Solid-State PSU Cooling: Beyond the Pail!
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee, Devonavar
Solid-State PSU Cooling: Beyond the Pail!
After much investigation, I've decided to try solid-state cooling!
The idea is to get an empty 20-gallon plastic pail (think drywall mud pail), and after putting the (case removed) PSU inside, fill it with powdered material.
The material will have to be an electrical insulator, but good thermal conductor. After some research, I found that metal oxides work best.
It seems that Aluminum Oxide (Alumina) is relatively cheap and has a conductivity of 30 W*mK, which is very good for a non-metal. It can be purchased easily at lapidary shops.
The other idea is to remove the PSU case, and carefully coat it entirely with epoxy, then place it in the pail and fill it with metal powder (probably aluminum).
The idea is to get an empty 20-gallon plastic pail (think drywall mud pail), and after putting the (case removed) PSU inside, fill it with powdered material.
The material will have to be an electrical insulator, but good thermal conductor. After some research, I found that metal oxides work best.
It seems that Aluminum Oxide (Alumina) is relatively cheap and has a conductivity of 30 W*mK, which is very good for a non-metal. It can be purchased easily at lapidary shops.
The other idea is to remove the PSU case, and carefully coat it entirely with epoxy, then place it in the pail and fill it with metal powder (probably aluminum).
Last edited by Isochroma on Sat Apr 07, 2007 6:36 pm, edited 1 time in total.
wow
wow! interesting and very difficult endeavor
I definitely wish you sucess, but I would think that would will still need to cool the outside of the container somehow. the heat needs to dissapate from the powder to something else, if it's air, then will normal convection take care of that?
I definitely wish you sucess, but I would think that would will still need to cool the outside of the container somehow. the heat needs to dissapate from the powder to something else, if it's air, then will normal convection take care of that?
Yes, the plastic wall of the pail would be a poor conductor, but it would have a large surface area.
Brainstorm! I'll use a nice big stainless stock pot, acquired at the local Value Village for dirt cheap! My god, I even have one already, and it's just big enough! I'll even spray-paint it flat black on the outside for better radiative emission...
Though the idea may be beyond the pail, I don't think it will fail :)
Brainstorm! I'll use a nice big stainless stock pot, acquired at the local Value Village for dirt cheap! My god, I even have one already, and it's just big enough! I'll even spray-paint it flat black on the outside for better radiative emission...
Though the idea may be beyond the pail, I don't think it will fail :)
well...
well how much thermal energy are we trying to dissipate here?
if your computer uses 150W at idle, and your PS is 80% efficient, then we should have about 30W of heat to dissipate right?
the PS has 6 sides, so 30W / 6 = 5 watts per side doesn't seem like a whole lot as long as each side is exposed to air or other acceptable conducting material
I would first suggest packing the PS enclosre at tightly as possible with the powder and see if the heat is being transferred to the sides, they should feel warm .
I think this could actually work!!
if your computer uses 150W at idle, and your PS is 80% efficient, then we should have about 30W of heat to dissipate right?
the PS has 6 sides, so 30W / 6 = 5 watts per side doesn't seem like a whole lot as long as each side is exposed to air or other acceptable conducting material
I would first suggest packing the PS enclosre at tightly as possible with the powder and see if the heat is being transferred to the sides, they should feel warm .
I think this could actually work!!
Lol why didn't I think of that one b4.
Considering the size of the "cooler" surface area and the amount of heat a PSU can generate in real-life situations, and that the components can handle temperatures way above that of CPUs, this sounds practical and most importantly, easy (compared to the cooking oil concept, which is, er... oily also).
Heck, so easy I think it'd already work with a normal pail. 20-gallon is really quite an overkill, you'd want as little redundant sand as possible since more distance = higher temperature difference between mosfet and cooling surface.
Considering the size of the "cooler" surface area and the amount of heat a PSU can generate in real-life situations, and that the components can handle temperatures way above that of CPUs, this sounds practical and most importantly, easy (compared to the cooking oil concept, which is, er... oily also).
Heck, so easy I think it'd already work with a normal pail. 20-gallon is really quite an overkill, you'd want as little redundant sand as possible since more distance = higher temperature difference between mosfet and cooling surface.
Oh and now that you mention sand, I'll post a link which was very useful in evaluating the thermal conductivity of solid materials:
Thermal conductivity of some common materials as aluminum, asphalt, brass, copper, steel and many more: http://www.engineeringtoolbox.com/therm ... d_429.html
Sand: 0.35 (from other sources I found values as low as 0.27)
Aluminum Oxide: 30
Zinc Oxide: 17 (Thermal Conductivity of Ceramics and Other Insulating Materials: http://isochroma.com/Testfiles/Misc/SPC ... erials.pdf)
Thus, sand would be a poor choice; aluminum oxide is 85 times more thermoconductive! The metal oxides seem to be the best thermoconducting electroinsulators out there.
I haven't given up on pure aluminum powder (non-pyrophoric grade, remember!), and coating the parts in insulating resin.
Thermal conductivity of some common materials as aluminum, asphalt, brass, copper, steel and many more: http://www.engineeringtoolbox.com/therm ... d_429.html
Sand: 0.35 (from other sources I found values as low as 0.27)
Aluminum Oxide: 30
Zinc Oxide: 17 (Thermal Conductivity of Ceramics and Other Insulating Materials: http://isochroma.com/Testfiles/Misc/SPC ... erials.pdf)
Thus, sand would be a poor choice; aluminum oxide is 85 times more thermoconductive! The metal oxides seem to be the best thermoconducting electroinsulators out there.
I haven't given up on pure aluminum powder (non-pyrophoric grade, remember!), and coating the parts in insulating resin.
Ah, found a really excellent article on liquid-cooling a PSU by immersion!
How To H2Ocool Your Power Supply: http://overclockers.com/tips1240/index.asp
How To H2Ocool Your Power Supply: http://overclockers.com/tips1240/index.asp
Last edited by Isochroma on Sun Apr 08, 2007 1:04 pm, edited 1 time in total.
My system is far too hot, and there's only one slow 120mm in the case rear. Even now it overheats in the summer.
So why not remove the PSU's heat load by removing the PSU? The fact that its heat load must be removed via its noisy fan is the problem.
Rather than spend the $$$ on a fanless PSU, I realized that for less I could have fun making one of my two PSUs fanless in a novel way, and learn lots in the process!
So why not remove the PSU's heat load by removing the PSU? The fact that its heat load must be removed via its noisy fan is the problem.
Rather than spend the $$$ on a fanless PSU, I realized that for less I could have fun making one of my two PSUs fanless in a novel way, and learn lots in the process!
Some time back when there was a thread on the mobo immered in oil,it was noted that the fans ran fine in oil,helped it circulate. That a PSU has much higher voltages inside would disturb me. If you don't get a nasty fire....it ought to work great.Isochroma wrote:Ah, thanks for the info! I didn't realize how much of a difference solid vs. powder made. So it looks like Coconut oil and natural convection, at $6.50 / lb...
It better work great, 'cause the noise does surely grate! (my nerves, that is)
The reason why noise is so evil to my situation, is because each night I settle down in my plush gov't-surplus chair, and watch my 32" LCD from about 4' away. It connects via VGA to my PC, which is 5.5' away (under the desk).
Display technophiles have as one of their primary goals/beefs "black level", which is the darkest black a display can show. LCDs have poorer black level than CRTs, and many also underperform plasmas.
In the area of sound, the ambient noise is the auditory equivalent of "black level", excepting the fact that it usually doesn't originate from the sound system. No sound quieter than the ambient "black level" can be properly heard, and the ambient "black level" adds distortion and reduces SNR on all above-threshold sounds. Since sound is much less directional than light (from the human reception perspective), secondary sources are important.
No matter what the cost, I will shortly find the ideal solution, and report here in detail the implementation details, followed by longer-term reliability and performance statistics.
The reason why noise is so evil to my situation, is because each night I settle down in my plush gov't-surplus chair, and watch my 32" LCD from about 4' away. It connects via VGA to my PC, which is 5.5' away (under the desk).
Display technophiles have as one of their primary goals/beefs "black level", which is the darkest black a display can show. LCDs have poorer black level than CRTs, and many also underperform plasmas.
In the area of sound, the ambient noise is the auditory equivalent of "black level", excepting the fact that it usually doesn't originate from the sound system. No sound quieter than the ambient "black level" can be properly heard, and the ambient "black level" adds distortion and reduces SNR on all above-threshold sounds. Since sound is much less directional than light (from the human reception perspective), secondary sources are important.
No matter what the cost, I will shortly find the ideal solution, and report here in detail the implementation details, followed by longer-term reliability and performance statistics.
Mineral oil is a problem for a non-sealed system, because it evaporates. This lead to two things:
1. Mineral Oil loss
2. Mineral Oil does terrible things to the lungs when inhaled as vapor/mist. It does not absorb but accumulates and causes cancer, chemical pneumonia, etc. Here is a great Google search that finds all the nasties:
http://www.google.com/search?num=100&hl ... l%22+lungs
Remember, any system that uses mineral oil must be either vacuum sealed or vented to the outside.
1. Mineral Oil loss
2. Mineral Oil does terrible things to the lungs when inhaled as vapor/mist. It does not absorb but accumulates and causes cancer, chemical pneumonia, etc. Here is a great Google search that finds all the nasties:
http://www.google.com/search?num=100&hl ... l%22+lungs
Remember, any system that uses mineral oil must be either vacuum sealed or vented to the outside.
Stick an ohmmeter in the oil before you drop your PSU in, to make sure it's safe, and please post the results. Also, does the oil smell? If it does, you may have to seal your PSU inside, in an air (okay, oil-vapor) tight manner.
You may consider using a steel pail, or perhaps a pail with low electrical and high thermal conductivities. Hmm, an Alumina pail?
You could do like THG and throw the whole PC in the bath, fanless. I bet the large mass of oil would deaden any vibrations from coil whine, too.
Would it be safe to float a hard drive on pontoons, blowhole-side up, in the oil?
You may consider using a steel pail, or perhaps a pail with low electrical and high thermal conductivities. Hmm, an Alumina pail?
You could do like THG and throw the whole PC in the bath, fanless. I bet the large mass of oil would deaden any vibrations from coil whine, too.
Would it be safe to float a hard drive on pontoons, blowhole-side up, in the oil?
Seal off? I don't need to seal anything, 'cause the circuit board will be suspended by its cables, hanging in the middle of the pot. I'll screw a hook into the underside of my desktop, and hang from there... I'll cut a hole in the container lid for the cables to go thru.
And when I want to move it, I just shut off the power for a couple hours, and it all hardens up into a solid block, perfect for moving! Far superior to liquid oil, for sure.
And when I want to move it, I just shut off the power for a couple hours, and it all hardens up into a solid block, perfect for moving! Far superior to liquid oil, for sure.
Isochroma, I think Endo's right that oil will enter your wires, between the insulation and the copper, and creep upward against gravity due to adhesion. Despite the sorta-large vertical distance, you may find traces of oil on your mobo and elsewhere.
I looked up the electrical resistivity of several common materials. Your oil should be fine in that regard.
Another potentially suitable (but also flammable) substance would be paraffin wax. Actually, it would be rather amusing to fill your PSU with molten candle wax. As a bonus, it melts around 60°C. Thus, if you ran in to trouble with thermal overload or entrapped bubbles, the wax would melt near the troubled component and convection would kick in. Of course, convection is very weak in such viscous fluids.
Why not entomb a hard drive in candle wax? I think I'll run it through a simulation and maybe try it on a 2GB HDD.
I'll keep watching this thread. It will be interesting.
I looked up the electrical resistivity of several common materials. Your oil should be fine in that regard.
Another potentially suitable (but also flammable) substance would be paraffin wax. Actually, it would be rather amusing to fill your PSU with molten candle wax. As a bonus, it melts around 60°C. Thus, if you ran in to trouble with thermal overload or entrapped bubbles, the wax would melt near the troubled component and convection would kick in. Of course, convection is very weak in such viscous fluids.
Why not entomb a hard drive in candle wax? I think I'll run it through a simulation and maybe try it on a 2GB HDD.
I'll keep watching this thread. It will be interesting.
won't the oil solidify before getting anywhere near the mobo? melting point's 20~25CI think Endo's right that oil will enter your wires, between the insulation and the copper, and creep upward against gravity due to adhesion. Despite the sorta-large vertical distance, you may find traces of oil on your mobo and elsewhere.