4 ms·

Sony develops thermal sheet as good as paste for CPU cooling

28 points by ukdm 14y ago

vecinu 14y ago
This is great news and hopefully, they'll bring it to the consumer market too. Thermal paste is definitely messy and hard to get right.
vonmoltke 14y ago
Both articles are rather thin on details about this material. I'm not sure how much (or if) its better than products that have been in the mil/aero industry[1] for years. Hopefully it will be easier and cheaper to get ahold of than the alternatives. [1] Such as http://www.bergquistcompany.com/thermal_materials/sil_pad/sil-pad-2000_properties.htm http://www.bergquistcompany.com/thermal_materials/sil_pad/si...
robomartin 14y ago
According to this article the claimed thermal resistance is in the order of 0.4-0.2K cm2/W: http://news.softpedia.com/news/SONY-Demos-Best-Thermal-Pad-in-the-World-281342.shtml http://news.softpedia.com/news/SONY-Demos-Best-Thermal-Pad-i... This is nothing new or earth-shattering as far as I am concerned. I've been using various thermal interface materials over the years for interfacing elements needing to be cooled to heat-sinks. These components range from MOSFETs to LEDs and FPGAs. You can easily achieve thermal resistance down to 0.11 K cm2/W with form-in-place sheets from companies such as Fujipoly: http://www.fujipoly.com/products/sarcon-thermal-management-components.html http://www.fujipoly.com/products/sarcon-thermal-management-c... In some cases the tradeoff is between high thermal conductivity and electrical conductivity. Where application requirements dictate that the thermal interface material (TIM) not conduct electricity (high-power LEDs typically have the thermal slug connected to the anode) you have to select TIMs that might not contain aluminum or other conductive compounds. In these cases you might take a thermal conductivity hit. That said, there are manufacturing techniques for such assembles that enable the use of electrically-conductive high-thermal-conductivity TIMs while greatly enhancing the thermal characteristics of the entire assembly. One that comes to mind is the use of aluminum-clad printed circuit boards as an intermediate step. EDIT: Used "thermal conductivity" when the numbers referred to "thermal resistance"
- vonmoltke 14y ago
  Those numbers are thermal resistance/impedance, not conductivity.
  - robomartin 14y ago
    You are right. I tend to think in terms of thermal conductivity, which is not the same as thermal resistance. Editing post to use proper terminology.
zxcdw 14y ago
I think the article tells nothing of importance as it is right now. Since when has applying thermal paste been messy and requiring trial and error to apply "correctly"? These days the thermal load(W/cm²) between the CPU heat spreader(the nickel alloy(?) coated copper plate on top of the CPU package) and the heatsink base is so low compared to the thermal load between the CPU core and the CPU heat spreader, that even if you manage to use ten times more efficient interface material in between, the actual heat transfer from the CPU core to the heatsink base wouldn't increase all that much. It's easy to understand, when you consider that the heat spreader has so much more surface area than the actual CPU die. In simplified layman terms, no matter what you put between the CPU and the heatsink base, you get the same temperature give or take a few degrees celsius. Of course, the thermal compound starts to matter much more when you overclock and the thermal load may double or even triple. I guess new inventions like described in the article are welcome though, but personally I am no excited at all, regardless of the claimed thermal conductivity and "real world performance".
- mnicole 14y ago
  > Since when has applying thermal paste been messy and requiring trial and error to apply "correctly"? For people not accustomed to building their own rigs or using thermal paste, it can have a learning curve. Even having done it before, I had an issue two months ago where applying a little too much kept me from entering the BIOS. After some worry I reseated it and it worked fine, but anything that makes the process of building your own system easier should be a welcome change.
  - zxcdw 14y ago
    How is that even possible? How could applying too much material between the CPU heatspreader and the heatsink keep one from entering BIOS? Even if you spilt it all over your motherboard and socket, it shouldn't do a damn, as (at least most?) thermal pastes aren't conductive enough to cause any short-circuits or other harm.
    - mnicole 14y ago
      I couldn't tell you, to be honest. It was originally an issue with the video card overheating, and I decided to clean and re-organize everything. The thermal was the only thing I wasn't 100% on, and when I reseated it with less paste, it worked again.
astrodust 14y ago
Is there a reason the thermal interface of a CPU is a smooth surface and not, say, a series of thin, deep grooves into which a fan can be slipped in for a snug fit with a minimal amount of paste? You could easily increase the contact area ten fold, possibly even more with some clever engineering. The only down-side would be requiring a fairly precise fit with the heat-sink where now any old hunk of metal will do.

28 points by ukdm 14y ago

This is great news and hopefully, they'll bring it to the consumer market too. Thermal paste is definitely messy and hard to get right.

vonmoltke 14y ago

Both articles are rather thin on details about this material. I'm not sure how much (or if) its better than products that have been in the mil/aero industry[1] for years. Hopefully it will be easier and cheaper to get ahold of than the alternatives. [1] Such as http://www.bergquistcompany.com/thermal_materials/sil_pad/sil-pad-2000_properties.htm http://www.bergquistcompany.com/thermal_materials/sil_pad/si...

robomartin 14y ago

According to this article the claimed thermal resistance is in the order of 0.4-0.2K cm2/W: http://news.softpedia.com/news/SONY-Demos-Best-Thermal-Pad-in-the-World-281342.shtml http://news.softpedia.com/news/SONY-Demos-Best-Thermal-Pad-i... This is nothing new or earth-shattering as far as I am concerned. I've been using various thermal interface materials over the years for interfacing elements needing to be cooled to heat-sinks. These components range from MOSFETs to LEDs and FPGAs. You can easily achieve thermal resistance down to 0.11 K cm2/W with form-in-place sheets from companies such as Fujipoly: http://www.fujipoly.com/products/sarcon-thermal-management-components.html http://www.fujipoly.com/products/sarcon-thermal-management-c... In some cases the tradeoff is between high thermal conductivity and electrical conductivity. Where application requirements dictate that the thermal interface material (TIM) not conduct electricity (high-power LEDs typically have the thermal slug connected to the anode) you have to select TIMs that might not contain aluminum or other conductive compounds. In these cases you might take a thermal conductivity hit. That said, there are manufacturing techniques for such assembles that enable the use of electrically-conductive high-thermal-conductivity TIMs while greatly enhancing the thermal characteristics of the entire assembly. One that comes to mind is the use of aluminum-clad printed circuit boards as an intermediate step. EDIT: Used "thermal conductivity" when the numbers referred to "thermal resistance"

vonmoltke 14y ago
Those numbers are thermal resistance/impedance, not conductivity.
- robomartin 14y ago
  You are right. I tend to think in terms of thermal conductivity, which is not the same as thermal resistance. Editing post to use proper terminology.

zxcdw 14y ago

I think the article tells nothing of importance as it is right now. Since when has applying thermal paste been messy and requiring trial and error to apply "correctly"? These days the thermal load(W/cm²) between the CPU heat spreader(the nickel alloy(?) coated copper plate on top of the CPU package) and the heatsink base is so low compared to the thermal load between the CPU core and the CPU heat spreader, that even if you manage to use ten times more efficient interface material in between, the actual heat transfer from the CPU core to the heatsink base wouldn't increase all that much. It's easy to understand, when you consider that the heat spreader has so much more surface area than the actual CPU die. In simplified layman terms, no matter what you put between the CPU and the heatsink base, you get the same temperature give or take a few degrees celsius. Of course, the thermal compound starts to matter much more when you overclock and the thermal load may double or even triple. I guess new inventions like described in the article are welcome though, but personally I am no excited at all, regardless of the claimed thermal conductivity and "real world performance".

mnicole 14y ago
> Since when has applying thermal paste been messy and requiring trial and error to apply "correctly"? For people not accustomed to building their own rigs or using thermal paste, it can have a learning curve. Even having done it before, I had an issue two months ago where applying a little too much kept me from entering the BIOS. After some worry I reseated it and it worked fine, but anything that makes the process of building your own system easier should be a welcome change.
- zxcdw 14y ago
  How is that even possible? How could applying too much material between the CPU heatspreader and the heatsink keep one from entering BIOS? Even if you spilt it all over your motherboard and socket, it shouldn't do a damn, as (at least most?) thermal pastes aren't conductive enough to cause any short-circuits or other harm.
  - mnicole 14y ago
    I couldn't tell you, to be honest. It was originally an issue with the video card overheating, and I decided to clean and re-organize everything. The thermal was the only thing I wasn't 100% on, and when I reseated it with less paste, it worked again.

astrodust 14y ago

Is there a reason the thermal interface of a CPU is a smooth surface and not, say, a series of thin, deep grooves into which a fan can be slipped in for a snug fit with a minimal amount of paste? You could easily increase the contact area ten fold, possibly even more with some clever engineering. The only down-side would be requiring a fairly precise fit with the heat-sink where now any old hunk of metal will do.