What are the failure modes of high power LEDs?

Do they fail open or shorted when operating temperature is exceeded?

How about a case of a number of them connected in series, and the
current rating is exceeded? Do they fail open, thereby potentially
saving a chain reaction or do they fail shorted?

Thanks,

-Martin

What are the failure modes of high power LEDs?

Do they fail open or shorted when operating temperature is exceeded?

How about a case of a number of them connected in series, and the
current rating is exceeded? Do they fail open, thereby potentially
saving a chain reaction or do they fail shorted?

It's probably safest to assume they will fail shorted. Different brands
and their manufacturing processes may have different outcomes.

It sounds like I need to do some destructive testing once we decide on
a device. I am trying to understand what the circuit might have to
survive if an LED fails shorted. Opens are not a big deal as all
current stops flowing. However, a short means that the Vf of that
diode just went to somewhere between 0 and Vf. A linear constant
current driver would have to deal with this. This is, in principle,
not a problem, but there are components that all of a sudden might need
to dissipate a lot more heat. It might pay to try to detect a shorted
LED and simply have the microprocessor shut the thing down.

Much safer to assume the worst case of a short circuit and allow for
that.

Here's a rather grim example of one of the first uses of Chinese white
LEDs when they were just getting to grips with the technology.

Much safer to assume the worst case of a short circuit and allow for
that.

The problem is that I'll have 10 to 12 high power LEDs in series. I
wouldn't want to allow for any more than one shorted LED before
calling for a shutdown. The application cannot tolerate even one bad
LED anyway.

Here's a rather grim example of one of the first uses of Chinese white
LEDs when they were just getting to grips with the technology.

Which raises another question. My application requires anywhere from
500 to 1000 Seoul Semiconductors P4 LEDs. Not a single failure is
tolerable. I think we need to be very careful about thermal
management in order to ensure that they survive. I've been thinking
about a liquid-cooled heatsink plate for this one.

-M

Much safer to assume the worst case of a short circuit and allow for
that.

The problem is that I'll have 10 to 12 high power LEDs in series. I
wouldn't want to allow for any more than one shorted LED before
calling for a shutdown. The application cannot tolerate even one bad
LED anyway.

Here's a rather grim example of one of the first uses of Chinese white
LEDs when they were just getting to grips with the technology.

Which raises another question. My application requires anywhere from
500 to 1000 Seoul Semiconductors P4 LEDs. Not a single failure is
tolerable. I think we need to be very careful about thermal
management in order to ensure that they survive. I've been thinking
about a liquid-cooled heatsink plate for this one.

-M

I think you should have a very serious look at the subject of
electronic devices/eqipment reliability prediction and get some
understanding of the subject before you go any further.

Then come back to the group with any questions you (almost certainly
will) have.- Hide quoted text -

Seoul P4 uses the Cree EZ Bright dice so pull the data sheets ther, its
laso protected with ESD diodes on the die, my own experience is if you
abuse them enough with over current the internal bond wires melt.

We've built a rig with 250 of these LEDs already. No failures
whatsoever. That's why I was asking. I don't know how these things
fail. We tried convection and forced-air cooling with good results. We
have a thermistor every 10 LEDs built into the circuit boards. They are
all mounted on metal clad PCB's thermally coupled to a 1/4 aluminum
plate with thin fins welded onto the back. It's a pretty amazing rig
to look at (well, you don't really want to, directly at least). I
can't wait to build the 1000+ LED version. The small unit does not
have any protection mechanism whatsoever. I'll design-in some of this
into the new boards.

Are the power supplies for the LEDs switch-mode current regulators?

Overall efficiency might be about 90% for the AC to DC SMPS and 95% for
our custom LED driver, for a total of 85.5% from AC to LED. We could
improve upon the efficiency of the linear LED driver circuit, but it
would require much tighter tolerances for all system components. I'm
not sure that this is worth considering once you take into account
component tolerances, thermal range as well as needing some headroom to
with possible failure modes. Like a lot about this field, there's a
lot to be learned.

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.

-Martin

--
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.

I think you should have a very serious look at the subject of
electronic devices/eqipment reliability prediction and get some
understanding of the subject before you go any further.

Then come back to the group with any questions you (almost certainly
will) have.- Hide quoted text -

Nice thought. If I had the time, I would. There's a reason we test
fly airplane designs. When all the smoke and bullshit clears out, you
need to find out what happens in the real world. I could model this
thing to death with FEA tools and I'd still have to build it. :-)

We've built a rig with 250 of these LEDs already. No failures
whatsoever. That's why I was asking. I don't know how these things
fail.
snip