Heatsinking Materials

Flashlight Modding and DIY Parts
1

Hi all,
I have been doing some digging to see what the best material option for passive heatsinking is. Not surprisingly, copper is still the king, (well silver is technically better, but who has that kind of money?)

The other thing i was looking for was the best Thermal interlock material i.e. Arctic silver or Arctic alumina. This is where things started to get more interesting. I know everyone here knows that Arctic silver is pretty costly, so after a little research, i think i might have come up with an alternative.

This link was probably the most helpful single source. Just a list of the thermal conductivity of common materials.

The most unusual thing i found is that hardware store J-B weld has a conductivity of .55 W (m K) and Arctic Silver has a conductivity of .95 Watts per meter Kelvin. These statistics came from a seemingly trustworthy PC forum,which i cannot seem to find at the moment. ( i will link it if i find it) That is somewhat surprising IMHO. I cannot confirm these numbers, but it makes sense.
I should mention that J-B weld is capacitive, but only very, very slightly. I wouldn’t worry about it until you get above 12v, and then only on massively high currents, and very closely spaced contacts.

Another thing I’d like to mention is that brass is a poorer conductor of heat than Aluminum. I’ve gotten the general idea from reading various threads on here and CPF, that most people tend to think that it is better, due to it’s copper content. This is not the case at all, the addition of zinc interrupts the copper-copper bonding and 3/4ths of the thermal conductivity is lost.

Also, different alloys of the same metal can affect its conductivity. Brass is an alloy of copper and zinc, yet its conductivity is lower than zinc alone.
2011 aluminum has lead and bismuth added to make it more machinable, but the slight addition of these heavy metals destroys the aluminum-aluminum bonding, and the thermal conductivity suffers. Also, cold finished tends to have worse thermal properties than extruded, due to the small tensions and warping of some of the crystal structure.

The last point i want to make is solder. Here is a list of the thermal conductivities of different common (and not so common) solders. The best is clearly tin-silver, although i have never used any myself. If any of you electronic gurus have, please share your experience. The issue seems to be its very low melting point. Good ol’ 60/40 is 50 W (m K)

Just some food for thought

2

Great reading. Thank you for the links. I just have some old Dow Corning stuff I’ve been using for some time now. I’m really happy to know that about JB Weld. I’ve got tubes of the stuff but have always been afraid to use it for HS purposes. For solder I’ve been using an old blue roll of Kester Ultrapure Sn63/Flux-RA, whatever that means. :stuck_out_tongue:

3

While brass has 1/2 the thermal conductivity of aluminum, it has 3x the mass by volume. That should mean that in 2 identical pills of each material, the brass should be 50% better as a heatsink as it would weigh 3x as much as the aluminum one.

I’d appreciate it with someone with a better grasp of the calculations will verify.

From Wikipedia:
“In SI units, thermal conductivity is measured in watts per meter kelvin (W/(m·K)). The dimension of thermal conductivity is M1L1T−3Θ−1. These variables are (M)mass, (L)length, (T)time, and (Θ)temperature.”

4

Brass would only take longer to become heat saturated, it is not 3x better than aluminum. Brass is only popular because we can solder copper stars to them. Just my humble opinion

5

What did you conclude to be the best thermal interlock material?

Some other info here .

6

Interesting, you may be on to something. If that’s the case, would the brass conduct heat better than aluminum once the aluminum has fully soaked, and the brass hasn’t? It might turn out that the advantage over time for brass vs aluminum would look like a bell curve with the beginning and end in the negative region, but an advantage between the time that the aluminum is fully heat soaked and the brass does.

I wonder if any of the shade-tree scientists around here have tested any of this and have temp/time graphs showing the temperature of both the LED and the opposite side of the same size of heatsink out of both materials? I vaguely recall a thread with someone talking about doing something like it.

7

I never really came to a concise conclusion, but what I can say is that for lower power applications, JB weld should be fine. Arctic silver doesn’t have much better characteristics, and is several times more expensive. If you are severely overdriving an emitter, I’d say solder it, but even at the high end of its rated current,you won’t notice much difference between JB or AS

8

Not that I know of, but I think I’m going to try it once I get some spare cash. I’ll make 2 pills, one in aluminum and one in brass, and monitor the temps of both ends.

9

The topic has been discussed several times since Ive been here. Topics including pill material, MCPCBs and thermal pastes and solder options. Sources of studies of the thermal capabilities of the thermal pastes came from PC overclockers too.

IIRC, the most important aspect was a direct thermal path MCPCB, again didnt really matter if copper or aluminium. The pill or shelf for said MCPCB to sit on didnt matter much which material it was made from, just as long as it did have a pill or shelf to move heat to the body of the light.

63/37 solder paste IIRC was touted as the nicest to work with and ample for our purposes.

10

There are another 2 facets to this issue that have to be accounted for. How quickly the metals emit energy (and there is a term for this that is slipping my mind) and heat bottlenecking.

I know that aluminum emits heat much fast than most other metals, partly due to its low density, but perhaps more of a quirk of its physical chemistry. This is fairly was to proove. If you heat a peice of aluminum and an equal volume of brass to the same temperature, the aluminum will cool much more quickly.

Heat bottlenecking is an issue for brass and other low-conductivity metals. While they do take much longer to become heat saturated, the heat sits under the emitter star for longer as a result. Copper is denser than brass, but it also has 4x the conductivity, so it is better all around.

The objective is to move heat away as fast as possible, so I think that aluminum still has a slight advantage over brass. After the point of saturation, I am not sure what will happen. Once the brass is saturated, it’s stuck that way for a while. The aluminum will cool faster, and potentially fast enough to equal the rate of is absorbing energy. This is all speculation though. Time for testing ! J)

11

The difference between brass and aluminum is negligible. Both brass and aluminum transfer heat better than all of the thermal paste and adhesives. That is the weak link in thermal transfer. Also, people love to throw out the number that aluminum is 3-4 times the thermal conductivity of brass which is just wrong. Maybe for pure aluminum, but nobody uses pure aluminum in the manufacturing of flashlights, they are aluminum alloys, which bring the thermal conductivity down. It’s a big NO NO to use brass and copper interchangeably, but aluminum and aluminum alloys should not be used interchangeably either as they have much different properties. T6 aluminum is probably the most common and cheapest and is 1.5-2x higher thermal conductivity than brass depending of the composition of the brass. Add in the fact that the PCB can be soldered to brass and not aluminum eliminating the bottleneck in heat transfer, I believe that the advantages of brass outweigh aluminum.

12

I agree 100% that thermal pastes are a massive weak link.

However, T6 is only a temper, not an alloy.
The statistic for 6063 T5 is 209 W (m K) as evidenced Here

1100 F is 99% aluminum, and is the most pure commercially available alloy. It’s conductivity is 218 W (m K). The point is, if you plan to use a paste, aluminum is much better (and easier on the wallet too).

13

Sorry, I meant to post 6061-t6. It is one of the most common aluminum alloys and is listed at 167 wmk here

I agree, if using paste and not driving hard, aluminum is adequate and cheaper. A DTP soldered to a brass pill is more effective than aluminum until it is completely heat soaked, but still falls short of copper.

14

I think both the higher mass and lower transfer coefficient are why brass takes so much longer to cool from the same temperature when they are the same volume. What we’re talking about is the ability to absorb, move, and release the same amount of ENERGY. When you heat them to the same temperature you’ve put substantially more heat energy into the brass than the aluminum because of its higher mass. At saturation I guess I would expect the aluminum to maintain a lower temperature than brass because it is easier for it to move it to the outside air.

Whether the differences between the pills actually makes a difference to the LED is more important to us I guess, and we may find that the amount of heat going into the two materials may matter too. Likely there will be no noticeable difference on reasonable sized pills at moderate power levels. Running 25+ watts on a 16cm³ pill is probably where the differences will become large enough to measure with any consistency with the light meter.

This thread reminds me of the capacitor threads on DIYMA.

I’m going to order a few analog temp sensors for my arduino. If I ever get some time and motivation I’ll try to do this test too. Keep the ideas coming guys. I feel like I have a hypothesis to test now, but we all still have a lot of opinions/ideas/crackpot theories to share, some may even be right! :wink:

When all is said and done, I bet altering the geometry of the pill to maximize it’s surface area contact to the body makes a bigger difference than the material (alum vs brass only being considered here) or it’s final weight(assuming a DTP star that’s mounted well to the pill). Then maximizing the surface area of the body to the outside air. After all, I guess that’s the biggest insulator in the entire heat path, the air that the body is radiating the heat to. Getting the heat to spread across as much of the body as possible will have a big effect. Long threaded or press fit pills should be getting more attention than the material they’re made from perhaps?

15

No pill designs should perform better than flashlights with a pill. I wish more lights came as a one piece design so the heat dissipates to the body better. Aluminum would be great as it would be able to adequately transfer heat to the entire head. I think the Jacob A60 had a good design. Increase the mass and it would have been perfect. I wish more would adopt this