation. I think silicones can have greater resistance to thermal breakdown as well.

Re: White Leds degrade a lot faster than we think.

[ QUOTE ]
paulr said:
When overdriven, the white leds take on a blue cast, which means that the yellow layer already isn't operating so well. I wonder if the yellow comes back if you lower the overdrive.

[/ QUOTE ]

I've noticed that very thing, to some degree. I've always assumed it's the combined result of both 1). a non-linear relationship between the phosphor's emmission, and the die's (ie; @ increased drive, the die output increases more than the phosphor output) and 2). "leakage" of the die's blue photons past the phosphor layer.

Just an assumption.

I'm sure there are many here that know the REAL answer, though...

Re: White Leds degrade a lot faster than we think.

Luxeon specs on half-life seem to be a bit more realistic.

The problem with white LEDs seems to be the durability of the phospohr rather than the life of the chip.

Re: White Leds degrade a lot faster than we think.

This is old news to me. I read that paper about a year ago, and I've noticed this phenomenom for myself. It seems to be especially prevalent with the cheaper LEDs from Hong Kong, although it also affects Nichias. I'm using a ChiWing white LED for the power light in my PC case. I plan to compare its brightness in a year to a new LED. I've also noticed that their UV LEDs degrade to half brightness after only a few hundred hours at normal current.

The degradation is caused by many factors and can be mitigated by driving the LEDs at reduced current and heatsinking them. It appears that both heat and current accelerate degradation. Life is an exponential function of drive current, so halving drive current will result in several times the life while overdriving can reduce life to hundreds or even tens of hours. All things being equal, a few thousand hours life isn't horrible considering that one can get white LEDs at $0.25 each these days. All one need do is to compare them with the old grain of wheat bulbs. Those cost a dollar each even twenty years ago, had efficiencies of 2 or 3 lm/W if driven to specs (compare to 15-20 lm/W for even the cheap HK LEDs), and burned out after only a few hundred hours. In fact, I used to overdrive them to get higher efficiency and whiter light, and I was lucky to get ten hours out of them. I'll take today's white LEDs, even with the "disappointing" far less than 100,000 hour life, any day of the week over yesterday's alternatives.

BTW, white LEDs are not the only color which doesn't last the predicted 100,000 hours. The shorter the wavelength the more encapsulant degradation becomes a problem. As I mentioned, UV LEDs don't seem to last more than a few hundred or at most a thousand hours. I've heard blue also fall far short of the claimed 100,000 hours, although by a lesser margin than white because there is no phosphor degradation. Aqua and true green also don't last 100,000 hours. If I had to guess, I might figure lifetimes of 15000, 30000, and 50000 hours for blue, aqua, and true green LEDs, respectively. Anything with longer wavelengths than that does indeed last 100,000 hours or more. In fact, many manufacturers of red LEDs are now claiming lifetimes of 200,000 hours, and I have little basis to doubt this claim.

Re: White Leds degrade a lot faster than we think.

Wow, all interesting. It all kind of suggests 3 alternatives:

1. Go with Luxeons (probably no more than 3W ones for now and 1W ones are probably better) well packaged with good heat dissipation. I was kind of thinking about the $55 Arc clone light on eBay which can be switched down to 200 mA for probably better than Lumileds spec'd lifetime...

2. In 5 mm lights, go for ones with multiple LEDs driven lower rather an over driven single LED lights of the same size and purpose, e.g. haveblue's Xnova AA clone over something like an Ultra G (jtr1972's comments about Chinese LEDs notwithstanding - probably get more life than you give up). Of course - I like my Ultra G a lot for its toughness and wide beam, but I like my Xnovas a lot too. Or..

3. Decide that - especially for those of us with a lot (a LOT) of flashlights - 10's of hours of operation at 80% original efficiency or better probably constitutes a practical lifetime of use and just not worry about it. It kind of reminds me of firearm ads and reviews I saw when I was younger which proclaimed that some gun would easily last a lifetime. They were true, but it turned out that a lifetime just isn't that long for a gun which is shot once every 5 years or so, i.e. that the weakest part of your gun (or flashlight) toy may well be you. Your offspring who care will probably be lighting their way with something generations beyond Luxeons or maybe even LEDs entirely.

Just a thought... /ubbthreads/images/graemlins/sleepy.gif

Re: White Leds degrade a lot faster than we think.

As a result of Newbie's post, I have pasted a very enlightening portion of one of the Luxeon pdf files he recommended for reading.
The text below comes from page 8 of the pdf file at this link location..... http://www.lumileds.com/pdfs/RD25.PDF
Note the last sentence......

General Lumen Maintenance Characteristics
LEDs experience a gradual permanent reduction in light output during operation. This phenomenon is called light output degradation and can either be caused by a reduction in the light generating efficiency of the LED die or a reduction in the light transmission of the optical path within the LED package. In general for the LED die, the rate of light output degradation is higher during the first few hundred hours of operation and then slows down afterwards. Therefore, the light output degradation of the LED die varies approximately as the logarithm of time. In general, the Luxeon is expected to provide an average of 70% lumen maintenance after 50,000 hours provided that the Luxeon is driven at a dc drive current of 350mA for Luxeon I and 700mA for Luxeon III and the junction temperature is maintained at or below 90°C. In addition, Luxeon III is expected to provide an average of 50% lumen maintenance after 20,000 hours when driven at 1000 mA and the junction temperature is maintained at or below 90°C. <font color="red">Note that the human eye is insensitive to small changes in light output and that a change of about 50% is needed in order to create a noticeable change.</font>

Re: White Leds degrade a lot faster than we think.

[ QUOTE ]
luxlover said:
As a result of Newbie's post, I have pasted a very enlightening portion of one of the Luxeon pdf files he recommended for reading.
The text below comes from page 8 of the pdf file at this link location..... http://www.lumileds.com/pdfs/RD25.PDF
Note the last sentence......

General Lumen Maintenance Characteristics
LEDs experience a gradual permanent reduction in light output during operation. This phenomenon is called light output degradation and can either be caused by a reduction in the light generating efficiency of the LED die or a reduction in the light transmission of the optical path within the LED package. In general for the LED die, the rate of light output degradation is higher during the first few hundred hours of operation and then slows down afterwards. Therefore, the light output degradation of the LED die varies approximately as the logarithm of time. In general, the Luxeon is expected to provide an average of 70% lumen maintenance after 50,000 hours provided that the Luxeon is driven at a dc drive current of 350mA for Luxeon I and 700mA for Luxeon III and the junction temperature is maintained at or below 90°C. In addition, Luxeon III is expected to provide an average of 50% lumen maintenance after 20,000 hours when driven at 1000 mA and the junction temperature is maintained at or below 90°C. <font color="red">Note that the human eye is insensitive to small changes in light output and that a change of about 50% is needed in order to create a noticeable change.</font>

[/ QUOTE ]

Yes, until you put two side by side. Then the human eye is very discerning. The lower the light level, the more discerning your eyeball is.

I've ran a test with 20 pilots. I put two units side by side, much like this one I was a part of:

http://www.garmin.com/products/mx20/gallery/image03.html

If I separated them by a good distance, the pilots told me they were the same brightness, even though one was at 200 footLamberts, and the other was at 170 footLamberts. This is a difference of 15%. As I approached 25% difference (150 fL), they were able to pick out which one was brighter, in increasing accuracy, until they were at 100% accuracy at 25% difference.

Now, I took the two units and set them side by side. All 20 pilots were able to pick out the brighter one, in a "dim" lab, and in full August noontime sunlight.

One guy could even pick out a difference of only 5%, when they were side by side.


luxlover, by your reasoning, the up and comming ARC5 needs to be 50% brighter, otherwise why make it, because you'd never notice the difference...

[ QUOTE ]
nerdgineer said:Maybe 200 hours is enough (20 AA batteries through an Ultra G), or maybe Leds have gotten better since 2000

[/ QUOTE ]

Actually it has not changed AFAIK and I mentioned this report when it came to the public. We discussed it a lot and it was more of a problem 4 or 5 years ago when white LEDs were much more expensive.
At this time (maybe still now) you could tell the more honest manufacturers from the rest.
The first ones used to drive the LEDs at 20mA or lower, they other ones used higher currents and got less efficiency and less lifetime.
When selling lights, many people just ask for brightness. so many lights are optimized for brightness at the selling time.
I've sold about 50 LuxBrites and they run their LEDs at ~17mA each.
And people compared it with cheaper unregulated '3 AAs and a fistful of LEDs' lights.
As they main users were/are cavers who use their lights many hundreds of hours this makes a difference.
I guestimate that I approach 1000 hours with my main LED caving light now.

If flashlight manufacturers made the LED a replaceable part, then the LED lifetime would be far less important. The new Photon Freedom light has its LED installed in a socket which would make it pretty easily replaceable, but most other flashlights would have to be either junked or skillfully modded when the LED gives out.

Even the very inexpensive $6 Dorcy AAA would be worth installing a new LED in vs. just throwing the flashlight out. I'd like to see sockets used more. It would also make the flashlights more upgradeable as LED technology advances.

The point of Lumen maintenance is a good one. The Lumileds ones last longer than others for a couple of reasons:
- TG and Nichia use UV to activate the phos - good for color, but very bad for lifetime. Ever see the effect of UV on plastic ?
- Lumileds uses blue + Phos - and special phos, for longer life.
- Heat sinking - ever wonder how the heat gets out of a 5mm LED - down the wire.

Lifetime is also an issue for simple colors like blue, except for a few brands for the same reason.

There is no way I would pay big $s for a flashlight with any LED except a Lumileds, for the obvious stated reasons - if you actually use the light, it will degrade with time.

Re: White Leds degrade a lot faster than we think.

[ QUOTE ]
HarryN said:
The point of Lumen maintenance is a good one. The Lumileds ones last longer than others for a couple of reasons:
- TG and Nichia use UV to activate the phos - good for color, but very bad for lifetime. Ever see the effect of UV on plastic ?
- Lumileds uses blue + Phos - and special phos, for longer life.
- Heat sinking - ever wonder how the heat gets out of a 5mm LED - down the wire.

Lifetime is also an issue for simple colors like blue, except for a few brands for the same reason.

There is no way I would pay big $s for a flashlight with any LED except a Lumileds, for the obvious stated reasons - if you actually use the light, it will degrade with time.

[/ QUOTE ]

I sorry, but I believe most of this post is inaccurate. While UV could be used, I believe most manufacturers use blue. It was Nichia that first invented a high power blue and introduce this form of white if I'm not mistaken.
Yes, the shorter the wavelength the faster it will affect epoxy. And the thermal stress of the epoxy on the wire bonds could contribute to failure as well. Some of the heat of epoxy encased LEDs is absorbed and radiated by the epoxy but this is one of the limiting factors in how much power can be put in a single one of these.

Lumiled has addressed the issue of power in a single case with their improved heatsinking design. And reduced the wirebond stress by having the lens optically coupled to the LED with a silicone compound that also allows a higher operating temp. Have not seen anywhere were they claim an improved phosphor. Though with they have improved the distribution of it. In fact it appears the phophor is the likely reason the Luxeon V portable (white) has such a short life!

Re: White Leds degrade a lot faster than we think.

There is one error, and that is that TG and Nichia are trending tending toward the UV activated for their advanced products, as opposed to being fully there, and most Nichia's are currently blue activated.

I can assure you that traditional epoxy fillers are not able to handle (well) what Lumileds sells as royal blue (around 440 nm), and especially the 400 nm range versions that TG and Nichia are heading into. Only a few silicones are able to resist this chemical scission induced breakdown for a period, and most will at least yellow.

Heat is also a problem, which is a significant factor in the 5 watt white - no doubt. As other firms start to release their 1 watt products (like Nichia), they are well aware of the challenges.

There are plenty of reliability studies presented at conferences on LED (not just white, but also blue and green) reliability (lumen maintenance), and only a few firms are in the game.

If you doubt it, just ask someone from Singapore about their Country's experience with LED stop lights.