XPG fails "closed"?

Gather App
Candle Power Forums

Help Support Candle Power:

K

kananga

Newly Enlightened
Joined
Nov 1, 2008
Messages
27
I've got one non responsive R5 in a series circuit: It's failed nearly short-circuit. I'm measuring 1.1 ohms across the mcpcb contacts. Is this something an XPG will do, or would this be more likely related to the sloppy soldering of the led to the board? It's significantly off center and out of alignment.

I've only been feeding these led's 500ma, and otherwise it's been just fine for a full year of regular use. Failure was basically instant; no gradual loss of output, it just up and quit.
 
Got it - I was successfully able to heat up the board and slide the led off without damaging anything. Board failure. The led is fine.
 
It is very rare for diodes to fail closed, especially on unregulated circuits since a close failure would result in a rise in current which results in an inevitable open failure. If I see that happening I generally look for external causes.
 
qwertyydude said:
It is very rare for diodes to fail closed, especially on unregulated circuits since a close failure would result in a rise in current which results in an inevitable open failure. If I see that happening I generally look for external causes.
Click to expand...
I had experience of strings of coloured (R/G/B) Rebels dying due to driver failure, where the driver was a buck regulator normally driving 4-8 LEDs in series from a 24V supply.
When the driver failed and put the full 24V across a string, typically it ended up blowing all the LEDs in the string short-circuit, before something else in the driver failed, the input fuse blew, or the source PSU shut down.

With this happening something like 6 times, I'm not sure if any LEDs actually failed open-circuit.
 
Last edited:
qwertyydude said:
It is very rare for diodes to fail closed, especially on unregulated circuits since a close failure would result in a rise in current which results in an inevitable open failure. If I see that happening I generally look for external causes.
Click to expand...

Actually, I know for a fact that the new Rebel ES is designed to fail closed. Can't speak for the old Rebel LEDs or the XP-G.
 
My electrically-destroyed XR-Es failed closed.

Edit: I'd like to add that I REALLY killed 'em electrically. I put about 18v across two of them :(
 
Last edited:
every led that ive had fail, failed open

this rebel es, is it a real thing? like being offered in the near future?
 
Optical Inferno said:
Actually, I know for a fact that the new Rebel ES is designed to fail closed. Can't speak for the old Rebel LEDs or the XP-G.
Click to expand...

That sure sounds like a disaster for any direct-drive setup, or any driver that doesn't like being shorted.

Hey Optical, I see you're in Niagara Falls? I'm a Thorold born, raised and escaped boy myself :)
 
Hi, I spoke with a Lumileds sales guy some time back about the fail open / closed aspect. He told me that the older K2 / Lux IIIs tended to fail open, but customers wanted them to fail closed / shorted. For that reason, and others, the Rebels fail closed.

The idea is that if the LEDs are in fact in a string with a constant current driver running them, there could be an LED failure and the string would keep running at the constant current.

I had mixed opinions about this, but it makes sense. If you think about it, if you have 3 strings in parallel sharing a constant current driver, and one LED fails, regardless of "fail open" or "fail closed", you are kind of in trouble. If you have a driver on each string and it fails open - one entire string goes out, but not if it fails closed.

As a practical matter, the failures seem to be pretty rare, and might be more assy. driven than a fundamental bad part issue when it comes to Cree and Lumileds parts.
 
After this little incident, I can see why some designers might prefer a fail-closed mode for current controlled series lights!

That board that shorted out was part of a two-led series circuit bicycle headlight, powered by a hub dynamo. I've always heard the hub dynamos are current limited, but recently a friend checked his out with an ammeter and a steep hill. I don't think I'd have had that kind of confidence, but... 500ma, dead short through the meter, all the way up to 40 mph! Anyway...

That board shorted 8 miles from home at 11:00 PM, on the wrong side of the I-5 bridge over the Columbia River. Because it failed closed, and because the current supply was limited to 500ma regardless, I still had one xpg to show me the way safely through the pinball machine that is that particular nonmotorized river crossing. If an led is going to fail in this circuit, that's how I'd like to see it go!

On the other hand, it wasn't an led, it was a product quality failure. So, is there a small-quantity source for board-mounted r5's that isn't on the opposite side of the world as Cree yet? :hitit:
 
Correct me if I'm wrong, but the old K2's, L1's, L3's, and L5's used a bond wire to supply the power to the LED die. The AlInGaP rebels also have the same bond wire. So all of them would fail open taking out the entire series strand.

The new Rebel ES and the InGaN rebels (Ithink) don't have bond wires. So they fail closed. I may be mistaken. The only one I know for sure fails closed is the new Rebel ES.

And Canuke, I also am a Thorold boy born, raised and escaped but not as far as you were able to get.:wave:
 
Surely the failure mode depends on the nature of the power supply and driver?

Even for single-LED installations, and even if there are bond wires, if an LED fails short-circuit, it's still possible for currents to flow that take out a driver component or shut down a power supply, but which aren't enough to burn out a bond wire.

http://www.candlepowerforums.com/vb/showthread.php?t=245291

Though I can't find the threads, I do remember past threads where the idea seemed to be that LEDs failing closed-circuit wasn't an unusual occurrence, and I don't remember that being a Rebel-specific issue.
 
one was a 24v power supply with a transistor based driver, xr-e q5 from dealextreme, i might have had one more experience but i dont remember the details
 
High powered LEDs, unless they "blow up", i.e. act as a fuse, do fail short.

HarryN, I think that the Lumileds guy was feeding you a line that they designed them that way. The old Luxeon's would do that too though not as often. It is inherent in the manufacturing of the LEDS when you over-voltage and/or hit them with a big static shock.

Semiman
 
We have just had our first XPG-R5 LED fail in one of our flashlights after several thousand in service. Only one
half of the outer band would light up. The interior 'fingers' were also dark. This was in a Stainless Steel
Night Patrol. The drive current to the LED in this light is 1.1 Amps, less than in our First Responder FR400 and
the Search and Rescue SR450 that are both being driven at 1.5 Amps. The drive circuit output is within
parameters and the heat-sink also seems to be fine. Just one of those things. Otherwise a great little LED.

Curt
 
SemiMan said:
High powered LEDs, unless they "blow up", i.e. act as a fuse, do fail short.

HarryN, I think that the Lumileds guy was feeding you a line that they designed them that way. The old Luxeon's would do that too though not as often. It is inherent in the manufacturing of the LEDS when you over-voltage and/or hit them with a big static shock.

Semiman
Click to expand...

You are probably right that the older products could fail open or short, but the rebels have no bond wires, so the odds of an open are very low. The presentation that I saw at a conference showed the rebel connections to be eutectics, not solder, so that failure mode should also be very low. I will admit that this was several years ago, so my memory of the details is imperfect.
 
You must log in or register to reply here.

Latest posts

Back
Top