3*XML-U2 diving flashlight mod - Intel 775 heatsink??

stroodl

Newly Enlightened
Joined
Mar 18, 2012
Messages
10
Hi,
I want to mod Technisub VEGA 2 with three XML-U2's. I wondered to use Intel 775 CPU heatsink. The question is, whether it is possible to cool those leds driven at 2.5A each with this heatsink in a full plastic body? Unfortunately it is impossible to transfer the heat to water. The heatsink (any) would have to absorb all the heat produced in about half an hour.

Is there any other way how to mod this flashlight with these leds?
http://bludive.pl/data/include/cms/Technisub_Vega_2.jpg
http://www.ixbt.com/cpu/images/s775-cpu/s775-cooler-bottom.jpg

Thanks,
stroodl.
 

illumiGeek

Enlightened
Joined
Oct 28, 2003
Messages
499
Location
Kapolei, Hawaii
CPU heat-sinks work by transferring the heat energy to air. In a sealed plastic body there will be no place for the heat to go. The air sealed in the light body will get hot, effectively negating the function of the heat sink. With a plastic bodied light you are pretty much SOL.

Halogen bulbs do not mind running hot (in fact they have to for the Halogen scavenging cycle), so they are okay in a thermoplastic body. LED emitters do mind running hot, so there must be a way to transfer the heat out of the body. You would have to mod the body to get the heat out into the water or you risk damaging the LEDs.
 

Klem

Enlightened
Joined
Mar 26, 2010
Messages
581
Location
Perth Australia
Unfortunately not.

All that fancy CPU heatsink will do is turn your torch into a fan-forced oven. Everything will be more evenly 'cooked'.

The only way to run 24Watts in a plastic housing is to have an efficient, seamless metal-to-water thermal path. Either have a seamless metal path to the outside, or bring the outside to the back of the emitters via cooling tubes.
 

illumiGeek

Enlightened
Joined
Oct 28, 2003
Messages
499
Location
Kapolei, Hawaii
Hi,
... Unfortunately it is impossible to transfer the heat to water. The heatsink (any) would have to absorb all the heat produced in about half an hour. ...
Just to clarify, heat sinks do not absorb heat, they transfer it from the source to something else (usually air, but it could be water or some other coolant).

And it's not "impossible" to transfer the heat to water. It would be difficult to do while maintaining the integrity of the waterproof seal, but it could be done. Dive lights are not completely sealed, they open up to change bulbs & batteries, and switches often pass through sealed channels. You would just have to mod the body to pass the heat out to the water.

I do not know how you plan on replacing the halogen lamp assembly with an LED array, but perhaps you could pass some thin aluminum bar stock out through a slot cut in the body. The slot could be sealed with a heat resistant epoxy or silicone sealer. The bar stock could be folded to extend down the side of the body to increase the surface area in contact with the water.

There are other ways it could be done as well, you just have to use your imagination to come up with something that would be compatible with your internal mod.

EDIT: @ Klem, cooling tubes... interesting and imaginative. Hadn't thought of that. It's a cool idea (pun intended)! Round holes would certainly be easier than cutting slots...

EDIT2: I wonder if convection would move enough water through the tubes? As long as you were swimming the motion would move water through, but if you were to hold the light stationary to examine something for an extended period... then again, water is 25 times more efficient at absorbing heat than air, and the heat transfer path would be more direct, so perhaps you wouldn't need much flow. It's really an intriguing idea.
 
Last edited:

Klem

Enlightened
Joined
Mar 26, 2010
Messages
581
Location
Perth Australia
EDIT2: I wonder if convection would move enough water through the tubes? As long as you were swimming the motion would move water through, but if you were to hold the light stationary to examine something for an extended period... then again, water is 25 times more efficient at absorbing heat than air, and the heat transfer path would be more direct, so perhaps you wouldn't need much flow. It's really an intriguing idea.

Yes, the idea of cooling tubes was discussed a while back among the regular posters, all contributing ideas based on a combination of their torch-building and diving experience.

You don't need to worry about the efficiency of water around a torch. Even small buckets of essentially still water when testing you can see the particulate slowly moving around the heat transfer areas. When diving, there is no such thing as absolutely still water rendering the heat sink essentially, 'infinite'.

The bottom line is while I am interested to see a solution here I would not recommend a plastic housing for anything more than 12Watts of LED. Especially one which houses both the emitters and batteries in the same insulated container (i.e. not a canister design). Thinking of the time, effort and risk of flooding here it will likely be more than if you start from scratch with a metal housing.
 

stroodl

Newly Enlightened
Joined
Mar 18, 2012
Messages
10
Thanks for the replies.
I know how the heatsink should work but it definetly has the ability to absorb some heat, its still a piece of copper and aluminium. The problem with transferring the heat to water is, that the heatsink must be removable so I can easily change batteries. I didnt come up with some easy and smart solution how to make easily removable heatsink perfectly transferring heat to the parts epoxied in the body.
I can use a copper tube like 2-3cm in diameter going in the centre through the whole body flooded with water on the back side and with copper base with installed LEDs on the other side. Couple of 18650's in parralel around the tube. This seems to be the best cooling possible - 2mm copper plate with fresh water on one side and 3 XML-U2's on the other.. What do you think?
edit: some sketches
 
Last edited:

Klem

Enlightened
Joined
Mar 26, 2010
Messages
581
Location
Perth Australia
Yep, that would certainly work.

Excellent idea with the cooling 'tunnel'.

No need to go too wide with the diameter of that tunnel as moving water against a copper base will be very efficient. I say that because I am thinking the wider you make that tunnel the less space you will have inside for batteries, switch and driver.
 

SmokedCPU

Newly Enlightened
Joined
Jan 9, 2010
Messages
85
Location
Quebec, Canada
Hi

This look like what i made some time ago as light in my old Farallon scooter.
Body is aluminium, so no heat problem.

I run it from a separate battery and light is trigger by reed switch. I bypassed one of the die to get a lower working voltage, and i supplied a smaller flat cpu fan on the led supply. however, assembly is not as sturdy as i would, but it work. Desing requirement needed flat assembly, i did not have space for a reflector.

4000LumensS.jpg
 

stroodl

Newly Enlightened
Joined
Mar 18, 2012
Messages
10
Switch is located on the back plate of the body, I'd like to use it to connect the (-) poles of the cells to the copper tube and use it as 'ground', while (+) will go direct to the leds. (isn't there any problem with the - pole connected to water?) Driver could be located on a copper plate soldered on the tube between two cells. Could you please recommend me a proper driver? I'd like to use 4-6 18650 or 26650's (optionaly) in parallel and 3 LEDs i parallel at about 8A total.
Thanks a lot, Ondra
 

DIWdiver

Flashlight Enthusiast
Joined
Jan 27, 2010
Messages
2,725
Location
Connecticut, USA
I think the ID of the tube does need to be large diameter. Otherwise the water will just sit there, not flow through and carry the heat away.

What about a solid copper rod? It wouldn't have to be that thick - a rod 0.5" diameter and 4" long conducting 24 watts would only drop about 48 degrees C. That's reasonable. You can get a foot of that rod for $16 at McMaster-Carr, which isn't known for great prices. A 3/4" rod would be even better, but more expensive.

Or how about a heat pipe? Effective thermal conductivity can be vastly greater than copper. You can buy them all over the place now, and they aren't that expensive. Attach one end to the plate your LEDs are on, the other to something that sticks out into the water at the back end.
 

stroodl

Newly Enlightened
Joined
Mar 18, 2012
Messages
10
There is 10 times bigger copper surface to transfer the heat to water with tube than with rod. I think both solutions would work perfectly under water, but would it be possible to use the rod-coolled LEDs for some time out of the water? I can also use a 5cm diameter tube and solder some copper heatsink in it - there would be a huge copper surface to transfer the heat to water/air(?).
 

Klem

Enlightened
Joined
Mar 26, 2010
Messages
581
Location
Perth Australia
For above water in air I imagine the 'dead-end' nature a hollow rod would not be efficient enough to give you very long at 24W. That internal fan will still be quite useless being a fully sealed dive-torch and it would pay to use a driver with a temperature cut-out (like a Taskled) to protect your emitters from damage.

There is such a big difference between water and air that I am thinking you might get 2 minutes at 24W in air. Obviously it's all guesswork but I can see that heating up pretty quickly on full power. You could with the right driver have a lower mode for above-water.
 

DIWdiver

Flashlight Enthusiast
Joined
Jan 27, 2010
Messages
2,725
Location
Connecticut, USA
There is 10 times bigger copper surface to transfer the heat to water with tube than with rod. I think both solutions would work perfectly under water, but would it be possible to use the rod-coolled LEDs for some time out of the water? I can also use a 5cm diameter tube and solder some copper heatsink in it - there would be a huge copper surface to transfer the heat to water/air(?).

Actually that wouldn't be a huge area, compared to 24W and natural convection air cooling. You'd want about 75 sq. in. (484 sq. cm.) and a path for the hot air to rise up through it in order for that to work very well.

I agree that either would work well in the water. The only reason I suggested the rod is because it's so much more compact. The problem isn't getting the heat from the copper to the water, that works REALLY well. The problem is getting the heat to the metal/water interface, then getting the water (and heat) away.

There's definitely some validity to using the metal to soak up the heat for a little while in air:

Copper has a specific heat of 0.386 J/gm-K. Assuming you will allow a 35K temp rise on the rod, have 20W of heat flux (that assumes the LEDs are just under 20% efficient), and the hypothetical 1/2" dia x 4" long rod weighing in at 115gm. That gives you 0.386 J/gm-K * 115gm * 35K = 1554J. Divide that by 20W and you get 77.7 seconds.

On a volumetric basis, water and aluminum are similar to copper (actually water is a little better). But that big tube full of water might have 10 times the volume, so you could run it a good long time before having to change the water. And that's with NO heat being transferred to the environment beyond the end of the tube!

Interesting what math can show you, eh?
 
Last edited:

Klem

Enlightened
Joined
Mar 26, 2010
Messages
581
Location
Perth Australia
77 or 120 seconds...DIW's empirical logical positivism, or Klem's 'Rule of Thumb'. My money's on DIW. My hat goes off to you Sir.

The point being, you would be wise to keep this as a below-water torch only.

Or...start the stopwatch when you turn it on!:popcorn: :poof: :mecry:
 
Joined
Feb 1, 2008
Messages
889
Location
US
hmm ... have you calculate weight of that rod or tube ( Al or Cu) and compare it with ''neutral '' buoyncy of whole torch ?

Otherwise ...look on ''Hanchan '' projects
 
Last edited:
Top