Heatsinking - is thicker better?

Bleeding away the excess heat from our beloved and inefficient LED-lights is vital, we all agree. There seems also a common agreement that a huge metal heatsink housed in a big aluminium chunk serving as a body will do the trick and you've done it, the perfect temperature-effective LED-light.

I dare say this is wrong, at least partially. Let me explain my heretic thoughts:
We can agree on the huge heatsink on which the emitter is to be mounted. This is a necessity for quick heat removal. On the heavy body we disagree. A huge mass of metal can absorb heat nicely, at least in the case of aluminium. But when heated up, the efficiency of this heatsinking will suffer greatly.
IMHO the most important part in heatsinking is a good path for the thermal energy from the LED to the heatsink to the body and from there on ... to the air or flesh. A heavy body can offer temporary relief for short runtime applications, but will be useless for longer burn, or let me put it better: the big chunk will be just as effective as thin tube of the same size in thermal relief.
For this thermal relief there are basically two pathways: transfer the heat to the air by increasing the contact surface, e.g. with cooling fins, knurling etc. ... or ... transfer the heat to the human body by maxing out the contact surface, too, but with the opposite solution: by making it smooth.
Anyway, for long-run applications the deciding factor is heat removal from the flashlight to the surroundings, a thick massive body is not really necessary to accomplish this, the most important thing would be a good pathway from the LED via a heatsink to the body and away from it quickly, much quicker than with a short-burst-style light that can take its time bleeding away the heat from the thick and hot body.

Conclusion:
Short runtime applications would benefit from a thick body as a "thermal energy buffer"-heatsink whereas long runtime applications would not need a thick body since they must be able to bleed away all the heat that is created in the same time, and the limiting factor is not the aluminium body but the transfer from the body to the surroundings. Those lights should be built differently then and could be noticably lighter and smaller.

Just had to voice my thoughts regardless of my technical ignorance and stupidity /ubbthreads/images/graemlins/grin.gif, thanx for reading.

bernhard

But you could go too far. You still need enough thickness to let the heat spread over the surface easly. Imagine a luxon on a typical slug, mounted in a foil tube. The slug will give a good pathway away from the LED, and the foil has lots of surface area. But there is a 'trafic jam' where the slug meets the foil, even if the foil and the slug are one piece.

Of cource this is an extream example, just to ilistrate the point.

Yeah, heatsinks are most effective when there is a way to dissipate the heat. I would imagine that is why many LED flashlights (particularly Luxeons) are made of aluminum. The body of an aluminum flashlight IS the heatsink. The aluminum slug attached to the LED acts as a bridge (to spread the heat to the body). Making this slug thicker won't do a whole lot of good since it doesn't dissipate heat well by itself.

Surface area is the key to effective heatsinking. Think about a computer CPU. You don't stick a solid block of metal on the CPU - you use a thin plate of metal with lots of fins on it.

Dan, yes, of course. that's what I meant by "good path", sorry for my poor wordings. This minimum thickness should be covered by the minimum necessary to achieve a certain physical strength of the body though ...
bernhard

P.S.: a good example is the Arc LS - a tiny light that still manages to bleed away the heat for a very long runtime, and does this best when held in a hand. more metal wouldn't do good here.

i think we all agree that using a bigger heatsink = better.

i agree that surface area to mold the heatsink with the flashlight body is the most important thing. in all my heatsink designs this has been one of the priorities.
but more mass will help also. maybe not as much for the long runs, but it will help initially and for the short runs.

so while a good path to the outside is of primary importance, why not have as much mass as seems reasonable too?

furthermore, mass will help at some point. if you have a small light (firefly, for example) it is so small that if there is a good thermal path from the emitter to the outside, then the whole thing will heat up quite fast.
having more mass to suck in the heat will help, because the body may not be able to dissapate this heat as fast as heat is being created. therefore more mass will help.

Wouldn't the conductivity of the material being higher help more than having a large mass?
Isn't copper a much better conductor than aluminum?
I have a 2c MAg that has a thin copper heatsink with alot of contact with the barrel of the flashlight.The sink is very effective in transferring heat to the flashlight.
This will probably get moved to the mods foum , but has anyone had any experiences with the whole light being made of copper?

copper is indeed a better conductor than aluminium.
however, aluminium is better in dissipating heat.

Keissling,

I agree on many of your points. Having a large body provides thermal inertia which works great until you saturate it. Once saturated, it's virtually worthless as temperature will start increasing rapidly again. The key is to remove this heat from the mass that's absorbing it. One way to do this without forced air is to increase surface area, i.e. fins like in the KL3. But then you get back to the large body which I don't like because I won't carry the light if it's too big. Another way to remove the heat from the flashlight body is to use the human's cooling system to conduct the heat away from the flashlight body. I think this has been discussed many times before but it's a point that is greatly overlooked. If you take a small bodied light like the arc and left it on in free air, it would get very very hot once the body saturates. However if you held it in your hand, it would run cooler. In this case, fins are worthless as it reduces the contact area between the light and your skin. You actually want a very smooth finish to provide maximum contact between light and hand. Mass does help but if the mass saturates, then it's worthless. Of course the larger the mass, the longer it would take to saturate but you could go on ad infinitum and come up with a flashlight the size of a hummer and say "now that's a perfect heat sink" /ubbthreads/images/graemlins/grin.gif

CM

Led-Ed,
From what I gather in CPU heatsink discussion forums, copper is almost the best material to use. Silver would be better but....
Copper VS aluminum Copper absorbs heat faster than copper but aluminum radiates the heat away faster than copper. If blasting the heatsink with 60+ watts, copper is used to conduct the heat away faster than aluminum. Getting rid of the heat in the copper is another story. Some CPU heatsinks have a copper base with aluminum fins as a hybrid.
Copper is very, very heavy and soft as well. I have a copper heatsink in my computer, Thermalrights are heavy! My video card only kicks out around 20-25 watts and an aluminum heatsink effectively cools it. Both of those heatsinks are assisted with 80mm case fans blowing on them. Generally speaking, if you have enough mass to sink the heat to, 30 watts can be cooled without a fan.
When LED lights start hitting around 10 watts, the larger D cell lights will need to worry about heat, for now it is no big deal. If I was going to design a say... 4 123a 12V 10 watt LED flashlight, it would be a hybrid. Copper slug soldered to an aluminum finned tube for the flashlight body. The copper slug would quickly conduct the heat away to the aluminum finned tube to dump the heat with a much larger surface area.
Fins, pins or rod heatsinks are worthless if stuffed inside a flashlight (no air flow) It is better to have a solid heatsink that resembles the hotlips.

My own 2c is to suggest 3 stages to bleeding off the heat. First stage is the massive heatsink on which the LED sits. This will help in short runs. This heatsink should be well connected to deep aluminium fins in the head and a very flat knurl in the body. These are the second and third stages.

I have found that even the Arc-LS 500mA hybrid does run very hot even when the heat is sunk to my hand. Some days, it just becomes plain impossible to hold, not just uncomfortable. Hence my suggestion to have fins in the head as well.

The main point to realise is the limit to which a block of metal stuck inside a flashlight body can absorb heat. To effectively keep cool that heatsink inside the flashlight must dump the heat to an even bigger sink; the environment. Going through our hands is one way. The only other possibility is via the surrounding air. One more, though impractical, way is evaporative cooling. It is only impractical because you have to carry a water tank that needs refilling every so often. But the Harrier jet does it, allowing for a full minute of hover time without the turbines melting.

On another note, we have been using solid blocks of metal because it is simple and easy to make, but has anyone heard of heat pipes?

[ QUOTE ]
Steelwolf said:

On another note, we have been using solid blocks of metal because it is simple and easy to make, but has anyone heard of heat pipes?

[/ QUOTE ]

Yes - after two years of designing embedded systems, you tend to learn about every method of moving heat from a confined space.

The operative term being 'moving', which is what a heat pipe will accomplish. In the end the heat must still be dissipated through (in our case) a hand or the ambient air.

So we still must strike a balance between cooling the LED and reducing the bulk of a flashlight. I've been playing around with stirling engines as a method of absorbing heat, but it makes for a really clumsy looking flashlight! /ubbthreads/images/graemlins/thinking.gif

Roth ... I do not want as much mass as possible because I do want my light as small as possible.

CM ... exactly! thanx for wording it so that everyone might have a chance to understand it. this technical stuff is a bit tough on my English ... /ubbthreads/images/graemlins/icon15.gif

Copper-aluminium hybrid heatsinking sounds cool, and couple that with Steelwolf's 3 stages including a bare body with a finned head, this combo should be quite efficient.

so ... overall it would be:

high power and short runtime --> big massive light, good to go until thermal saturation is reached, worry about cooling of the whole thing later

long runtime needs good thermal conductivity of the whole light, especially in the last step, the transfer from the body to the enviroment. the mass of the light is of minor concern here, conductivity rules. the restricting factor would be the amount of energy that could be transfered to the enviroment, that would be the max. power at which the light could be driven in long run applications.
tricks include: copper-alu-hybrids, flat body, fins on the head, big contact area of all involved parts etc.

bernhard

The optimum cooling for a given body size would be achieved, if the total surface and the heat sink is on the same temperature. I think the L4 is a good example, that we are far away from this situation (the head is much hotter than the body). Therefore I think, that a thicker sink and body material, will lead to a more equalized temperature and to a larger effective cooling area (by radiation or convection).

PeterB

The hard anodize seems to insulate the KL4 head from the body. I think if they left the threads unanodized, you'd get better heat transfer from head to body. The McLux is a good example. My McLux head and the body seems to be at a much closer temperature to each other than on the L4.

CM

I think all of you have very good points made here.

But, what needs to be done, is to find a good equilibrium.
The large heat sinks are good, becuase they have alot of surface area touching the inside of the body of the light. That makes the heat transfer to the body better and faster.

But, at the same time, as said above, that large heat sink will actually HOLD the heat in the light.

One of the very best designs out there, is the Mclux.
The head IS the heatsink, and the flutes (grooves) on the outside are GREAT for trasferring heat to the outside air. Because they make the light have alot more outside surface area.

The problem with our ussual M*g mods are, they have thermal joints. IE the sink to the body.

And the head of a m*g does basically nothing. Sence the body is only transferring heat to the heat by the threads, which isnt very good at all.

ah CM ... I made the same observation! my McLux and L4 (both 5W) behave differently with the McLux getting much hotter at the body than the L4 while both heads are quite hot. if the HA acts like an insulater here, it does the job remarkably well.
BUT wouldn't that mean that LED-lights should not be HA then? would bare aluminium lights transfer the heat better to the air and hand?
bernhard

btw ... when holding my McLux BB750 5W for a longer time I sometimes whish it was an L4 ... this thing gets *very* hot, even at the body! technically speaking this is a very good design, but, ahem, well ... I am just a weenie. /ubbthreads/images/graemlins/smile.gif

.

.

Hmmm, does black anodize perform better than the natural color type III anodize then? Black is supposed to be a better abosrber of heat IIRC. Seems like the ones without anodizing between head and body transfer heat better. My Arc driven at 667mA transfers heat better to the body (2AA) than the L4 does. The L4 can get VERY hot, with the temperature differential between it and the body is quite high. Same thing with my McLux PR bare aluminum runnning a 5W LED at 750mA.

.