Main Page | Report this Page
 
Loading..
   
Science Forum Index  »  Engineering - Lighting Forum  »  LED failure mode
Page 2 of 2    Goto page Previous  1, 2
Author Message
Victor Roberts
Posted: Mon Mar 05, 2007 11:51 pm
Guest
On 5 Mar 2007 14:26:22 -0800, "RHRRC"
<h.lewis@connect-2.co.uk> wrote:

Quote:
On 4 Mar, 00:28, Victor Roberts <x...@lighting-research.com> wrote:
On 3 Mar 2007 10:05:46 -0800, "m" <martin.use...@gmail.com
wrote:

What are the failure modes of high power LEDs?

They usually don't fail in the traditional sense, but their
light output decreases over time to a point where they are
of little value.

"They usually don't fail in the traditional sense"?

Wouln't it be nice if that were true.

Do you mean that as well as failing in the traditional sense they have
an extra failure mode over other discrete semiconductors in that they
also fade away.

Most discrete semiconductors do not generate light. What I
mean is that the practical life of an LED is determined by
the loss of light output which occurs long before 99.9% of
them will have failed due to shorts or opens.

--
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.
Back to top
m
Posted: Tue Mar 06, 2007 2:30 am
Guest
Quote:
have a look at an old Agilent (now Avago) note "Application Brief
I-004" and have a serious read of these led-specific failure
mechanisms and note that these do not include other of the basic
semiconductor technology relevant failure mechanisms.

Would you happen to have a link to this document. I tried to search
and didn't get very far. I'd definitely like to read it.

Thanks,

-Martin
Back to top
RHRRC
Posted: Tue Mar 06, 2007 6:28 pm
Guest
On 6 Mar, 06:30, "m" <martin.use...@gmail.com> wrote:
Quote:
have a look at an old Agilent (now Avago) note "Application Brief
I-004" and have a serious read of these led-specific failure
mechanisms and note that these do not include other of the basic
semiconductor technology relevant failure mechanisms.

Would you happen to have a link to this document. I tried to search
and didn't get very far. I'd definitely like to read it.

Thanks,

-Martin

the agilent/avago I-004 is quite old but does indicate some of the
failure mechanisms of 5mm leds to give you a feelfor the subject.

Application Brief I-004 - Reliability of Precision Optical Performance
AlInGaP LED Lamps in Traffic Signals and Variable Message Signs.
http://literature.agilent.com/litweb/pdf/5965-8097E.pdf

Lumileds gives some more relevant
http://www.lumileds.com/pdfs/RD25.PDF

these are also instructive

http://www.dfrsolutions.com/Information/MAE%20LED%20article.pdf

http://mae.pennnet.com/Articles/Article_Display.cfm?Section=Articles&ARTICLE_ID=230223&VERSION_NUM=2&p=32

Cree have a note here

http://www.cree.com/products/pdf/XLamp_Reliability.PDF


Whatever you must try and get some data on the P4's and cast a very
critical (cynical) eye over it
for example
impressively Cree cite, for the RTOL (room temp operating life test),
testing to JESD22 Method A108-B.
included in the critereon for failure to this test is a >10%
change in ouput at 25degC within 1008 hours.
Wow. In near perfect conditions repeating this test 8 times
sequentially means a greater than 50% reduction in output
in 1 year is acceptable under this criteria
Ow!!

the reputable companies dont lie, many of the el-cheapo far eastern
companies do.
How relevant the information provided is to your application only you
can decide.

Hope this helps

Good luck with the project
Back to top
RHRRC
Posted: Tue Mar 06, 2007 6:42 pm
Guest
On 6 Mar, 03:51, Victor Roberts <x...@lighting-research.com> wrote:
Quote:
On 5 Mar 2007 14:26:22 -0800, "RHRRC"





h.le...@connect-2.co.uk> wrote:
On 4 Mar, 00:28, Victor Roberts <x...@lighting-research.com> wrote:
On 3 Mar 2007 10:05:46 -0800, "m" <martin.use...@gmail.com
wrote:

What are the failure modes of high power LEDs?

They usually don't fail in the traditional sense, but their
light output decreases over time to a point where they are
of little value.

"They usually don't fail in the traditional sense"?

Wouln't it be nice if that were true.

Do you mean that as well as failing in the traditional sense they have
an extra failure mode over other discrete semiconductors in that they
also fade away.

Most discrete semiconductors do not generate light. What I
mean is that the practical life of an LED is determined by
the loss of light output which occurs long before 99.9% of
them will have failed due to shorts or opens.

--

The fellow has indicated a use of 1000leds per product.
Using a degradation figure commonly banded about of 30% output
degradation in 50,000 hours:
If this is the predominant failure mechanism (and a 30% drop in output
is deemd a fail) then the rest of the assembly must have a FIT of
below 20!!
Spread across at least 2000 soldered loints and 1000 parts - let alone
anything else - this is impressive.





Quote:
Vic Robertshttp://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.
Back to top
Victor Roberts
Posted: Tue Mar 06, 2007 7:51 pm
Guest
On 6 Mar 2007 14:42:50 -0800, "RHRRC"
<h.lewis@connect-2.co.uk> wrote:

Quote:
On 6 Mar, 03:51, Victor Roberts <x...@lighting-research.com> wrote:
On 5 Mar 2007 14:26:22 -0800, "RHRRC"





h.le...@connect-2.co.uk> wrote:
On 4 Mar, 00:28, Victor Roberts <x...@lighting-research.com> wrote:
On 3 Mar 2007 10:05:46 -0800, "m" <martin.use...@gmail.com
wrote:

What are the failure modes of high power LEDs?

They usually don't fail in the traditional sense, but their
light output decreases over time to a point where they are
of little value.

"They usually don't fail in the traditional sense"?

Wouln't it be nice if that were true.

Do you mean that as well as failing in the traditional sense they have
an extra failure mode over other discrete semiconductors in that they
also fade away.

Most discrete semiconductors do not generate light. What I
mean is that the practical life of an LED is determined by
the loss of light output which occurs long before 99.9% of
them will have failed due to shorts or opens.

--

The fellow has indicated a use of 1000leds per product.
Using a degradation figure commonly banded about of 30% output
degradation in 50,000 hours:
If this is the predominant failure mechanism (and a 30% drop in output
is deemd a fail) then the rest of the assembly must have a FIT of
below 20!!
Spread across at least 2000 soldered loints and 1000 parts - let alone
anything else - this is impressive.

I don't know where you get 30% at 50,000 hours. That much
degradation would make the LEDs virtually useless by then.
Lumileds uses 70% at 50,000 for most of their LEDs and 50%
for certain very high power LEDs.


--
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.
Back to top
Adam Aglionby
Posted: Tue Mar 06, 2007 8:54 pm
Guest
On 6 Mar, 23:51, Victor Roberts <x...@lighting-research.com> wrote:
Quote:
On 6 Mar 2007 14:42:50 -0800, "RHRRC"



h.le...@connect-2.co.uk> wrote:
On 6 Mar, 03:51, Victor Roberts <x...@lighting-research.com> wrote:
On 5 Mar 2007 14:26:22 -0800, "RHRRC"

h.le...@connect-2.co.uk> wrote:
On 4 Mar, 00:28, Victor Roberts <x...@lighting-research.com> wrote:
On 3 Mar 2007 10:05:46 -0800, "m" <martin.use...@gmail.com
wrote:

What are the failure modes of high power LEDs?

They usually don't fail in the traditional sense, but their
light output decreases over time to a point where they are
of little value.

"They usually don't fail in the traditional sense"?

Wouln't it be nice if that were true.

Do you mean that as well as failing in the traditional sense they have
an extra failure mode over other discrete semiconductors in that they
also fade away.

Most discrete semiconductors do not generate light. What I
mean is that the practical life of an LED is determined by
the loss of light output which occurs long before 99.9% of
them will have failed due to shorts or opens.

--

The fellow has indicated a use of 1000leds per product.
Using a degradation figure commonly banded about of 30% output
degradation in 50,000 hours:
If this is the predominant failure mechanism (and a 30% drop in output
is deemd a fail) then the rest of the assembly must have a FIT of
below 20!!
Spread across at least 2000 soldered loints and 1000 parts - let alone
anything else - this is impressive.

I don't know where you get 30% at 50,000 hours. That much
degradation would make the LEDs virtually useless by then.
Lumileds uses 70% at 50,000 for most of their LEDs and 50%
for certain very high power LEDs.

Way I read it was 30% degradation, 70% brightness left. Though these
are accelerated life tests so real world experience may be different
depending on application. Thats also Lumileds predictions with soft
gel lens I think, where as other manufactures trying different
solutions. Remember the early 5mm whites fading due to resin
yellowing.

Adam


Quote:

--
Vic Robertshttp://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.
Back to top
Victor Roberts
Posted: Tue Mar 06, 2007 10:12 pm
Guest
On 6 Mar 2007 16:54:38 -0800, "Adam Aglionby"
<ledlight@gmail.com> wrote:

Quote:

The fellow has indicated a use of 1000leds per product.
Using a degradation figure commonly banded about of 30% output
degradation in 50,000 hours:
If this is the predominant failure mechanism (and a 30% drop in output
is deemd a fail) then the rest of the assembly must have a FIT of
below 20!!
Spread across at least 2000 soldered loints and 1000 parts - let alone
anything else - this is impressive.

I don't know where you get 30% at 50,000 hours. That much
degradation would make the LEDs virtually useless by then.
Lumileds uses 70% at 50,000 for most of their LEDs and 50%
for certain very high power LEDs.

Way I read it was 30% degradation, 70% brightness left.

You are correct. I had misread RHRRC's post.

--
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.
Back to top
m
Posted: Wed Mar 07, 2007 2:27 am
Guest
Quote:
The fellow has indicated a use of 1000leds per product.
Using a degradation figure commonly banded about of 30% output
degradation in 50,000 hours:
If this is the predominant failure mechanism (and a 30% drop in output
is deemd a fail) then the rest of the assembly must have a FIT of
below 20!!
Spread across at least 2000 soldered loints and 1000 parts - let alone
anything else - this is impressive.

You might want to refer to my original post.

I am not concerned with light output degradation after so many hours.
The failure modes I am looking for might be classified as
catastrophic. In other words, it stops working 'cause it let the
smoke out or something else happened. With 1,000 LEDs I am concerned
about anything that could cause a chain reaction.

An LED that fails open is benign. One that fails shorted causes an
increase in heat load to the driver components. If more than one fail
shored (in the same series connected string) failure of the drive
components could be precipitated. This is what I am interested in
understanding.

-Martin
Back to top
 
Page 2 of 2    Goto page Previous  1, 2   All times are GMT - 5 Hours
The time now is Mon Oct 13, 2008 5:27 am
Science Forum Index

Powered by phpBB & ML | Disclaimer



Contact Us