Puppy Linux Discussion Forum

I had seen some guy do this on another video where his original fan was broke on a freebie computer and new one was $30. He just wanted to get it going to see if worth more effort (and to make a silly youtube video) so left top off and mounted a mountain of a heat sink with no fan. Thermal paste plus zip ties didnt work. So he super glued it. And yea it worked. No clamps needed. Now as the above video points out, there is a downside if you want to ever remove the heatsink... And nobody has mentioned how it holds up long term. I mean even tooth paste lasts for couple days until the water evaporates. Oh yea, dont use epoxy, apparently the vapors of the processor cooking it are not beneficial. Oh and no gummy bears. Yea one guy tried using a gummy bear. Just made a gummy mess with no benefit.

@mouldy

Superglue in cured form is plastic which doesn't have very good heat transfer, plus it starts to de-polyrmerize when used above 140-160F for extended periods. Other properties also make it a poor choice for any permanent attachment between dissimilar materials. Most adhesives are polymers with poor thermal transfer properties unless they have a metal fill exceeding 70%.

The "high" performance thermal compounds are mostly marketing hype. Main thing is to keep the compound layer as thin as possible. Some legitimate thermal compound substitutes that will work are anti-seize compound or diaper creme (zinc oxide).

wizard

anti-seize compound

I've used a high temp, silicon anti-seize grease successfully. Fixed the drive belt tension pulley (squeaky and very loud) on our home laundry dryer and using the same stuff a thin layer for heat sink contact on a CPU. The heat sink is held by a clamp assembly. Still working...this grease is monster...used it on the job to lubricate the focus mechanism on film lamps that get really really hot.

The trick to applying any kind of thermal paste is doing it without trapping voids or air bubbles, which, being air, don't conduct heat well at all. A drop of paste in the center, smashed so it spreads to the edges, is the way I've always done it. To avoid trapping air bubbles, I don't spread it out before assembly.

The first thermal paste was silicone, but it was soon noticed that the silicone oil in the paste coated everything in the vicinity. Not many solvents effectively remove silicone oil, so it causes an unsightly mess that's difficult to clean off completely. Silicone oil on electrical contacts apparently breaks down into silicon dioxide (possibly from microscopic arcing,) preventing the contacts from making contact. This happened to me several times before I realized why it was happening.

wizard wrote: ↑Sun Jun 30, 2024 5:21 pm

@mouldy

Superglue in cured form is plastic which doesn't have very good heat transfer, plus it starts to de-polyrmerize when used above 140-160F for extended periods. Other properties also make it a poor choice for any permanent attachment between dissimilar materials. Most adhesives are polymers with poor thermal transfer properties unless they have a metal fill exceeding 70%.

The "high" performance thermal compounds are mostly marketing hype. Main thing is to keep the compound layer as thin as possible. Some legitimate thermal compound substitutes that will work are anti-seize compound or diaper creme (zinc oxide).

wizard

I have never been a big super glue fan for any use, seemed like a cure all type product that yea probably work for some things but not as well as implied. I did at one point use it to repair some cracks in a RARE EXPENSIVE plastic air intake tube on a car. Works much better on plastic (well some kinds of plastic) if first mixed with baking soda. Be careful, baking soda cause it to cure incredibly fast, but also much stronger. Air is metered on modern cars with an electronic gizmo and the leaks throw off the readings which in turn throws off air-fuel mixture. Course used to be same with leaky carburetor base gasket. Sometimes there was just no alternative to using 5min epoxy to seal them. though best to still use a gasket so if you did have to remove it, the gasket could tear apart where the epoxy wouldnt. And there are some industrial epoxies not available to general public that are far stronger and with much greater resistance to heat, etc.

My interest in the super glue was first that it worked at all for a processor, even in the short term, but also that video I didnt link that guy used cause he didnt have a good way to clamp a non-OEM heat sink to the processor. The super glue both acted as a thermal compound plus held the heat sink tight. Seriously engineering a mechanical hold down for a non-OEM heatsink can be quite a challenge. Now how exactly you get the heatsink off after using this for experimentation is not clear....

I have used silicone grease cause usually have it around for automotive wiring connections and to mount ignition module on cars to heat sink. Well on GM cars, not even sure current ones have such things. The others usually had the heatsink built into the module. But GM had them mounted to the distributor casting. Though actually they lasted much longer if mounted to chunk aluminum say under the dash away from engine heat. When Ford and Chrysler stopped making modules for the old cars, the ones from China had minimal heat sinks to save on cost of aluminum I guess, and thus didnt last. Found out on non-computer cars anyway you could use the GM module mounted to chunk aluminum or old computer heat sink. Once they started tying everything into computer or series of computers it became lot more complex to change anything in non-factory way. Modern vehicles are more computer than car. And not friendly user programmable computers. More like trying to custom reflash a bios to change them.