Liquid Cooled LEDs?

I'm trying to decide if liquid cooling makes sense for high power LEDs. I know lotsaluck had a project going where he would mill tiny veins into the LED heatsink and let the heat of the unit circulate cooling water. That makes sense, but it's too sophisticated for me to do myself.

I was looking at an LED glued onto its heatsink and thinking that the front half of the LED ought to be cooled too. If I housed a thin layer of fluid on top of the LED I ought to be able to get some cooling by convection.

The ideal liquid would be highly tranparent, have a high thermal capacity (thermal capacity = ability to absorb and hold heat) and a low viscosity so it would flow readily. Water is out because it could freeze. I'm thinking mineral oil or alcohol would be worth trying. There will of course be transmission losses, no way around that.

Has anyone here worked with liquid cooled LEDs? Is it worth pursuing? Comments?

I have also though about something like the oil they use in compasses but in addition to the index of refraction issue, there is the need to keep bubbles out of the bath regardless of temperature. You could have a diaphram that could equalize pressure issues but I am not sure if the gain would be worth the expense of trying to add some additional thermal relief. I actually doubt it.

On a related issue, I have imagined a light that was completely flooded in some type of oil for a dive light idea. you would not need to worry about the light flooding as the oil would not compress, anyway. The batteries in the oil would get heat from the LED and if the light case has a good thermal resistance, this could help in the problem I recently became aware of with batteries being cooled down too much. /ubbthreads/images/graemlins/icon3.gif

I think there are some practicality issues in all of these conjectures.

Unless you design a light with a radiator where the heated liquid can go to release heat away from the LED and design it where the liquid will actually go there with efficiency you would probably be better served to use a larger air cooled heatsink. If you have limited space and lot of high powered LEDs liquid cooling would be a better choice.

I would suggest looking at GPU (video card) liquid cooling systems for ideas as that is probably the closest available working model.

> in addition to the index of refraction issue,

Tell me about this?

>there is the need to keep bubbles out of the bath regardless of
> temperature. You could have a diaphram that could equalize
> pressure issues but I am not sure if the gain would be worth the
> expense of trying to add some additional thermal relief.
>I actually doubt it.

A simple approach would be to submerge the unit in fluid before sealing it. Anyway, if your fluid area is wide enough, bubbles ought to rise up out of the way of the beam unless you point it straight up, no?

> I think there are some practicality issues in all of these conjectures.

True, but we at this forum routinely go where no sane manufacturer would tread, do we not?

On a purely speculative level, it would be possible to design a light that could be "bleed" just like your brake lines or any other hydrolic system is. The LED die might not need the silicone and lens or dome configuration it typically sports so you could deal with the light and how it migrates through the internal optic system and then exits through the window. Whatever the fluid used is, compatability with other components must be considered as well.

I would think of problems would occur with a liquid filled head of a flashlight, breakage causing leakage, contamination causing optical problems, and extra weight causing the light to weigh more and be unbalanced more.

I could see having liquid pass through a radiator perhaps into the body of a flashlight to help increase cooling at the full length of the light better, perhaps even fins at the bottom like a radiator also would possibly be a solution. One test about the weight/balance issue is to take a flashlight and fill the head up with water and compare it to one that is empty. Chances are a light with little liquid would not benefit from such cooling as much as one with a larger head which would hold more liquid.

Imagine a maglite head filled with cooling liquid and the extra weight.

When you start looking into the thermal properties of liquids with respect to thermal conductivity and specific heat, I think water majorly outshines oils and silicone fluids. I once considered using Hg to cool my computer until I looked up the numbers! /ubbthreads/images/graemlins/icon15.gif Industrial heat transfer fluids other than water are usually selected for high boiling points or low freezing points.

Larry

I think liquid cooling of a star is a good idea if you really wanted to run it that hot, but with the extra space and weight you'd need to really get convection you'd be better off with an extra two or three LED's - it's more efficient anyways.

I think you hit the nail twentysixtwo. If the Luxeon emitter or star is generating too much heat without proper relief, it is likely being driven too hard for the package it is in and it is also likely that the package is too small to house reasonable cell capacity. The real expense of any type of fancy cooling could be overcome at a fraction of the cost if one went to clustered LED's. Granted, you can't do the same thing photon wise with the same amount of real estate but I think the ultimate answer will come from newer and more efficient LED's which simply make more light and less heat! Frankly I am not sure the K2 is going to offer up anything more than the ability do drive a LED harder and/or have it hosted in a less than ideal thermal package; neither of these are that big of a bonus in the small portable dedicated light packages that we are messing with. /ubbthreads/images/graemlins/icon3.gif

On comment here: if you fill the flashlight completely with water, you can not maintain the mag-style screw-unscrew focusability. It changes the inside volume of the flashlight and liquids don't compress well.

One way to solve this problem is to build a small balloon (or something like that) filled with air inside the flashlight. When screwing the flashlight's head, it would adjust the changing volume. They use similar kind of technique in oil-filled shock absorbers.

Best luck with liquid cooled flashlight project! I have also considered building one but so far it has been only a plan.

- Nereus

"but I think the ultimate answer will come from newer and more efficient LED's which simply make more light and less heat! . . ."

There was some research that apperared in Science Magazine a while ago -- which was not well interpreted in trade magazines -- where they used tiny holes to force the quantum mechanics of photon generation in LEDs.

The jist of it is that the holes are arranged to prevent the formation of photons that remain trapped in the medium and turn to heat. So they redirected energy that would have gone to heat into more light.

If memory serves, they obtained 5x more output and 15x more efficiency than their "unholy" LEDs.

I suppose you would still need thermal management though. Perhaps if you used a strongly conducting fluid you could use the current to the LED to circulate it magnetohydrodynamically. Then you could have a plastic light with a sealed pumpless coolant system that carried heat to an appropriately radiative material for disposal.

Just a thought.

Scott

Liquid cooling would make sense for a recoil arrangement (like the Pelican M8), where the LED is mounted near the exit lens and is directed backwards to the reflector.
Another advantage would be the reduced reflection loss at the LED dome and the window, but the refractive index of the liquid leads to a widened beam.

I have always thought of using a solid state Peltier device
to transfer the heat, it can run from the same supply as the LED

I can just see it now... honey can we pull over to a gas station...... my flashlight ran out of water and is about to burn up. Or how about.. I need to take my flashlight to the radiator shop and check the freon level it it seems to be burning the flesh off my hand now.

Heatpipes solve alot of the issues brought up.

Being around a dollar in small volume, and in the five dollar each range for 20, not a bad idea.

A fella will need to obtain the right variety, as they have different working temperature ranges.

If an LED is being driven hard enough that ordinary heatsinking and conduction aren't enough, then chances are you're in the very high power/output range where you'd be better off using an incandescent lamp.

[ QUOTE ]
TORCH_BOY said:
I have always thought of using a solid state Peltier device
to transfer the heat, it can run from the same supply as the LED

[/ QUOTE ]

And consume 10x the power of the LED to provide marginal improvements in thermal performance.

Just some thoughts on this:

The heatpipes are a good idea for the transfer of heat.

As far as the cooling goes, there are two ways to look at it, active or passive cooling.

Most all lights today are passively cooled. Adding active cooling such as a flowing radiator cooled system or peltier device would add weight and draw power from the system. For performance testing or static light positions, this wouldn't be a bad option.

For a passive cooling like adding fluid to the head of the light: This would be very problematic due to thermal expansion and contraction of the liquid. THe benefit would be interesting if you could have an exposed LED without the dome construction around it, the solution would have to have low electrical conduction for this. Also you would have to deal with the extinction coeffecient of the light, and the refractive index of the solution. Also depending on properties of the liquid, you might start getting unusual artifacts in the beam caused by the thermal gradient that would exist in the liquid itself.

A passive liquid cooled system would probably do better with a more effecient thermal pathway and heat sinking which would involve altering the design of existing LED's.

Actively cooled LED lighting would not be very effecient for lumen/power consumption.

I think IC's that monitor temperature like in the Gladius combined with more heat sinking is the way to go.

I think that the main advantage in liquid cooling would be passive cooling, like heatpipes to transport heat from the source to a heatsink. For most flashlights, this is not necessary, since there is continuous heat-conductive metal from the emitter base to the bulk of the metal in the body. Hetapipes might speed the transfer of heat, but whether they'll aid thermal conductivity in the long run is questionable.

If you had the emitters located somewhere in "thermal isolation" from a heatsink, then that would be an excellent case for heatpipes to transport the heat to a remote heatsink/radiator.

These are rather large for an EDC, but they do have a radiator to heatsink the LEDS.
http://www.electronicsweekly.co.uk/Articles/2005/07/29/35918/DutchtownilluminatedbyLEDstreetlights.htm