Hi all,
As part of my project, I had ordered four Cree XLamp XP-C LEDs (Blue). I connected them in series and they are driven by a voltage source followed by a current limiter(LM317 configured as a limiter). It limits current to 350mA, rated current of the LED.
The SMD type LEDs are mounted onto an MCPCB and then onto a CPU heat sink. I had calculated the junction temperature and it is within the maximum tolerable value.

The circuit worked just fine and the LEDs lit up today morning and in the afternoon, but now its not working. The LEDs seem to have just... failed? I used a digital multimeter and set it in the 200 ohm range to test the LEDs individually after detaching them from the circuit. This method of testing used to work before, but the LEDs does not light up now. The circuit itself does not light the LEDs. I'm pretty sure the current was well within the rated value, so have my LEDs really busted? What did I do wrong?
At the same time, green and red LEDs that I have along with this are working fine.

I had adjusted the current through the LED (within the range of currents) a few times, is this harmful to the LED?

 

If the current was within the LED ratings and failed then I suspect they became overheated.
Are you sure your heatsink arrangement was sufficient to keep the chips cool enough?
What was the thermal resistance between the LEDs and the MCPCB?
Did you use thermal grease between the MCPCB and the CPU heat sink?

 

Check for shorts where the wires are soldered to the MCPCB..
Got a schematic of the whole setup?
thermal resistance of the CPU heatsink?
overheat the LM317?
true crees or chinese knockoffs?

350ma should be totally fine as long as you keep them cool..

 

I experiment mostly with high wattage, cool white LEDs; sometimes they will flash when applying too much current.

If I shut them off quickly, when I come back they work again.

 

Hi all,
Thank you for the fast replies!
I used a CPU processor heat sink. Four blue MCPCBs were soldered onto it. When the setup lights up, the heat sink heats up, but slowly. If you light it for a few minutes, it'll feel warm on touch.

The LEDs are real cree, I brought them from element14. (Farnell).
LM317 was provided with an adequate heat sink.

I assumed that a CPU heat sink would have near 0.5 °C/W thermal resistance. Put it as 1 °C/W for safety.
LED Heat Sink Design
Cree XLamp XP-C ideal Junction temperature Tj=25°C
Maximum Tj = 150 °C
Junction to solder point thermal resistance
of red, Rjsr= 10°C/W
of Green, Rjsg=20°C/W
of Blue, Rjsb=12°C/W
Thermal grease thermal resistance is assumed to be zero.
Thermal resistance of MCPCB (Rm) is near 10°C/W
Thermal resistance of heat sink (Rh): 1°C/W

We use three heat sinks, so consider 4 LED per heat sink (12 in total)
Power dissipated (Power not converted to light) is 60% of the total input power. We consider the worst case scenario with 75% power dissipation.
Power dissipation in:
Red LED, Pdr=0.77 W X 0.75 = 0.5775 W
Green LED, Pdg=1.19W X0.75 = 0.8925 W
Blue LED, Pdb=1.115W X 0.75 = 0.83625 W
Where Pd is the total power dissipation.
Assuming ambient 55°C,
Tj=Ambient+ Rjs of each LED*its Power dissipated+ MCPCB temp. drop + heat sink drop

Red LED:
Tj = Ta+ 4 X Pdr* Rh+ Pdr*(Rjsr + Rm)
= 55°C+4 X 0.5775 X 1+0.5775 X (10+10)
= 68.86 °C
Green LED:
Tj = Ta+ 4 X Pdr* Rh+ Pdr*(Rjsr + Rm)
= 55°C+4 X 0.8925 X 1+0.8925 X (20+10)
= 85.345°C
Blue LED:
Tj = Ta+ 4 X Pdr* Rh+ Pdr*(Rjsr + Rm)
= 55°C+4X0.8363 X 1+0.8363 X (12+10)
= 76.74 °C

Calculated junction temperatures are less than the maximum junction temperature (150 °C).

The circuit diagram is attached.
What did I do wrong?