Slashdot Mirror
The Thermal Paste Revolution
arhines writes "ZZZ is running an article about an interesting new thermal paste which surpasses even solder in thermal conductance by 33 percent. If this paste makes it to the market sometime soon, we'll all surely be thinking about putting it in our boxes. In fact, if use of the paste becomes commonplace, it may even give the semiconductor industry a little speed boost."
" Lots of OEM or low end cooling setups use either a thermal interface pad (TIM) or that white goop you get at radio shack. The fact is that neither of those does a great job of transferring heat from the processor to the heatsink. While they work ok, they don't exactly assist Moore's law in fulfilling itself by limiting clock speeds with heat." Actually, that's hardly true at all. RS's compound has been found to be one of the best out there. Just take a look at some reviews that include it.
I got a +5, Troll
Working in a test lab for PCs I encountered one problem with heat conducting paste: Since Intel introduced the mPGA 478 housing for their CPUs the ZIF socket on the mainboard is much smaller than the heatsink above. That means, if the paste between processor and heatsink is too much adhesive it is like the processor is glued to the heatsink and every time you remove the heatsink (e.g. for changing the CPU) you pull out the processor from a closed ZIF socket! Ok, so far the processors survived but I don't think that this is nice anyway.
I have a tube of Arctic Silver 2 (yeah, I'm like two generations behind) but I'm not sure that I really needed it. This dude tried out several non-conventional thermal transfer compounds, including vegemite.(!) When properly applied, there wasn't a huge difference between them. In fact, in the (extremely specific) conditions, the vegemite and toothpaste outperformed the Arctic Silver! (Obviously, you should read the article for details.)
The article's point isn't that you should be using toothpaste; rather, it's that make sure you properly apply whatever thermal compound you do use, and don't expect miracles. No matter how effective your thermal transfer, you've still got to dump the heat somewhere. If you're running close to the edge of thermal failure, there are almost certainly other, much more effective cooling solutions. This new paste is probably a good thing, but don't expect miracles.
About 10 years ago I was working on a product that used 200 Amp IGBT's for a traction drive. I spent about a month researching the thermal circuits used to cool these devices.
The conclusion: The best thermal contact is metal to metal. The best way of acheiving this is by "lapping" the contact area's together with a fine abrasive. Once your have done this the application of a minute amount of thermal grease improves conductivity by less than 0.5%. We also discovered that applying more than a fine film or grease significantly decreased the conductivity (10% or more).
Lay off the grease!
Oh dear lord. You haven't heard a "Windtunnel" G4 at full howl, have you? ;)
I have a QuickSilver G4 on my desk, and let me assure you, it is far from fanless. The Windtunnels (AKA "Quad Nostril") took it a step further.
I agree that some current CPU designs are downright absurd (Intel so completely gave up on the concept of being able to cool their highly-clocked - over-clocked if you ask me - chips that they designed in a feature to slow down/turn off sections of the chip in order to lower temperatures), but passive cooling only gets you so much. I really doubt VAXstations could operate efficiently with a pair of 15,000rpm drives inside the case, which my QS handles with reasoanble aplomb (though I splurged on additional cooling - mostly a PCP&C external supply exhaust fan - to keep temps at reasonable levels).
The last fanless Apple desktop computer I can think of were the 2nd-generation iMacs (the ones with completely clear sides, or flower-power, and whatever god-awful color scheme they came up with at the end). Everything else from their recent lineups, even portables, have had fans.
Supposedly the G5s will feature quieter operation, by virtue of seperated thermal zones with different thermally-controlled fans exhausting air from each zone - they only spin as fast as that zone requires. To some extent, it is probably the wave of the future in this regard, in that increased thermal needs have butted up against all the buffont bettys tranquility requirements.
But if another artist whines about how loud their 10,000+rpm drive is, and how their tuned-for-quiet-not-performance-operation IDE drive at home is quieter, I swear I'm gonna stuff a high-CFM 120mm fan in their pie-hole... and maybe a high-cfm 80mm fan in their corn chute.
Nah. That'd be cruel.