Middle LED in a series burning out

kuksul08

Enlightened
Joined
Jun 4, 2007
Messages
783
Location
California
At first I thought it was a dud LED, and now I know something is wrong.

I am using an LM338 in constant current mode. So it uses a .47ohm resistor and I get around 2.6A into a series of 3 XM-Ls from a 13V source. I've been using this for a while in another setup without issue.

I connected power and it worked correctly for about 10 seconds. Then, the 2nd LED in the series burned out. When it burned out, the 1st and 3rd LEDs still shine, but dimly. I replaced the LED thinking it was just crap. I checked the current when I first turned it on and it was correct at 2.6A. Then about 10 seconds later, it turns off.

So, I then bypassed that middle LED and connected only the 1st and 3rd LEDs - it worked fine, except for more heat being dissipated by the regulator.

How can this possibly be?

edit: I busted out some "junk" old MC-E's that I had laying around and wired it exactly the same way, but monitoring current for every second and making sure it was heatsinked in the same manner. It ran just fine. A solid 2640mA for 3 or 4 minutes. Got pretty hot and still remained the same. Could I have just had two duds in a row? Are warm white XM-Ls limited to a lower current?
 
Last edited:

Epsilon

Enlightened
Joined
Sep 19, 2010
Messages
463
Location
Netherlands
2.8A should not kill an XML, regardless is the tint.

If they really are all getting 2.8A, then there should not be a problem. But it can if the LEDs are placed right next to each other. Then the middle one is getting hotter than the other two, resulting in a dead LED. The second time might just be a bad thermal path.

How big is the heat sink and what paste are you using?
 

kuksul08

Enlightened
Joined
Jun 4, 2007
Messages
783
Location
California
That's what I thought. When testing the running circuit, I should only need to monitor current, correct? The voltage across each LED will follow just depending on the characteristics of the LED.

They are attached to 20mm aluminum stars and then with silicone/ceramic paste to a large aluminum heatsink. Any noticeable heat is only evident after about 45 seconds of running, and these LEDs died within 10 or so seconds.

I'm starting to think that they are not thermally connected well to the star board, from my supplier.
 
Top