LED heatsink size

[dsvilko]

I've been trying to determine how large heatsink will I need for my 3xCree R2 bike light. To calculate this I wrote a small perl script that given the Vf, If and the number of LEDs can calculate the maximal thermal resistance of the heatsink on which they are mounted. The formula is from the Cree documentation. I am no expert (didn't know a thing about the thermal resistance a few days ago) so I am not completely confident that the results are correct.
If anyone is interested, here is what I get for the 'typical' values:
Vf=3.6V
If=1A
Tj_max=145C
Ta_max=40C - maximal ambient temperature

N Rh_max [C/W]
1 20
2 9.6
3 6
4 4.3

From the http://homepages.which.net/~paul.hills/Heatsinks/Heatsinks.html
we can determine the approximate heatsink volume:

N V [cm^3]
1 10?
2 25
3 50
4 90

I've got a 5x5x2.5cm=60cm^3 heatsink so, theoretically, it should be enough but it's really too close to the limit for my taste.

Can anyone tell me, from their experience, do the figures for the heatsink volume make sense? Would you be comfortable with 3 leds on a 5x5x2.5cm heatsink?


Here is the perl script:

#!/usr/bin/perl

#syntax: script_name Vf If n

$Tj_max=145;
$Ta_max=40;
$Rj=8;
$Rs=1;
$Vf=$ARGV[0];
$If=$ARGV[1];
$n=$ARGV[2];
$P=$Vf*$If*$n;

$Rh_max=($Tj_max-$Ta_max-$Rj*$P/$n-$Rs*$P)/$P;
print "$Rh_max\n";

 

[marcopolo]

I would go with at least 100cm^3 of metal as a heatsink. That way if I had a crash the lights would help break my fall and other large objects like boulders.!!

Are you being serious man? 60cm^3? as a heatsink? A big lump of sqaure metal?

Do you mean 60cm^2 - ie surface area?

Marco.

 

[dsvilko]

I would go with at least 100cm^3 of metal as a heatsink. That way if I had a crash the lights would help break my fall and other large objects like boulders.!!

Are you being serious man? 60cm^3? as a heatsink? A big lump of sqaure metal?

Do you mean 60cm^2 - ie surface area?

Marco.

No, apparently when using the real heatsink (as opposed to a flat sheet of metal) the parameter that best determines the thermal resistance is the 'volume' of the heatsink, not (as I also logically thought) the total surface area. By volume I mean width*length*height, regardless of how many fins it has (within reason). That is, it you are using it passively (without a fan). With forced airflow the things are different - now the important parameters are the airflow speed and the total surface area. It's all explained in the linked webpage.
As I would like to have a light that wouldn't burn out even if I rode slower than 40km/h, I plan to design with the former rather than latter rule-of-the-thumb.

PS: You are aware of the fact that a 100cm^3 is a cube 10cm on a side? Such an aluminium cube would weight 0.27kg, hardly enough to influence things in an event of a crash

 

[znomit]

No, apparently when using the real heatsink (as opposed to a flat sheet of metal) the parameter that best determines the thermal resistance is the 'volume' of the heatsink, not (as I also logically thought) the total surface area. By volume I mean width*length*height, regardless of how many fins it has (within reason). That is, it you are using it passively (without a fan).

The idea here is fins make no difference because two fins facing each other just radiate the heat back and forward, not out. Its a very conservative approach and valid for metal heatsinks inside enclosures. On a bike the airflow is by far the significant source of cooling so surface area is important.

 

[dsvilko]

Actually, with such a low temperature difference between the heatsink and the ambient temperature, heat transfer by radiation is not very efficient. Almost all the heat is transfered by convection (warm air rises so the heatsink makes it's own airflow even in a perfectly still air). With a flat sheet heatsink almost a third of the heat can be transfered by radiation but that figure is much lower for a finned heatsink.
I think there is a different reason why no. of fins makes little difference. Heatsink with more fins has a larger surface area which is good. On the other hand, because of the smaller fin spacing air flows slower which is bad. These two cancel out.

Anyway, as I have found a small (4cm) cooling fan that works with 5V (0.7W) i will add it to my heatsink. It's silent enough not to be noticable and should reduce thermal resistance to about 2C/W which is more than enough. It will turn on when more than one led is lit. With a fan I can not only cool the leds but also the batteries and the current regulators so I can make the light extremely compact (5x5x12cm with everything but the remote switch).
Now only if the current regulators have not gone out of stock the day I ordered

 

[marcopolo]

:laughing:

 

[marcopolo]

PS: You are aware of the fact that a 100cm^3 is a cube 10cm on a side?

Not on my planet it isn't.

 

[dsvilko]

Well, I was a bit closer
In any case, cube with a 4.64cm side is not that big.

 

[marcopolo]

It's an awful lot of extra mass when you consider a lot of people have been usiing triple&quad R2's with just a small slug behind the MCPCB of only around 6cm^3 coupled with thermal paste to a case roughly 6cm long by 4cm diameter weighing only 100g - tried and tested by many here.

And I've heard of no-one running 1A or under suffer from excessive overheating.

Sometimes you just have to throw theory out the window and follow the crowd!!! Baaaaaaaaa.!!!

Marco.

 

[dsvilko]

Sometimes you just have to throw theory out the window and follow the crowd!!! Baaaaaaaaa.!!!

Maybe you are right. Maybe it would work even without a fan. On the other hand, maybe not. My small heatsink will be the only metal part so it has to remove all the heat on it's own. There will be no flashlight body to help.

 

[LukeA]

Maybe you are right. Maybe it would work even without a fan. On the other hand, maybe not. My small heatsink will be the only metal part so it has to remove all the heat on it's own. There will be no flashlight body to help.

This is a bike light. It's not that hard to have hundreds of cfm of air passing by it usefully, and that's a hell of a lot more air than that CPU fan is going to move. Even though air isn't very efficient at absorbing heat, if you move sufficient quantities past your heatsink you will remove enough heat from it.

I think it would be worthwhile to look at the thermal resistance of the heatsink/air interface and work backwards from there.

 

[Steve K]

Maybe you are right. Maybe it would work even without a fan. On the other hand, maybe not. My small heatsink will be the only metal part so it has to remove all the heat on it's own. There will be no flashlight body to help.

I think your instinct is fairly good, but I'm not sure how reliable the numbers are. I've made a couple of dynamo lights, running around 4 watts or so. They don't require more than about 4 or 5 square inches of 1/16" aluminum in the airstream to get rid of the heat. Of course, the dynamo light has the advantage that there's no heat generated when the airflow is stopped.

By contrast, I've also converted a battery powered light to a 3W Luxeon. The housing is plastic, so I added a 3 or 4 square inch piece of 1/16" aluminum. If the bike is going to be stopped for more than a couple of minutes, I have to turn the light off, or the heatsink will become uncomfortably hot. Not a good thing.

I'd suggest trying out various sizes of heatsink on the lab bench in still air. That should give you an idea of the worst case scenario, without having to commit to a lot of expense. I'd avoid solid cubes of aluminum, choosing instead some simple aluminum sheet stock or extrusions that you can buy at the local hardware store.

good luck,

Steve K.

 

[snaps]

This is a Cutter tripple R2 run at 1 amp - it's my helmet light & has an 8mm x 34.9mm round heatsink that is coated in heat transfer compound fitted tightly into the alloy Marwi housing.
I my living room at 20C ambient at 1 amp it triggers the bflex 50C temp limit after 12 minutes but out on the trail it is barely tepid.
I also use quad R2s in the same housing with no thermal problems.