h a l f b a k e r yExpensive, difficult, slightly dangerous, not particularly effective... I'm on a roll.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
This heat sink is like a metal donut that sits on top of the chip. The outer walls of the donut are made of copper or brass or whatever it is they use in regular heatsinks. However, the innards of the donut are a metal with a low melting point, like lead. ( Maybe a lead/tin alloy for greater health
safety. ) As the donut heats up, the inner metal begins to liquefy.
A cooling fan blows air through the donut, so theoretically, the donut should be cooler on top. The metal on top cools, and flow down through the donut. When it gets to the bottom, it picks up heat and rises again.
I'm not saying that the inner metal will flow circularly through the donut, but I would guess there would be a general flow of metal up and down on both sides of the donut.
NaK
http://en.wikipedia.org/wiki/NaK
Eutectic alloys of alkali metals [8th of 7, May 17 2008]
Menger Sponge
http://en.wikipedia.../wiki/Menger_sponge
Perhaps an ideal shape for the heatsink [lawpoop, May 17 2008]
Table of elements sorted by thermal conductivity
http://environmenta...riodic/thermal.html
For reference [csea, May 18 2008]
NanoCooler's liquid metal cooling loop
http://www.techpowe...mg/05-05-05/p10.jpg
A prototype. [phoenix, May 18 2008]
Heat Pipe
http://en.wikipedia.org/wiki/Heat_pipe
[MisterQED, May 18 2008]
Indium-Gallium
http://www.indium.c...ons/liquidmetal.php
[hippo, May 19 2008]
Indium-Gallium (thermal conductivity)
http://www.indium.c...on/thermalklist.php
The top indium alloy has better thermal conductivity than Aluminium, but not quite as good as copper (see link below) [hippo, May 19 2008]
Thermal conductivity of common materials
http://www.engineer...uctivity-d_429.html
[hippo, May 19 2008]
[link]
|
| |
Hmm. You'd want something with a lower
melting point than lead, I think. However,
there are plenty of alloys that melt at
anywhere from room temperature
upwards. |
|
| |
On the other hand, maybe you'd be better
off using liquid/vapour cycling, a la
heatpipe? Those things are awesomly
effective at heat transfer. |
|
| |
the heat has to go somewhere... even if you take it off the chip with a liquid metal, you're gonna have to cool the liquid metal before you use it again. Sapphire (ATI OEM) were going to do just that, though; a graphics card cooled by a liquid metal, pumped by a linear motor a few years ago. They never got it into production :( |
|
| |
Fast-breeder fission reactors use NaK, a eutectic alloy of Sodium and Potassium as their coolant. Apart from its slight tendency to catch fire spontaneously on contact with air, it's an excellent heat transfer medium (78% potassium and 22% sodium, is liquid from -12.6 to 785 °C). |
|
| |
Bakeable, but maybe a little too much like building an incendiary weapons technology into a commercial appliance (not that that's a problem at all). |
|
| |
Or you could use Mercury. Excellent heat transfer capablility, and in its pure metallic form, nowhere near as toxic as the lilly-livered environmentalist scaremongers would have you believe. |
|
| |
[Toaster]I was thinking that there could be some cooling mechanism, such as a fan, that cools off the top of the donut, drawing heat from the molten metal inside. |
|
| |
Not sure why a donut (torus) shape is specified, as its point of contact with a flat surface would be a circle. One would generally prefer a larger surface contact. |
|
| |
Gallium / Indium / Tin / Zinc alloys have fairly low melting points and are used for this purpose. |
|
| |
[8th] i think the problem with Hg is with disposal, not usage... I've got a number of old mercury switches around here that aren't poisoning anybody... but would if I tossed them out and the stuff got into ye olde ecosystem. Given that people ("they") buy new computers every couple years, with new incompatible components... |
|
| |
[csea] Probably a torus is not ideal, but I thought there should be somw way to get air flow *through* the structure. I guess ideally this heatsink would be something like a Menger sponge, but that would be a b*tch to manufacture. |
|
| |
Some tradeoff between airflow versus surface area. |
|
| |
[lawpoop] I wasn't familiar with Menger sponges, so looked it up. Pretty interesting, has lots of surface area, but probably not optimal airflow. |
|
| |
Most practical heatsinks for electronics use aluminum, as its thermal conductivity is surpassed only by silver, gold and copper, some exotic alloys, ceramics and carbon nanofibers. |
|
| |
You're right about the tradeoff. Sintering of sperical particles can give very high surface area/volume ratios, and decent airflow if forced air is used. |
|
| |
Most of the liquid metal heat transfer methods are used where the heat source must be at some distance from the radiating elements. |
|
| |
I have to look it up, but two things come to mind, the Alaska Pipeline and my ice cream scooper. Both use a phase change material to transfer heat much more effectively than a solid metal. The ice cream scooper uses the warmth from your hand to keep the ice cream from sticking to the scoop and the Alaska Pipeline uses posts filled with a salt to conduct away the heat that bleeds out of the hot oil and keep it from melting the perma frost and causing the pipeline to sink. |
|
| |
The Alaska Pipeline has the right idea, what you are looking for is a heat pipe (link). |
|
| |
Couldn't that be used as a MHD generator? Maybe power its own fan... |
|
| |
Conversely, a magnetic field could be used to pump the liquid metal without moving parts? |
|
| |
[TIB] I think that is how it is pumped. |
|
| |
The possibilities, i suppose, would be gallium, mercury, various alkali metals (quite hazardous, i would think), Fields metal or Woods metal. I don't know how conductive they are. |
|
| |
// a magnetic field could be used to pump the liquid metal without moving parts // |
|
| |
A Helical linear induction pump, or an an Einstein-Szilard pump (q.v.) |
|
| |
// I don't know how conductive they are // |
|
| |
Conductivity isn't the issue, it's their heat transfer capability. |
|
| |
Well, thermal conductivity - it's the same thing. My father-in-law used to cool silicon mirrors he was reflecting very powerful lasers off by pumping liquid indium-gallium alloys through them. These alloys are liquid at room temperature and have really good thermal conductivity characteristics (see link). |
|
| |
// indium-gallium alloys // |
|
| |
If the thing's going mass-market then uing rare earths will probably be too expensive. Sodium and Potassium are plentiful, and the technology for extraction and refination from the ores is well developed. |
|
| |
or you could just use water, in a closed system, at about 0.2 bar. |
|
| |
Hey, why does it have to be liquid? Rotate a metal slatted belt with a motor. Case modders would love to have the top of it protruding from the case, whirling round like a minigun. |
|
| |
Or use lots of little metal balls; slowly driven up out of the case with an Archimedes screw and dropping into the cooling hopper, before rolling along a path back into the internal hopper. |
|
| |
Miniature, solid copper wheeled vehicles, powered by internal tiny stirling engines, which power themselves up a spiral ramp off the chip; when they run out of power they run backwards under their own weight to the bottom again. Their front ends are wedge shaped so the ascending ones can pass beneath the descending ones. |
|
| |
[pocmloc] That could work only by adding a balance mechanism that would keep the processor always at the bottom, in case you have the case standing on one side. |
|
| |
Yes, you either need gimbals, or alternatively, bi-metallic levers which are heated by the over-heating chip and which press outwards to flip the case upright. |
|
| |
Wouldn't the original doughnut idea need to be upright as well anyway? |
|
| |
Why not make the CHIP into a hollow cylinder and
pump metal through it? |
|
| |
Google "Woods metal" which has a very low melting point. You might be able to roll your own. There's some stuff you can buy; brand names Cerrotru and Cerrobend. |
|
| |