5mm epoxy white LED longevity?

[JohnR66]

I just bought a batch of 5mm white LEDs. They are supposed to be quality
units as claimed by the supplier. Considering no abuse of them, I wonder the
typical lifetime and failure modes of these LEDs? One thing is certain, they
are far brighter than ones I purchased 4 years ago.

John

 

[Clive Mitchell]

I just bought a batch of 5mm white LEDs. They are supposed to be quality
units as claimed by the supplier. Considering no abuse of them, I wonder the
typical lifetime and failure modes of these LEDs? One thing is certain, they
are far brighter than ones I purchased 4 years ago.

If they are cheap Chinese units they will lose brightness significantly
in a short time, assuming they don't suffer film breakdown and start
flickering before going dim or out completely.

If they are from a reputable manufacturer like Nichia or Cree then you
can probably look forward to a long and useful working life.

The claims of many Chinese suppliers (often claiming to be
manufacturers) can be taken with a pinch of salt.

20,000mcd 5mm LEDs for instance.........
(Yeah, perhaps if you stick them across the frickin' mains!)

 

[JohnR66]

If they are cheap Chinese units they will lose brightness significantly in
a short time, assuming they don't suffer film breakdown and start
flickering before going dim or out completely.

If they are from a reputable manufacturer like Nichia or Cree then you can
probably look forward to a long and useful working life.

The claims of many Chinese suppliers (often claiming to be manufacturers)
can be taken with a pinch of salt.

20,000mcd 5mm LEDs for instance.........
(Yeah, perhaps if you stick them across the frickin' mains!)

I got them from superbrightleds.com. They claim these are quality units, but
anybody can make a claim. I plan to torture test a couple of them. One will
be driven at 40ma (rated 30ma absolute max) static test and reverse voltage.
For the hell of it, I may even microwave one : )
John

 

[JohnR66]

LEDs come on a lot in 4 years, also 4 years of reasonable use will have
seen intial light output drop considerably.

Luxeon 1W white with much better thermal management and silicone rather
than epoxy encapsulant , Lumileds have quoted 70% of intial output at
50K Hours.

5mm LEDs have some problems with epoxy degradation, phosphor
degradation and lowering output of the blue die at the back. Personal
opinion would place useful lifespan of a decent quality white 5mm LED
nearer 20K hours.

That is without the random failures Clive has mentioned.

Adam

20K hours is more than adequate for me if this is at continuous max current.
John

 

[Clive Mitchell]

Clive has previously suggested testing LEDs with extremely low
currents, microamps, in general those that will light with a lowest
current first, show best reliabilty in full current service.

To elaborate on that. I found that you can test Gallium Nitride LEDs
with a very low current like 75uA which will cause a good LED to glow
slightly while an LED with a film fault (manufacturing or static damage)
will have enough resistance to pull the voltage down to a point where
the LED will not light.

I'm not sure how infallible this technique is, but it seems to work OK
for me. I use a little tester that has a PP3 9V battery and a switch
with two resistors feeding a standard miniature Molex socket that the
LEDs can be plugged into for testing at either 75uA or 20mA.

 

[JohnR66]

To elaborate on that. I found that you can test Gallium Nitride LEDs with
a very low current like 75uA which will cause a good LED to glow slightly
while an LED with a film fault (manufacturing or static damage) will have
enough resistance to pull the voltage down to a point where the LED will
not light.

I'm not sure how infallible this technique is, but it seems to work OK for
me. I use a little tester that has a PP3 9V battery and a switch with two
resistors feeding a standard miniature Molex socket that the LEDs can be
plugged into for testing at either 75uA or 20mA.

Thanks for the tip. I tested the LEDs at 50 and 5 uA. At 50 microamps, they
lit up all about the same brightness. At 5, they lit up in varying degrees
of brightness, but they all lit.
What do you mean by film? The junction area?
John

John

 

[Don Klipstein]

I just bought a batch of 5mm white LEDs. They are supposed to be quality
units as claimed by the supplier. Considering no abuse of them, I wonder the
typical lifetime and failure modes of these LEDs? One thing is certain, they
are far brighter than ones I purchased 4 years ago.

By-and-large, better quality white LEDs have claims to last 50,000
hours, and I think that's when operated at "characterization current"
(typically 20 mA for 5 mm units) and favorable temperature (25 C ambient
with the LED mounted in a not-worse-than-average situation for temperature
rise).

One study done in 2002 is cited by Lumileds to say that 5 mm white LEDs
(which Lumileds competes against) fade to half their initial brightness at
6,000 hours. Lumileds does not state the test conditions in that study.
So I think they should usually do at least that well unless they are junky
cheapies.
http://www.lumileds.com/pdfs/AB07.PDF

I have had some white LEDs of various sizes fade at a rate where
halflife would be half a year to a year. I have had a nightlight with
some apparently cheapish ones fade with an "eyeball estimate" halflife of
about or maybe a bit under 6 months.

Then again, there are now pedestrian walk signals with white LEDs, and
if there is a big life expectancy problem there then some slop is going to
hit the fan!

Meanwhile, if you need a white LED to last many many years, I would
suggest considering underpowering it.

As for failure mode: The usual main problem with most white LEDs is
that the phosphor degrades. Most non-white LEDs do not have a phosphor
and have a longer life expectancy.

I have heard of complaints of a few worse cheapie junk Chinese LEDs
just conking out or flickering and then conking out, and I have had a
couple conk out (failing open) before I had them so much as 6 months or 50
operating hours. I do not know who made the ones that conked out on me.

- Don Klipstein ([email protected])

 

[Victor Roberts]

I just bought a batch of 5mm white LEDs. They are supposed to be quality
units as claimed by the supplier. Considering no abuse of them, I wonder the
typical lifetime and failure modes of these LEDs? One thing is certain, they
are far brighter than ones I purchased 4 years ago.

The main failure mechanism, at least on properly
manufactured LEDs, is going to be lumen depreciation. The
LED may operate for 100,000 hours or perhaps even 200,000
hours, but will not be producing a useful amount of light.
You need to decide what percentage of initial light is
acceptable for your application and then calculate the
useful life of the LEDs used in that application from the
lumen depreciation curve. However, I suspect you don't
have a lumen depreciation curve since, as far as I know, no
LED manufacturer publishes one. Some of the better LED
manufacturers do give relative output at one point. For
example, some Luxeon LEDs are rated for 70% of initial
lumens at 50,000 hours while other Luxeon LEDs operated at
higher current are rated for 50% of initial light at 20,000
hours. Neither Lumileds nor any other LED manufacturer
publishes a full lumen depreciation curve so we can't
calculate mean lumens.

Lumileds does shows the early part of a lumen depreciation
curve for one of their Luxeon devices plotted against 5mm
devices measured by the LRC many years ago. The paper on
that original LRC experiment is still available. Go to:

http://www.lrc.rpi.edu/programs/solidstate/SSLLRCAuthored.htm

and scroll down to year 2000 and click on the link to
Narendran, N., N. Maliyagoda, A. Bierman, R. Pysar, and M.
Overington. (The only paper on this list for year 2000.)
Then see Figure 10. As you can see the data is quite old,
considering the advances made in LED technology over the
past 6 years.

My opinion is that if the supplier will not provide you with
a lumen depreciation curve for their devices or at least the
percent output after some number of hours, such as 20K or
50K, then you are taking a very big risk using these LEDs
for any commercial project. They also need to provide you
with minimum guaranteed output, not just typical, and give
you the output vs. junction temperature curve and the
thermal resistance between junction and case before they can
be used in any serious commercial project. Lumileds
provides this data, and other quality manufacturers such as
Cree and Nichia provide at least minimum output data, output
vs. Tj and thermal resistance from junction to case. Many
other LED manufacturers provide only typical output at 25C.
This data alone is useless for serious applications.

--
Vic Roberts
http://www.RobertsResearchInc.com
To reply via e-mail:
replace xxx with vdr in the Reply to: address
or use e-mail address listed at the Web site.

This information is provided for educational purposes only.
It may not be used in any publication or posted on any Web
site without written permission.

 

[Victor Roberts]

By-and-large, better quality white LEDs have claims to last 50,000
hours, and I think that's when operated at "characterization current"
(typically 20 mA for 5 mm units) and favorable temperature (25 C ambient
with the LED mounted in a not-worse-than-average situation for temperature
rise).

One study done in 2002 is cited by Lumileds to say that 5 mm white LEDs
(which Lumileds competes against) fade to half their initial brightness at
6,000 hours. Lumileds does not state the test conditions in that study.
So I think they should usually do at least that well unless they are junky
cheapies.
http://www.lumileds.com/pdfs/AB07.PDF

I do not believe that the 2002 paper referenced by Lumileds
is available on line. However, I think the 2002 paper is
based on or similar to an earlier experiment run at the LRC
that is available in PDF format. Go to
http://www.lrc.rpi.edu/programs/solidstate/SSLLRCAuthored.htm

and scroll down to year 2000. The 2002 paper is listed but
a download link is not provided. Also, IES papers are
rather hard to find on line.

--
Vic Roberts
http://www.RobertsResearchInc.com
To reply via e-mail:
replace xxx with vdr in the Reply to: address
or use e-mail address listed at the Web site.

This information is provided for educational purposes only.
It may not be used in any publication or posted on any Web
site without written permission.

 

[Clive Mitchell]

Thanks for the tip. I tested the LEDs at 50 and 5 uA. At 50 microamps,
they lit up all about the same brightness. At 5, they lit up in varying
degrees
of brightness, but they all lit.
What do you mean by film? The junction area?

I believe the Gallium Nitride LEDs have a very thin film of material on
a substrate that makes them a bit sensitive to static damage.

 

[Don Klipstein]

I believe the Gallium Nitride LEDs have a very thin film of material on
a substrate that makes them a bit sensitive to static damage.

Gallium nitride LEDs are very intolerant of reverse breakdown, even at
very low currents in the microamps. I suspect some sort of electrolysis
mechanism causes a partial short (parallel resistance) in the area where
the breakdown occurred. I have seen GaN and InGaN LEDs apparently damaged
by static that do not glow at all until a few or several milliamps flow
through.

- Don Klipstein ([email protected])

 

[Clive Mitchell]

I have seen GaN and InGaN LEDs apparently damaged by static that do not
glow at all until a few or several milliamps flow through.

You mean like the ones that used to be supplied by ebay trader "Chi
Wing" that came in an ordinary plastic bag and were somewhat unreliable.

"I am genuine manufacturer of LEDs...." Uh huh? That's why you don't
use any anti-static procedures?

 

[Don Klipstein]

You mean like the ones that used to be supplied by ebay trader "Chi
Wing" that came in an ordinary plastic bag and were somewhat unreliable.

"I am genuine manufacturer of LEDs...." Uh huh? That's why you don't
use any anti-static procedures?

I have seen 4 Nichia LEDs with partial shorts. Two of them I even made
that way. One that I made that way got that way from handling without
adequate defense against static. The other had current-limited
rectified-filtered 120VAC applied in the wrong polarity.
Of the other two, one was known to be abused by a previous owner and the
other I probably zapped with static after unpacking.

- Don Klipstein ([email protected])

 

[Clive Mitchell]

I have seen 4 Nichia LEDs with partial shorts. Two of them I even
made that way. One that I made that way got that way from handling
without adequate defense against static. The other had current-limited
rectified-filtered 120VAC applied in the wrong polarity.
Of the other two, one was known to be abused by a previous owner and
the other I probably zapped with static after unpacking.

I'm suspicious that I may even have nuked some cheap LEDs with the
leakage from a mains soldering iron. They worked before they were
soldered in and I didn't dilly dally with the soldering iron.