I just remembered an additional material upgrade. Normal dtp pcbs are made out of copper or aluminium. There is another possible material onto which the LED can be directly soldered which transfers heat even better than copper: silver. Basti in the German TLF forum made such a pcb a few years ago. He made it just for the looks though.
The thermal conductivity of silver is 7% higher than that of the best possible (most pure) copper. Most common copper alloys are not 99.99% pure though, making the difference for pronounced.
I asked him if he would make one for me for my big thrower when I was planning the build, but he is not an active flashahlic anymore.
This is a variant of Panasonic’s PGS, and most importantly, soft PGS.
The two disadvantages of using PGS are higher contact thermal impedance compared to thermal paste if applied without enough pressure, and electrical conductivity.
Soft-PGS is better in this regard than pure PGS, since while it features much higher thermal resistance(20W-28W/k vs 300-1800W/k), PC users have tested that thermal performance is much better with soft PGS because of much lower contact thermal resistance as specified above.
It’s also because soft PGS has much better thermal transfer in the Z-Axis compared to the X/Y-Axis, meaning thermal transfer of high power density products is much better than regular PGS.
Link:
TLDR: Soft-PGS is better in almost all cases with high power density electronics, such as CPUs/GPUs and VHP LEDs such as XHP70.2s overdriven, CFT-90s up to 100W power levels without starting to fall behind thermal paste because of heat spots without very high pressure mounting. The best combination for massive thermal transfer capabilities are liquid metal below the soft PGS, and the soft PGS under the MCPCB itself.
Most flashlights allow providing some decent MCPCB to pill/shelf pressure. Does this mean a piece of 0.2mm soft-PGS under the emitter board is nice enough? Is it electrically conductive?
Yes, it is electrically conductive. Just be great for almost all LED builds, except that CFT-90 LED. That thing has huge power density, and needs liquid metal.
Another thing which I suspect won’t work but I’ll throw it anyway. I wonder if improvements to radiation either directly from the LED package or from the MCPCB could bring measurable performance increase?
-The one with 28 W/m·K is the EYGS series which comes in 0.2mm thickness and is electrically conductive.
-The one with high thermal conductivity in the z-axis (vs. x/y) is the EYGT series. This is graphite in a silicone matrix , it looks like this is NOT conductive (4*10^5 Ω·cm), but comes only in >0.5mm thick sheets and has a conductivity of 5 to 10 W/m·K (strangely, it goes up with thickness?). Has some sort of separator but the datasheet is not clear what its properties are or whether it needs to be removed before use.
This doesn’t make much sense to me. The LED and pcb radiate into the air inside the flashligth head. Air is a very bad conductor of heat. After this the heat would still need to go through the metal of the head or through the glass lens (glass is somewhat good at conducting heat). You would need to replace the air inside the head with something else that doesn’t transmit light any worse, but has a noticeably better thermal conductivity.
Would air absorb significant part of the radiation? I assumed no, but maybe incorrectly.
Otherwise goes towards the optics or towards the head sides or both, depending on what radiates it and what’s the optic.
Head can absorb it, conduct it towards the edges and remove.
Bi57Sn43/Bi58Sn42 should make your life a lot easier when soldering. Versus standard Sn63Pb37 or Sn60Pb40 you can do with around 50°C less in your iron.
Cheers :-)
Originally posted on Wed, 10/10/2018 - 19:06. Edited for a tiny explanatory addition.