Cree MC-E Catastrophic failure.

[John Tserkezis]

I have a Cree XLamp MC-E that suffered a catastrophic failure (open
circuit, and the clear reflector has come off the die).

I haven't been able to get a straight answer out of any googling or
datasheet perusing, so I'm asking here. The reliability spec sheet
absolutely abuses the leds before failure, from what I can guess, more
than I have here, so I'm at a a loss.

It lives in a bicycle light, I have the star led die bolted to an
aluminium block about 1cm thick, within a PVC pipe section, and a
further heatsink bolted to the other side of that.

All applicable surfaces were honed, (not polished, but smooth and
flat), and heatsink compound applied. The only surface I did not touch
was between the actual LED die, and the aluminium star surface it's
mounted on - that was left "factory standard".



My first test was in a still room (ambient was about 16C (60F) or so)
after about 15-20 minutes, the temperature probe (positioned near the
block) rose to about 65-70C (150F) before I shut it down. The heatsink
was quite warm to the touch at that point. Not _quite_ hot enough to
pull my hand away, but nearly there.

Since it's mostly sealed, I didn't have opportunity to measure lead
temperature, or anywhere near the actual die star for that matter.

On the road, it ran for a solid 2.5 hours or so absolutely perfectly,
with the heatsink temperature barely at "warm to the touch". (I didn't
have a temperature probe on it on the road).



The failure occurred the next day when I was demonstrating the light
to friends. It ran for several seconds before I moved the bike a bit,
where it failed. I had thought it was a wiring issue, till I measured
and then dismantled the assembly (it was O/C) to find bubbles in the
MC-E clear lens (mostly around the emitter output edges), and found the
lens came off quite easily with my finger.



Did the initial 'test' kill or stress it? Should I have a larger
heatsink? I am most certainly not the first person to use this led, so
what went wrong here?

Again, I didn't have access to the die temperature, and since it's
mounted on the star, how does one measure that anyway?
--
The ballot is stronger than the bullet.

 

[Cydrome Leader]

John Tserkezis <jt@techniciansyndrome.org.invalid> wrote:

I have a Cree XLamp MC-E that suffered a catastrophic failure (open
circuit, and the clear reflector has come off the die).

I haven't been able to get a straight answer out of any googling or
datasheet perusing, so I'm asking here. The reliability spec sheet
absolutely abuses the leds before failure, from what I can guess, more
than I have here, so I'm at a a loss.

It lives in a bicycle light, I have the star led die bolted to an
aluminium block about 1cm thick, within a PVC pipe section, and a
further heatsink bolted to the other side of that.

All applicable surfaces were honed, (not polished, but smooth and
flat), and heatsink compound applied. The only surface I did not touch
was between the actual LED die, and the aluminium star surface it's
mounted on - that was left "factory standard".



My first test was in a still room (ambient was about 16C (60F) or so)
after about 15-20 minutes, the temperature probe (positioned near the
block) rose to about 65-70C (150F) before I shut it down. The heatsink
was quite warm to the touch at that point. Not _quite_ hot enough to
pull my hand away, but nearly there.

Since it's mostly sealed, I didn't have opportunity to measure lead
temperature, or anywhere near the actual die star for that matter.

On the road, it ran for a solid 2.5 hours or so absolutely perfectly,
with the heatsink temperature barely at "warm to the touch". (I didn't
have a temperature probe on it on the road).



The failure occurred the next day when I was demonstrating the light
to friends. It ran for several seconds before I moved the bike a bit,
where it failed. I had thought it was a wiring issue, till I measured
and then dismantled the assembly (it was O/C) to find bubbles in the
MC-E clear lens (mostly around the emitter output edges), and found the
lens came off quite easily with my finger.



Did the initial 'test' kill or stress it? Should I have a larger
heatsink? I am most certainly not the first person to use this led, so
what went wrong here?

Again, I didn't have access to the die temperature, and since it's
mounted on the star, how does one measure that anyway?

how were you driving this LED?

 

[John Tserkezis]

Cydrome Leader wrote:

how were you driving this LED?

A pre-built module that supplies 700mA constant current. Ends up
being about 12.something volts for that LED module which looks within spec.

And I've checked, it works as advertised.
--
Make like a bottom and split.

 

[Uwe Hercksen]

John Tserkezis schrieb:

I have a Cree XLamp MC-E that suffered a catastrophic failure (open
circuit, and the clear reflector has come off the die).

I haven't been able to get a straight answer out of any googling or
datasheet perusing, so I'm asking here. The reliability spec sheet
absolutely abuses the leds before failure, from what I can guess, more
than I have here, so I'm at a a loss.

It lives in a bicycle light, I have the star led die bolted to an
aluminium block about 1cm thick, within a PVC pipe section, and a
further heatsink bolted to the other side of that.

All applicable surfaces were honed, (not polished, but smooth and
flat), and heatsink compound applied. The only surface I did not touch
was between the actual LED die, and the aluminium star surface it's
mounted on - that was left "factory standard".



My first test was in a still room (ambient was about 16C (60F) or so)
after about 15-20 minutes, the temperature probe (positioned near the
block) rose to about 65-70C (150F) before I shut it down. The heatsink
was quite warm to the touch at that point. Not _quite_ hot enough to
pull my hand away, but nearly there.

Since it's mostly sealed, I didn't have opportunity to measure lead
temperature, or anywhere near the actual die star for that matter.

On the road, it ran for a solid 2.5 hours or so absolutely perfectly,
with the heatsink temperature barely at "warm to the touch". (I didn't
have a temperature probe on it on the road).



The failure occurred the next day when I was demonstrating the light to
friends. It ran for several seconds before I moved the bike a bit,
where it failed. I had thought it was a wiring issue, till I measured
and then dismantled the assembly (it was O/C) to find bubbles in the
MC-E clear lens (mostly around the emitter output edges), and found the
lens came off quite easily with my finger.

Hello,

the heatsink worked well when you rode the bike, enough airflow at the
heatsink. But when you did your first test in a still room, there was no
forced airflow at the heatsink and the airflow caused by the temperature
differences, the thermal convection, was not enough.

The heatsink had some fins, I suppose. But what direction had these
fins, horizontal fins would be perfect when riding the bike, but in the
still room vertical fins would be better.

If you do another test with a new LED in a still room, you should use a
ventilator to simulate the forced airflow when riding the bike.

Bye

 

[John Tserkezis]

Uwe Hercksen wrote:

the heatsink worked well when you rode the bike, enough airflow at the
heatsink. But when you did your first test in a still room, there was no
forced airflow at the heatsink and the airflow caused by the temperature
differences, the thermal convection, was not enough.

The heatsink had some fins, I suppose. But what direction had these
fins, horizontal fins would be perfect when riding the bike, but in the
still room vertical fins would be better.

If you do another test with a new LED in a still room, you should use a
ventilator to simulate the forced airflow when riding the bike.

The original heatsink I had calculated to be enough (assuming the air
movement you're likely to get outdoors due to a breeze, but otherwise
standing still) and I had verified that while riding, it was no problem
at all.

It appears I've underestimated the worst case conditions for this
though, so this time, I'm overestimating the heatsinking requirements.

I'm using significantly larger heatsinks with vertical fins (double
sided to increase surface area).
What I'll also do is drill a small hole in the case near where the led
star base is, enough to position a temperature probe. So I'll get a
closer look at that it's running at.

The resulting heatsink is going to be absolutely ridiculously larger
than whatever I've seen suggested (or calculated) but hopefully the
visual appeal will make up for that.
--
Man invented language to satisfy his need to complain.

 

[Uwe Hercksen]

John Tserkezis schrieb:

The resulting heatsink is going to be absolutely ridiculously larger
than whatever I've seen suggested (or calculated) but hopefully the
visual appeal will make up for that.

Hello,

you may try a heatpipe between the LED and the heatsink, the thermal
conductivity is much better than that of aluminium, but a heatsink of
copper is also worth a try, not as good as a heatpipe, but better than
aluminium.

Bye

 

[John Tserkezis]

Uwe Hercksen wrote:

The resulting heatsink is going to be absolutely ridiculously larger
than whatever I've seen suggested (or calculated) but hopefully the
visual appeal will make up for that.

you may try a heatpipe between the LED and the heatsink, the thermal
conductivity is much better than that of aluminium, but a heatsink of
copper is also worth a try, not as good as a heatpipe, but better than
aluminium.

I don't have copper like that lying around, and I don' t fancy buying
something like that either.

I had thought of a heatpipe, but there are additional "problems" to
overcome if I do that.

The 1cm thick block of aluminium the led is bolted to is part of the
structural arrangement of the entire light, so making it thinner will
prevent me from mounting anything securely.

The heatsink can be bolted directly to this 1cm block so I don't
really need a heatpipe.

The most important bit, if I *did* use a heatpipe, I'd have additional
structural problems because I would have to mount it much more securely
than otherwise. (the heatpipe would move/vibrate and fracture).

The general rule when it comes to bicycles (same as go-karts) is that
however tight you make things, they're not tight enough. Crossing your
fingers helps.
Once they're bolted on, you need to test it by absolutely manhandling it.

If it comes loose in your hand, it WILL come loose on the road.
If it does not come loose in your hand, it MIGHT not come loose on the
road. This is where the crossing your fingers comes in.

You may have smooth roads where you are but here, in Sydney
(Australia) we do not. :-(
--
Junk - stuff we throw away. Stuff - junk we keep.

 

[yaputya]

"John Tserkezis" <jt@techniciansyndrome.org.invalid> wrote in message news:j2s7ta$qol$1@dont-email.me...

I have a Cree XLamp MC-E that suffered a catastrophic failure (open circuit, and the clear reflector has come off the die).

I haven't been able to get a straight answer out of any googling or datasheet perusing, so I'm asking here. The reliability spec
sheet absolutely abuses the leds before failure, from what I can guess, more than I have here, so I'm at a a loss.

It lives in a bicycle light, I have the star led die bolted to an aluminium block about 1cm thick, within a PVC pipe section, and
a further heatsink bolted to the other side of that.

All applicable surfaces were honed, (not polished, but smooth and flat), and heatsink compound applied. The only surface I did
not touch was between the actual LED die, and the aluminium star surface it's mounted on - that was left "factory standard".



My first test was in a still room (ambient was about 16C (60F) or so) after about 15-20 minutes, the temperature probe
(positioned near the block) rose to about 65-70C (150F) before I shut it down. The heatsink was quite warm to the touch at that
point. Not _quite_ hot enough to pull my hand away, but nearly there.

I have been using a 5W LED array (Bridgelux) as a reading lamp
with only a 6 x 5 x 2 cm CPU heatsink from an old computer - without
any thermal problems.
It runs warm to hot - but not too hot to comfortably maintain touch though.

Since it's mostly sealed, I didn't have opportunity to measure lead temperature, or anywhere near the actual die star for that
matter.

On the road, it ran for a solid 2.5 hours or so absolutely perfectly, with the heatsink temperature barely at "warm to the
touch". (I didn't have a temperature probe on it on the road).



The failure occurred the next day when I was demonstrating the light to friends. It ran for several seconds before I moved the
bike a bit, where it failed. I had thought it was a wiring issue, till I measured and then dismantled the assembly (it was O/C)
to find bubbles in the MC-E clear lens (mostly around the emitter output edges), and found the lens came off quite easily with my
finger.

It seems that the die lost thermal contact with the heatsink.
You obviously had a screw loose. It only takes a minor degree in the
torque attaching the LED die to the heatsink to make a big difference
in thermal conductivity.