I have a IR thermometer which has so far been reliable, cheap, uncalibrated, but it’s accuracy around 40*C has been confirmed.
All of my flashlights have noctigons now, all are XML/2. They are all 1s1p or 1s2p 18650, driven to 3A at max.
I have observed that (after a few minutes at 3A from fresh battery) the temperature of the body goes up to 60-70*C, depending on size of it, but the emitter temperature goes up to 125-155*C.
That’s in bit of contrast as to what I expected with noctigons, that big temp differential between the emitter and the body, that would mean that the thermal conductance is quite low.
edit:
that’s not nearly as appealing as shown here in IR photo for noctigon:
i have bought few leds soldered on noctigon, and found some were soldered very poorly, especially mtg2, there were visible gaps between a heatpad of the led and heatpad of the star, half of it was simply hanging in the air. (the heat pad and contacts on noctigon stars was at the time not leveled, heatpad would sit lower), it would sure hurt heat transfer, i returned those, and since than i never buy a led soldered on a star, all soldering i do myself. may be you have poorly soldered star, with air pockets under the led.
heat it with hot air, when the solder melts, gently press the emitter onto the noctigon by pushing down on the dome with tweezers.
I tried it on 3 different lights, always got the same numbers, 125-155 in thermal equilibrium, depending on battery state.time
anybody else got a IR thermometer to share their results? I was doing it by pointing down on to the very emitter die, and moving the thermometer around until I found the angle&position at which it was registering the highest temperature.
How are you attaching the MCPCB to the emitter shelf of the light? (Not the LED to the MCPCB)?
Notice in the sinkPAD add picture that they’re screwed down. They helps in more way than 1- it ensures a gapless transition from MCPCB to head and also the screws themselves carry heat.
The more TIM you use the worse, there is no better thermal transfer that that of perfectly smooth metal on metal.
The fact that you have a copper mcpcb mounted to an aluminum heat sink or aluminum shelf, makes it a bottle neck situation. Aluminum will take the heat slower than the copper is putting out, so heat will just increase in the copper, since there is nowhere to go. copper stars are great, but a complete copper head, with an integral led shelf is about the only way to get the heat out of the mcpcb quickly.
But, if you were really at 155C, the solder should be melted off and everything would be shorted out by then, so I question the accuracy a little.
Thermal conductivity of aluminum is cca 2x less than copper
The entire surface of the emitter contacting the mcpcb is (estimation) cca 5mm x 5mm = 30mm2,
The surface where it has direct contact with copper is 15mm2,
The surface where it is contacting the mcpcp with + and - is cca 10mm2.
I’d estimate that to equivalent of 20mm2 of direct copper contact, let’s call this surface A
The surface of the 17mm noctigon is cca 216mm2 (reduced by 10mm2 for the screw holes), let’s call this surface B
B/A = 10,8
Even though B is transferring heat to the aluminum shelf which has half the thermal conductivity of copper, it is still doing so across approximately 10.8 times larger surface, therefore effectively 5.4 times higher absolute thermal conductance. Of course this is not direct metal-to-metal transfer, so there is some loss of conductivity between the noctigon and the shelf, but it shouldn’t account to the calculation being wrong by order of magnitude.
The solder should melt around 180*C if I remember correctly, though I can try.
This is an interesting subject, please add if you wish so. I’ll do some more testing and ajduct calculation for copper-aluminum transfer when I find time.
yea, the biggest bottle neck is the heatpad itself, it does not matter if al. has less heat conductivity than copper, due to areas it will be more than enough for proper heat transfer. also any led has max die temp 150c if you ran a led at 155 i’d die rather quickly.
i came to conclusion, based on experience, ir termometers are not reliable when you mesure small parts with big temp. difference. the probe is the way to go, a lot more accurate and presice.
yea, believe it or not, but leds sometime do desolder, i remember i soldered 2 wires to sides of 5050 led, and hooked it directly to cell phone battery, it took 2 sec for 1 wire to desolder. strange enough, after i resoldered the wire and used resistor, led worked fine, i would expect the heat that melted solder to kill it, but it did not, not right away at least.
Update: I have a test setup, I don’t have a tube light at the moment but I’m gonna be doing a test with a convoy M2 host I’m building. I have a set of Micro K type t-couples, I’ll solder one to the thermal pad (the little dot on SinkPADs) then afix one to the head, I have 2 identical meters capable of monitoring them so that should allow decent accuracy.
I may not get finished today but once done I plan do make a new thread about it. The test will be to show whether there is a temp difference between using an Al oR Cu SinkPAD with all other factors the same.
LEDs desoldering, I had those, desoldering at 2.5A on regular MCPCBs with no heatsinking (MCPCB sitting in the air hanging by the leads), surely this can happen.
I have also a cheap IR thermometer that I want to throw away, since it absolutely useless at around 90-100C, limit is claimed to be around 105C, I have get something with greater range.
What is interesting about this 155C that if you would use a lightmeter to log the light ouput from start ( at ambient temperature) to the point it reaches what your IR meter says (155C) we should see a huge lumen sag in the graph, some reviews show little sag on P60 modules like the CUXM2, which we all know are not the best conditions for an LED. Even I have had very low lumen sag on CUXM2.
