Aluminium pill vs copper pill - share before and after results

Here I try to explain how I see the benefit of copper PCB's. This is by no means ment to be the only truth but a trial to set things in perspective.

Two years ago I made thermal measurements on a Solarforce L2 body with a standard brass pill, but instead of an LED I mounted some resistors to burn some controlled heat and measured the temperature of the pill's mounting face as well as the outside body temperature of the light. This is a slow affair as one has to await equilibrium at every power level.

The scenario was like this with some empirical (round) numbers that fitted the measurements:

Thermal resistance (in 'C/Watt):
From LED (Tj) to LED mounting base: 2.5 (XM-L)
From LED mb to PCB (Alum.): 0 (often spec'd as 0.25)
From PCB to PILL (Brass): 1 (This will be lower for Copper -see later for that!
From PILL to Host Body: 6 (I have seen 5-9 depending on the gab padding with alu etc.)
From Host Body to Ambient: 6
Total Thermal Resistance 2.5+1+6+6 = 15.5

All this can be simply calculated on as in Volt = Current x Resistance
where Temperature difference ('C) is Volt, Heat flow (Watt) is Current and Resistance of course is Thermal Resistance.

An example:
A XM-L driven at 3.3 Amp in a L2 host burns 10.1 Watt in the LED.
If we use the total resistance from abowe (15.5) it will give us a temperature rise of the LED of
dT = 10.1 x 15.5 = 157 'C or 25 + 157 = 182 'C in the LED .....WHAT?

Here we forgot the Luminous Efficiency of the LED (NOT Efficacy! -as this is an optical measure). The efficiency is around 30% (very empirical) so 0.30 x 10.1 Watt = 3.03 Watt is distributed as light and will not produce heat inside the flashlight.
Then dT = (10.1 - 3.03) x 15.5 = 110 'C and the led will be approx. 135 'C if ambient temperature is 25 'C.

Then what about the copper PCB?
If ideal it can only bring the total Thermal resistance down 1 'C/W (total resistance 14.5 'C/W).
The LED temperature will then be 128 'C and we will have 641 lumens out the Lens instead of 628.
That is only 2% more light with a copper PCB!

Edit: Also a copper pill can only change the value of "PILL to Host Body = 6" a little as most of the resistance in this construction part comes from the infamous gap in P60. I guess another 2% could be gained over an alu pill. In a C8 like host that could be more.

If you had an aluminum MCPCB, without the dielectric layer, and could solder a LED to it, it wouldn't be much if at all different from the copper boards. It's the dielectric layer that makes the difference not so much the material of the backing.

Aren't the AL Sinkpads like that? The solder doesn't actually bond to the center pad, it just touches (maybe it sticks a little, but not an actual solder joint)?

My position is that copper pills are better than aluminum.

Radiation heat transfer is a function of the difference in temperature to the 4th power. (Thot4-Tcold4)

Convective heat transfer is a function of the difference in temperature to the first power. (Thot-Tcold)

Both of them matter. Convection matters more to us than radiation due to the relatively low temperatures we operate our lights.

Why do I think copper is better? Inside the pill, whether it is aluminum or copper, there is a temperature gradient. It is hottest in the middle and coolest on the outside. It also varies over it's length in the same manner. Where copper outperforms aluminum is that the slope of this gradient is less since copper is about 89% better than aluminum at conducting heat. The outside temperature is closer to the inside temperature. (Luminarium observed that the pill gets hot quicker. I think this is commonly observed.) What this does, even if it is only a few degrees, is increase that temperature difference where heat is being exchanged (Thot-Tcold) at the body surface. That temperature difference is what drives the heat transfer. The larger the difference, the more heat is transferred. This is true of both radiation and convection. It is even applicable to conduction through the aluminum body of the light. So what this means is that for a given LED die temperature, the outside of the light will be hotter, thus convect and radiate more heat to it's surroundings. Stating it in another way, copper gets the heat to the outside quicker where a higher delta T transfers more heat away.

Secondly, barring any heat exchange at all, the copper pill is much more dense. For a given volume, it has roughly twice the heat capacity of aluminum. So for short term uses of a light, it will operate twice as long just using that copper as a heat reservoir than aluminum would.

I know some or most of you won't be convinced. What I have explained above is the cliff notes version of heat transfer. I've posted the equations that were derived by the giants of science in other threads and that didn't convince you so I won't bother to do it again here. Maybe someday I'll do a graphic on it. That seems to allow people to understand more easily.

I don't think anybody's arguing that materials with better thermal conductivity will not be better at conducting heat, but rather, is there enough difference between the common materials, in the shape of parts used in a flashlight, that you can measure a difference in the light that comes out the shiny end?

Like the guy doing all the testing of wrap materials for P60 pills - running tons of tests, but only being concerned with what the different setups did to the junction temperature, and completely unconcerned with if there's any real difference in the light output.

As a wise man once said "Sometimes good enough really is good enough, and sometimes good enough is better."

it does get bonded to the heatpad, i reflowed on al sinkpad2, it is not bare aluminum that touches led, it looks like electroplated copper, or something else, but solder sticks to it.

Reading all of this thread, posted up a similar thread when everyone started selling the copper pills. Anxious for the verdict.

Thanks for the info Bucket - I at least appreciate it very much! You explained it well. comfy has a point though - ultimately it's what the effect of the differences are in real life, in what we do and what we have on hand. Even the thinner cheap alloy aluminum stock pills, compared to the over sized copper ones that are now available to us. Even in simple volume, the copper ones from VOB and Ryan are much more robust.

0:)

Good enough is good enough if it’s good enough for you. I’m pleased by each light I finish but I still try to make the next one better. It seems pretty obvious that direct thermal path copper stars represent a significant and indisputable improvement over aluminum stars w/dielectric layer. Improving the performance at each material transition and minimizing the # of transitions is a goal to be strived for though as comfychair says, not idolized. Mods that don’t push the lumen/wattage envelope will still work with aluminum stars just fine. There’s still room in the world for a .5A 6V lantern.

A few years ago I was asked to build a light for bigChelis on CPF. It was a simple direct drive SST-90 MagLite and it was to have three heatsinks made: brass, copper, and aluminum. I never made the brass one. The end result? No difference in output from initial turn on until we had decided to turn it off. What did change was the output graph. With aluminum, the output dropped quickly as the heatsink and the outside of the light heated up. After a few minutes the system reached a saturation level and the output dropped more quickly. With copper the output dropped slowly and took longer to reach that saturation level at which point output dropped just like on the aluminum heatsink. Conclusion: copper draws the heat away from the LED faster and more efficiently, allowing you to run at high outputs for longer before heat becomes an issue.

Since then I’ve had the pleasure of owning a 3C MagLite that had direct to copper bonding of the emitter and it was insane. I think bigC measured 1300 lumens OTF on that light and it would take a minute or two to get to 1200.

Dale has some pretty sweet numbers posted up today in his Trifecta thread

I should also add that after machining the copper pill for Dale's M8 and his experience with the head of the light heating up more quickly, I think that the relative looseness or tightness of the pill threads has a large part in how well the small internal pills will transfer heat to the head. It stands to reason that loose, rattly threads won't have as much contact area with the head. I also think that whatever pill you have should be screwed in as tight as you can get it without making it permanent :)