Why no liquid-cooled LEDs?

smokin

Newly Enlightened
Joined
Oct 18, 2010
Messages
7
Location
Phoenix, AZ
The best metal heat sinks only draw heat from the back side of an LED, leaving all the front-side heat on the chip for cooking it. Wouldn't it be possible to have the LED die immersed in heat-conducting liquid to draw away ALL of the heat?
 
The best metal heat sinks only draw heat from the back side of an LED, leaving all the front-side heat on the chip for cooking it. Wouldn't it be possible to have the LED die immersed in heat-conducting liquid to draw away ALL of the heat?

Most liquids have really awful thermal conductivity. To get good performance (comparable to metals) you need to pump them - requires a pump, power, and more heat emission area.

Most liquids affect the LED optics. Losing even 1% light costs more than you're likely to gain even with pumped water cooling (or any exotic low-dispersion liquid you could think of). And how do you make with work with a reflector?
 
Immersing the LED in non-conductive liquid would not only steal lumens but also ruin the beam characteristics completely. Without circulation it would also eventually boil off and turn into endless, annoying moisture drops in front of your LED.

I thought initially you meant simply using a water-cooled heatsink like a personal computer's water block. In such a case..

LEDs are more efficient the lower they are driven. When you drive LEDs at their limit their efficiency is no better than HID lamps, yet unlike HID lamps you have overheating issues to contend with in addition to power consumption once you get to this level.

I guess the answer would be that the total cost of equipment, labor, maintenance and the risk of leakage if using lower cost conductive liquid far outweighs any advantages liquid cooled LEDs would have, if any, over HID lamps.

Maybe it would be a fun hobby project but there are good reasons why it hasn't been done yet.:nana:
 
The best metal heat sinks only draw heat from the back side of an LED, leaving all the front-side heat on the chip for cooking it. Wouldn't it be possible to have the LED die immersed in heat-conducting liquid to draw away ALL of the heat?
in short no.
almost every device is a compromise between price\complexity\practicality.

in theory it is possible to build what you ask, but it wont be no where near practical\simple\reliable.
 
Last edited:
this topic has been covered several times, look below at the similar threads near the bottom of the (full) page.
Think of liquid cooled computer processors and read up on them and then consider the huge drawbacks. You will not have a handheld self contained liquid cooled light, liquid cooling to be worth the effort will require external hoses and power cables and a backpack to carry it all. I doubt many of us want to be a ghostbuster for the expense of having the brightest light.
 
this topic has been covered several times, look below at the similar threads near the bottom of the (full) page.
Think of liquid cooled computer processors and read up on them and then consider the huge drawbacks. You will not have a handheld self contained liquid cooled light, liquid cooling to be worth the effort will require external hoses and power cables and a backpack to carry it all. I doubt many of us want to be a ghostbuster for the expense of having the brightest light.

http://www.candlepowerforums.com/vb/showpost.php?p=3451672&postcount=31

you sure about that? :naughty:

It'd certainly be possible to put that into a spotlight form factor :) watercooled CBM-360 in a handheld spotlight form factor certainly wouldnt be impossible.
 
http://www.candlepowerforums.com/vb/showpost.php?p=3451672&postcount=31

you sure about that? :naughty:

It'd certainly be possible to put that into a spotlight form factor :) watercooled CBM-360 in a handheld spotlight form factor certainly wouldnt be impossible.

It may be possible, but runtime would be so short that if it was just a massive block of aluminum with a fan and no water cooling it would probably work the same anyway. When you approach such levels of output requiring watercooling you are probably going to have to go to large lithium rechargeable batteries and push them to the limit for just a few minutes of light. I don't think anyone ends up really using a light with 5 minutes of runtime, it ends up only being for show and nothing else.
There is a thread where someone is running 8 SST90s off 18650s direct drive and only using a heatsinked aluminum body light.
http://www.candlepowerforums.com/vb/showpost.php?p=3453160&postcount=33
 
Last edited:
The Thermal Conductivity of water is only about 0.6, compared to about 200 for aluminum. Even cruddy stainless steel is about 16.

Other common liquids are all worse than water. But even .6 is much better than air (.025) and about the same as common thermal paste/compound.

So, water comes in handy as thermal paste substitute for quick easy experiments. But filling a copper pipe with water does not increase the total thermal conductivity much, unless the water moves due to convection or other flow.

But I do think there are some special situations where water could be useful even if it is not moving. Copper is the best common conductor, but it is both expensive and not readily available in bar forms. Copper pipe and fittings and solid copper wire are very available, both new and salvaged. Nesting 1/2, 3/4, and 1-in copper pipe would increase thermal conductivty - filling such a nested arrangement with water and sealing might be a good practical way to get good group heat flow. Even better, pack a copper pipe with new or salvaged #4 or #8 AWG solid bare copper, available by the foot from Lowes or HD. Filling with water to spread the heat might be much more convenient than trying to figure out how to solder it all together inside the pipe (and much less dangerous).

The makers of power LEDs design for the heat to flow out the backside. The heat does not care which way it flows -- it only cares how easily it flows. It is much more important to do a great job of getting the heat out the back than worry about the tiny amount of heat at the front (if the back is well cooled).
 
Many factory have used liquid to cool the led product. You can search in the internet. But as my opion, using liquid has many unsafety points. So I suggest not use this mothod. Otherwise, service will have more pressure.
 
Heatpipes are a very effective way of using liquid cooling - the liquid/vapour can move the heat away much faster than conduction in any solid !
 
Friend of mine at work patented a fixture that the LED would sit in that was essentially hollow clear plastic optics that doubled as a heat pipe with non-thermally conductive liquid in it that convected around to cool the LED. Sounds like what you want? Works pretty great but not economically feasible.
 
Sounds just like what I was wondering. Also answers my OP question with "not economically feasible", sounds too familiar!

Even if immersive liquid cooling did work, I'm still going the right way on my next light project. That's by adding more LEDs running at the max efficiency point, instead of pumping more current through them.
 
http://www.candlepowerforums.com/vb/...2&postcount=31

you sure about that? :naughty:

It'd certainly be possible to put that into a spotlight form factor :) watercooled CBM-360 in a handheld spotlight form factor certainly wouldnt be impossible.

My liquid cooled led array work like a charm, 12 XP-gs stacked in a 3x4cm area are driven at 2A and maintain 26C above ambient. Batteries are 11.1V 8200mAh lithium polymer, enough for 20+mins flight using the light full time.

I used liquid cooling, because using a large heatsink where the light is placed wasn't possible, so I moved the heatsink where practical and water carry the heat to it. Also, water have a high specific energy, so the light can be ON without airflow for longer.

The other day, my light temp went too high in flight (I have telemetry), the water had frozen in the pipe. I added car anti-freeze, problem solved :)

I'm building a second plane which should break 20k lumens, again liquid cooling will be used as the light is hidden in the nose without airflow. So, basicaly it'll be a foam flashlight with wings.
 
Looks like people missed the entire point of the OP. He was not asking about water cooling from behind but actually running a coolant over the LED die itself. This has merit for certain applications but I can't see it making it in the market place just yet. I am however working on just such a thing. It uses liquid silicone as the coolant and a micro pump designed for cooling lasers that fits into a flashlight. I will start a thread on it when I can get around to it.
 
Top