Help with heatsinking calc for a Cree Q5

Hi,

I'm designing a 6x Cree Q5 mountain bike light, and wondering what I can get away with in terms of heatsinking. Obviously its important when driving these at 1A, but weight is an issue. My design is 6 individual aluminium pods, like a section of pipe with one end closed, and the cree star sitting inside mounted on the back wall. It will be sandwiched between the optic holder and the aluminium with thermal paste inbetween. My orig design then had one big heatsink connecting all 6 pods together, but I'm playing with the idea of just using the aluminium surface of each pod as the heatsink.

Each pod will have around 24 square cm (3.84sqin) of surface area. Does anyone know a calculation for checking if this is enough for a Q5? Although there will be air flowing past the light as I ride, I want to try to design it for stationary 25degC ambient temps. I read somewhere that someone thought 3sqin per elec watt consumed was a good rule of thumb, but I don't know how accurate that is. Based on that, I would need 9sqin per led, if drawing 3W, which I don't have with just the pod. Hmmm.

How do other people decide on heatsinking? Thumb-suck? Trial and error? Calculations?

Thanks for any tips!

As always, this is turning into something a bit more complicated than I hoped... anyway, I'm making my way through this document, which seems to describe heatsink theory in relatively simple words... http://sound.westhost.com/heatsinks.htm

you could calculate most of this stuff till you are blue in the face, and the application of it via different methods , convection flow of heat, transfer, even the ACTUAL type of metals your using would toss the calculations right in the hopper.
things i observed , when fiddling with this stuff.
there is aparentally different types of aluminum that both transfer and dissapate better or worse, so you get one chart, and then you buy something, and you dont even know what it is. (aluminums with 1/2 the transfer rate)

they say (who is they?) you need about 7Sq inches for a hot running high powered led, but i have gotten away with as little as 4 ( every led is more efficent when run at it lower specs, more efficent means way less heat).
ex: i could run 20 crees with a dang inch of heatsinc at 120ma, or run 1 with 20cm of heatsink at 1200ma , and i would have much more light with the 20 leds.

copper transfers heat really well, so i get a copper end pipe, and it transfers heat , , , WHERE? around itself , well whoopitydo, i need the heat to depart the emitter (#1 priority) even if the whole rest of the contraption is the same temp at least the emitter itself is cooler.

i need to transfer the heat OUT into my hand , into the air, out of the unit, no mater how much heatsinc (i have a 2FOOT one) unless the air traveled across it, it was just a big lunk of metal.

i have hunks of POUNDS of copper, and given enough time the whle lunk is just holding heat and passing it around, sure the EMITTER is cooler, but things arent going very far. given enough time i have a big copper fry pan with a heater in the center

the smallest fan added to the smallest heat sinc , is more effective than a lame hunk of whatever they want to toss in a chart.

so be light, dont fret to much over Diamond transfer of heat, if your using a star item, you already spread the heat , and that is out of your control , you already lost a lot of transfer via the ceramics, and the star and the (etc etc) . Thermal speed has its biggest nessisity at the pont of thermal creation, once you HAVE a spread, speed of transfer is not as important, as long as it can go somewhere.
if you have emitter items, a copper spread, even thinner is a good start, but then you need to DISPERCE the heat, not TRAP and MOVE it around the copper, it needs to LEAVE the heat sinc.

someone said i was CRASY with that theroy, yet i was running a CPU more overclocked than they were , with the cheapest LIGHTEST heat sinc, because it was Designed well. Spread and Disperce , and remove. copper spread, aluminum disperce. hey i got a aluminum pan, and it disperces so much heat it takes way more heat to cook the same thing? why ? it aint in the CHARTS

it doesnt matter if you have 3 inches , or 50 inches, deep inside the gate its still HOT, so more is always better, and there is a level of practicality, and reality that you can have with the available parts, the more the merrier, and far less than to much will still work for Many hours, far to much will be around for longer than the technology itself.. (the led would like things much cooler, but there is only so much to do before its impractical)

For example, METAL beats the snot out of any thermal transfer glues , or sauces, so you can get $7 super a gram sauce, and i can drop 2c worth of solder in there, and blow it away instantally.
SO
locking down the metal base of the emitter, to the other metal item is a great start, put the sqeese on the thermal glue so it is VERY thin, and has much metal to metal contact.
BUT
so there is your first problem, when these devices make metal to metal contact, they CONNECT so bam ceramics, and goops, and isolation, and its insulation, without insulation you have connection, and THEY didnt put diamond in at the first point
(only so much you can do)

with seperate tubes ,and metal to metal (mostly) connected emitters, you can potentially get a much better lockdown of the intital transfer (the most important one) , but that means they still have to get OUT.

does it need to be waterproof?
for example, i can run a led Underwater almost 2 times the power it would normally fry with, but water not so good on components, Cap it, seal it, and close it, and you lose your most valuable thermal components immediatly, transfer OUT.

heat rises, and a chimney effect (rarely applicable to bikes) can assist greatly in air movement that moves it out.

the Heat sink calculator is the one in your BRAIN, that has the ability to ignore 500 pages of technical mumbo jumbo, and get the heat, off the main piece, and cool the heat sinc itself, without causing electrical problems (short). and realising that you have to work within the components your already given, and avilable metals, and logical assembly of the parts for waterresistance, and heat removal.

a FACE of leds that are kept from the elements, and a Backside that is fins, and some simple method of transfer from there to here.

Use the Force luke.

I'm planning to mount 8 MC-E's on a copper block 90x60x60mm (l*w*h) with a 60mm fan cooling it. When first introduced to LED's, this would have (IMO) been suicide. Now I've had a chance to fiddle with some LED's, I acknowledge that this design is probably pushing the envelope, but it should be fine. On a MTB, with forced air flow, and normally cold air at that, you'll be absolutely fine with (what you think is) a ridiculously small amount of heatsinking. My personal favorites are a lot of long fins, as that gives some kind of buffer and also more surface area to dissipate heat.

You guys are making things way too hard for yourself.

Here's a simplified rule of Jonathan's thumb.
1. A stock Mag head can sucessfully dissipate about 10W of heat in stationary position and stale air.
(almost everyone knows mag that's why I use this example).

2. Moving air flow double that heat dissipation.

Now if you must get technical then take a look at this thread.

ccoutts: Pictures are worth a thousand words. If you have model your design I can run a thermal analysis for you.

VidPro: I miss your post my friend. It's a pain in the A to read, but makes sense if you actually read and think it out.

Jarl: 8x MCE @ 1A is ~ 100W. Your copper block will not have enough surface area too cool such beast. You need more FINS.

Current breeds of CPU heatsinks are designed for over 120W with the fan zipping away @ 60+ dB
Now if I were you I'll get a Core2Extreme HS which is designed to keep the 200W thermal beast at bay.

VidPro: I miss your post my friend. It's a pain in the A to read, but makes sense if you actually read and think it out.

I been working. on more other stuff.
but i did get the new Acrichie leds working with high voltage DC , i put 10 of thier doubble emitter things into lighting up a room, with this DC method, and its a good thing i put in a LOW mode at 20W of total AC power, it lights up the whole room, replaced my 2x32W florescent circle lights.

i might have pics of it, although its UGLY (as usual) its very functional.

Thanks modamag, for the offer. I've sketched a rough drawing so you can get the idea. The Q5's and optics are behind a glass cover on each pod, and the pods are bolted to the heatsink with thermal goop inbetween. I added a heatsink as a "backbone" to the whole thing, but obviously the length of each fin is the only variable dimension. I selected 12mm, just because this gives around 7sqin of aluminium surface area per pod, which was a rule of thumb I read somewhere. So I guess the question is, how small can I make the fins, to keep my Q5's from overheating. There will be air flowing past most of the time, but I'd like to design to 25degC still air if possible. Thanks heaps for any advice!

Oh yeah, thanks VidPro for your essay... didn't really understand a lot of it but got the general idea you just trial/error it.

For useful demo heat sink design software, both forced and free are try
http://www.frigprim.com

If you use the demo version you will need to scale your results to the size you need. I found the demo version very useful for determining the fin spacing versus height and thickness.

matthew