By: Edward (IP: 24.5.93.*)
Written on: 29-09-2006 01:37
I think the issue is, primarily, that the CPU and Heat Sink surfaces do not contact properly, leaving air pockets. Air does not conduct heat well.
It seems that while aluminum foil may conduct well -depending on the the grade of the aluminum being processed into foil- it has a uniform thickness, whereas the thermal compound will "spread" filling air gaps where necessary and thinning down in "high" areas.
Also, the process of pressing aluminum onto the dies may itself create air pockets between the aluminum sheet and the substrates; also, being of uniform thickness, it may cause the "high" areas to be raised higher and/or "hard pressed" thereby extending the "low" areas into wider gaps and overstressing the CPU.
I think that if the heatsink and processor were engineered well enough that they matched up closely enough that the aluminum was able to be used unifomly across the joint, without any of the aformentioned liabilities, the two would have been matched well enough that you might not gain much benefit from the addition of the aluminum.
I might be possible to use a graphic mapping and manufacturing technology to match a particular CPU and Heatsink combination and manufacture an aluminum sheet custom mapped to accomodate that combination.
Edward |
By: Ray (IP: 84.67.182.*)
Written on: 29-09-2006 02:28
Another thing ... aluminium oxidises as soon as it is in contact with air, which is pretty much instantaneous, even if cleaned (which is why it is hard to solder aluminium).
So, although the idea is excellent, I wonder how good aluminium oxide is at thermal conductivity?
Wouldn't copper foil be better? (I know copper oxidises but it's relatively a much slower process so the foil could be cleaned before inserting between heatsink & CPU with less chance of oxidisation setting in).
Interesting stuff!
Ray |
By: Purfus (IP: 207.5.141.*)
Written on: 04-11-2006 16:51
It is a novel idea if your in a bind and have a lack of a usable compound. However the biggest problem with using aluminum is it is a very cheap light weight metal. It has very little heat capacity (thats why you can pull the aluminum off a turkey roast right out of the oven and not burn yourself). The only saving grace is it is also quite good at dissipating heat. But this is just not enough to make up for it's low capacity. I believe you would find that aluminum will do the job at lower temperatures but temps over 110f would start to exceed its capacity and the heat would be unable to transfer to the heatsink fast enough and it would start to build up in the core and temperatures would begin to rise even faster. When dealing in heat transfer you want high grade heavy metals.
On a similar note I did once use a finely ground zinc powder to refinish a crappy aluminum heatsink surface and it worked quite well. Zinc is a far cry from silver but it is better than the crappy aluminum the cheap heatsink was made of so I would be willing to bet the heat transfer capabilities of the zinc powder was not far from that of arctic silver. Only when used with a cheap aluminum heatsink of course. A nice copper core sink would probably be hindered by using zinc. Note I only used the zinc powder to imrpove the heatsink surface (ie fill in pours and prevent air bubbles). I also used a greese to facilitate the heat transfer so don't come barking down my door when your chip fries because you crushed up a zinc losenge and crammed it in between your chip and sink. Please don't do that, I did say a fine grain zinc powder... |
By: HardwareMan (IP: 72.130.46.*)
Written on: 12-12-2006 20:20
Most people miss the point of the thermal conductive compound used between two flat surfaces.
Air is a poor thermal conductor. Polishing the flat metal surfaces to improve contact is an expensive process.
The thermal compound is meant to minimize the deleterious air gaps that exist at a microscopic size. It is meant to be applied in the tiniest possible quantities to displace the air gaps. Two giant metal (copper/aluminum) plates in perfect contact are better transfer interfaces than thick gobs of thermal compound between two roughly machined surfaces.
Putting aluminum foil between the CPU and heatsink/fan (HSF) may even have the opposite effect. There are now four surfaces with air gaps to contend with. Only testing will verify the result.
Putting arbitrary substances in lieu of thermal compound is dangerous. The suggested polishing subtitutes will introduce volitile fluids that are designed for physical transfer of metal debris. They are not for long term, high temperature thermal transfer.
If you cannot obtain some, borrow a small application of thermal compound from a friend.
Conclusion
Premium thermal conductive compound sells for about US$10 at Radio Shack (generic alternatives are a few dollars). Some of these products are good for half a dozen applications. This is an inexpensive maintenance cost for a microprocessor (cost US$75 to $400) in an expensive PC system (cost US$500 to $2000).
|
|