I’m very slowly DIYing a spotlight to have a 12V HXP70.3 mounted on a computer CPU heatsink inside it. I did a test run with the LED/heatsink outside of the spotlight with 68F/20C ambient temp and no fan, 12v@1Amp for 30mins I got the heatsink to 104F/40C, cranked up the power to 12V@2Amps for another 30mins and got to 145F/63C.
Question: how hot can the CPU heatsink get “safely” before I start risking damage to the HXP70.3? I have 5 tiny 12v fans I want to stage using normally open thermal switches(I have a set 40,45,50,55,60,65,70,80,90,100,110,120,130C) but not sure at what heatsink temperature I should start cooling and at what heatsink temperature I should cut power to the HXP70.3.
You need to measure the temperature on the surface of the MCPCB, close to the emitter. So you can relate the emitter temperature with the heatsink temperature. 85°C is a reasonable temperature for the MCPCB. You can use a drop of water to know that the MCPCB is close to 100 °C.
Edit: I think I get it now. Max junction temp - thermal resistance - (MCPCB temp - heatsink temp) would be my MAX heatsink temp before I go over 150 junction temp.
Thank you both for the response, rerunning the test, now inside the spotlight while measuring the heatsink and the MCPCB next to the led temperatures. Looks like my heatsink is 10C cooler than the LED at 12v 1A and 13C at 12V 2A. Lets round that to 15C.
I am still not sure at what temperature measured at the MCPCB next to the LED should I maybe start to worry about damaging the led, and at what temp I will for sure kill it. Data sheet shows relative flux at 150C junction temp, does that mean in my case I can get my heatsink to 126C [150C-(0.2*12v*3.6A)-15C] before I kill the led?
I'm not sure if I would ever drive it that hot but I wouldn't want to cut power to something thinking it might be damaging it when it could easily handle 3 time the heat without problems.
You can get your heatsink to around that temperature before hitting the maximum rated operating temperature of the LED. It won’t kill the LED immediately if it gets hotter than that, but the lifespan (lumen maintenance, L70) is reduced as the temperature rises. Notice that the estimated lifetime hours are still >20,000 at 150C, and that is to 70% of initial output, i.e. L70.
The LED has a slightly higher Vf than 12V at 3.6A. Looks like the Cree Product Characterization Tool doesn’t exist anymore… so I can’t use that to see where the voltage would be, but it’s probably around 12.5V. This still calculates to 126C. You could take a little off of that for headroom.
The accuracy of your temperature measurement matters for that calculation. Ideally you should attach the probe directly adjacent to the LED and measure there. The probe tip should be placed into a blob of thermal grease.
There is some thermal resistance in the solder joint under the LED, but it’s small enough to be negligible I think. If your MCPCB is not a “direct thermal path” type, then the junction temp can be much higher than the MCPCB, but most MCPCBs these days are copper, and of the direct thermal path type.
But when will the LED actually be killed? The LED would desolder itself from the MCPCB before it would die. You’ll be in the dark when the LED desolders itself.