Lets discuss heatsinking.

LED Flashlights – General Info
21

It is actually very common to have the material closest to the heat producer be of very high conductivity material (copper) and then the rest of the heat sink be a lesser conducting material. The reason is cross sectional area.

Near the heat-producer, there is a very narrow freeway for the heat to move. Because of this, the heat has to move very quickly (high conductivity material). As the heat moves further from the producer, the freeway gets wider and wider, and a slower flow of traffic moves the same number of cars (heat).

Using copper as a pill would be the ideal. Aluminium and Brass are probably close enough that we may as well use Brass since it can be soldered to, unlike aluminum.

PPtk

P.S. See the common Intel heatsink for a perfect example. A copper slug at the heat-producer (CPU) mounted in an aluminum finned heatsink.

22

I suppose this is not what everyone meant by heat sinking

or this

23

I agree, but that Intel heat sink has a fan for active cooling, where a flashlight does not. The bottle neck is reduced dramatically by the fan cooling and it makes the package workable and more efficient. Air around the outside of a flashlight is in no way a replacement for active cooling, yes? no?

Hook up that Intel heat sink to the Intel CPU and leave the fan off it and see how long it takes to fry that CPU, correct?

24

a copper pill in an aluminum light has superior heatsink performance to a dimensionally identical aluminum pill in an aluminum light - every time.

25

I have been thinking about using diamond paste as thermal void-fillers. Is this idea crazy or not? The idea is to use the diamond grit paste as lapping compound, then manually rubbing the LED star against pill for a good while – creating a smooth lapped surface of the two PLUS creating a goo of aluminium+diamond which also doubles as thermally conductive void-fillers.

It’s just an idea with no backing data whatsoever. Really though, would this possibly work? Lapped surfaces + diamond&aluGoo = thermal conductive goodness?

26

The closer you are to the source of the heat the more important it is to have a material with very high thermal conductivity i.e. diamond, silver, copper (in that order). Of course there are cost constraints. This will act as a conduit for the initial heat removal from the source, get it out of there! Further out away from the source, as cross sectional area increases and “the lanes of the highway are greater”, We can switch to cheaper materials (notice how the best conductors of heat are also the most expensive). What I didn’t realize until eebowler pointed out in this post is that Aluminum actually has a higher heat capacity than copper or brass.(takes more heat to raise it in temperature) That means that later, after we have opened up a good heat path away from the led with say a copper star, it makes sense, both from a cost stand point and a sinking stand point to switch over to a material that is both cheaper and has more heat capacity i.e. Aluminum.
So having said that, if a flashlight has an led on a copper star fixed to a brass pill, screwed into an Aluminum flashlight body, in reality the copper star, brass pill is a high-conductivity heat path to the actual heat sink which is the flashllight body!

EDIT: Once the heat is in the flashlight body it can be conducted away through the users hand, radiated (black body radiation) or convected (moving air). In my thinking, I think of these 3 as a “drain” on the bottom of the “sink”. After all, even a sink will eventually over flow if there is no drain.

27

Not entirely correct. Depending on the Power-State of the CPU the fan can be slowed down. Actually while i’m writing this the fan in my PC is completly stopped and will stay so until the CPU throttles up again.
PC-Heatsinks are made for the dissipation of more than 100Watts of heat. If you using less power the mass of alu dissipates enought heat to the air in your PC-case even without active fan.

I think PilotPTK has it right. There is a small spot of heat (LED/CPU) where the heat has to be taken away fast. Therefor the copperstar/copperslug is better than the alustar/ bare heatsink. Once you have a wider thermal path the thermal conductivity gets less important and the thermal capacity / surface ratio gets more interesting. Therefor the fins for the CPU-Cooler and masses of Alu for the Flashlight.

@Scaru
You make me a diamond pill and i test it for you thoroughly. :money_mouth_face:

@Pulsar
Could probably work. Better than Arctic silver? I don’t think so. The carrier for the diamond particels is extremly heat resistant - thats for shure - but heat-conductive. Hmmm…. did you test it ?
LED-Star mounted to PC-Cooler (without fan) and thermometer at a fixed point on the cooler = highly scientific test equipment H)

Why not reflow-soldering/baking it directly to the heatsink/pill. Does someone know the composition of solder-lead and the thermal conductivity? Afterall the bare LED is also soldered to the Star for better heat-transfer.

Mo

28

copper has a higher volumetric heat capacity

iow, an aluminum heatsink must be larger than the copper one to benefit from aluminum’s higher specific heat capacity

edit, to avoid double post:

as far as soldering, thermal compound, direct mounting goes, here’s some food for thought

http://tangentsoft.net/elec/diy-hs.html

imo - soldering an emitter to heatsink is preferable because there isn’t a convenient way to lap and apply physical pressure to the emitter - but I would prefer to lap/screw down an aluminum pcb rather than solder it (and not just because aluminum solder paste is difficult to use).

29

In other words you are saying that because copper is more massive than aluminum, then for a given volume of each, copper will absorb more heat? If that is true then copper is better all the way. The only advantage for aluminum is cost? correct?

30

correct, but the calculation of how much more aluminum is required might be worthwhile. In reality, the difference would be less than the volumetric heat capacities imply, because a more massive aluminum flashlight would also have more surface area.

I think the cost of material is a factor, but the cost to machine is probably a bigger one…

even if cost weren’t an issue, a BTU shocker, TK70, etc. might be fairly impractical made of copper…imagine how heavy they’d be!

31

It is a lot easier and quicker to make machine aluminim pills. Unless you are pushing the limit either material I suggest would be fine. As most of us do here whatever material we have at hand that can be adapted to suit the application gets used. As far as cpu coolers go they can be either passive or actively cooled. The better passive coolers still use a copper base with aluminium fins.

32

My wife got new cookware and asked me to throw out the old. They are copper bottom. She doesn’t know it but they are stowed away in the garage!

33

I've seen several PC builds with passive CPU cooling, just a big aluminum heatsink.

http://www.modreactor.com/images/stories/reviews/cooler_master_hyper_z600/cooler_master_hyper_z600.jpg

Or this one:

http://img16.imageshack.us/img16/4782/thermalrighthr02cpucool.jpg

Here's the some info for that one, passive and active. The case still has some airflow from the other fans (they say)

http://www.hardocp.com/images/articles/1281274288qcyXNvNIQm_3_1.png

34

I was thinking about using some old copper? coins to sand down and fit in the p60 pill.

Will that help? (for heat-sinking I mean)

35

Most copper colored coins are not actually copper. A U.S. penny of recent vintage, for instance, is 97.5% Zinc and 2.5% Copper. The thermal conductivity of that coil will resemble Zinc far more than it will Copper. If you could find some coins made out of actual 100% (or close to it) Copper, then yes, it will help.

PPtk

36

Vintage coins would be a little rare and expensive. Copper isn’t that hard to get - just buy the electrical mains busbar, those 100-120A ones are thick and nice, and consist of almost pure copper.

They will be square shape though, but since we just need a small portion to wick heat quikly onto body, they should work just fine.

37

Nope, as I mentioned this was purely an idea with no backing data. I was hoping someone with the know-how would give some input for the feasability… or whether this is simply a stupid idea.

The carrier paste for diamond is exactly what I was wondering about - what is it made of, and how well would it do in heat transfer.

Lapping two surfaces evenly is definitely excellent for thermal transfer, now if only the remaining diamond+alu sludge would work as void filler, this method would make a simple and cheap solution for thermal management.

38

It would be interesting to know what percentage of heat is lost to the environment by what mode. I wouldn’t be surprised if a fairly big proportion of it is through radiation. Just plugging some numbers into the Stefan-Boltzmann law, using a P60-size cylinder, 0.1 as the emmisivity of aluminum, and 50 degrees Celsius as the temperature, I get about 0.7 W.

39

With even slight movement from pointing a flashlight around, the vast majority of thermal transfer away from the host will be due to convection. Conduction into the hand would also be significant. Radiation would be almost completely irrelevant to the calculation.

PPtk

40

hey guyz i dont want to come off as smart aleck but the only physical property (if you're comparing "numbers" of materials) relevant to the discussion is called thermal diffusivity α. Apart from the fact that material properties are always dependent of temperature T (and pressure P), thermal diffusivity is defined as lumped quantity

α = k / (ρ * cp)

see wiki: http://en.wikipedia.org/wiki/Thermal_diffusivity

Why is that? For the transient temperature field T(x,y,z,t) of a solid homogeneous body, say a chunk of pure copper, the 3-dimensional heat diffusion equation sufficiently describes the temperature field at any point (x,y,z) and at any time (t). And if you look at this famous partial differential equation, the only physical quantity relevant to the solution of the equation can be alpha (α) only, i.e. the thermal diffusivity, simply because there is no other physical quantity "left" in the entire equation. In other words you have to lump thermal conductivity (k), density (ρ), and specific heat capacity (cp), in order to get the "single number relevant" to the discussion/problem.

Basically the higher this number (HIGH thermal diffusivity) the faster high temperatures are disintegrated, i.e. the extended body cools down FAST by itself. The smaller this number (LOW thermal diffusivity) the longer the body stays hot by itself, e.g. streets and walls in spain. If you wanted to build/create a "heat reservoir", i.e. a chunk of material which you heat up in the evening and which should stay hot (by itself) until the next morning, then you'd be looking for a material which has

very high mass density (ρ),

AND very high specific heat capacity (cp),

BUT very low thermal conductivity (k),

because this will return an extremely low thermal diffusivity (α) number!

Back to the "problem of finding the best heatsink". How do you define "best heatsink"? If you define it as a material which transfers VERY FAST the generated heat away to the cooler parts of the material by heat conduction in solids, then yes the material should have the highest number alpha (α) possible.

Now it's up to you to look through the tables and compare the alpha's of Aluminum, Stainless Steel, Brass, Copper, Gold, Silver, Neodymium, ..

i am out :p