I watched the vids TELVM linked - interesting. Rotating the cpu 90-degrees [in the same direction] every 10 passes seems important - random action grinding will impart a parabolic surface as in grinding a telescope mirror.
wrote:I watched the vids TELVM linked - interesting. Rotating the cpu 90-degrees [in the same direction] every 10 passes seems important - random action grinding will impart a parabolic surface as in grinding a telescope mirror.
There are a few ways to do it, that is just one, Rotating the cpu 45 degrees wile far more tedious does a much better job.
I do 100 passes then rotate 90 degrees.
It's just my personal taste
Almost all kind of compound last 2-3 years, then they do transfer little heat (dry up).
When I build a machine i want it to last for a least a decade without disarm it (exept blowing dust) so what i do is a homemade old school heatsink lapping to mirrow level and a very smooth laping to ICs (lightly) and no compound at all, and i test they got pretty warm, so they're doind the job.
I use Icefusion on all my pre P4 machines and NT-H1 on the newer stuff. I also use the Icefusion on the GPU heatsinks, last week I pulled the heatsink off a MX 400 and the paste was dry caked dust.
wrote:... what i do is a homemade old school heatsink lapping to mirrow level and a very smooth laping to ICs (lightly) and no compound at all, and i test they got pretty warm, so they're doind the job.
For peace of mind I always drop a very small drop of thermal paste after lapping.
But superb metal to metal surfacing can achieve surprising results 😲 : https://www.youtube.com/watch?v=fnoVV-RWIWY&t=11m36s
Let the air flow!
wrote:I use Icefusion on all my pre P4 machines and NT-H1 on the newer stuff. I also use the Icefusion on the GPU heatsinks, last week I pulled the heatsink off a MX 400 and the paste was dry caked dust.
I would love to try one of those Shin-Etsu Japanese thermal pastes. It might be the best thermal paste money can buy. Although it's probably not worth using on retro stuff.
Has anyone ever solder a heatsink to a cpu?
wrote:Has anyone ever solder a heatsink to a cpu?
How about liquid metal TIM? Do you even need to worry about surface flaws then?
wrote:wrote:Has anyone ever solder a heatsink to a cpu?
How about liquid metal TIM? Do you even need to worry about surface flaws then?
I tried that stuff before. It's nice, but you have to be careful with it.
Does anyone know how the cpu internals are connected to the cpu heat sink itself? is the chip 'fused' into the metal sink during manufacture or does it employ some kind of mechanical connection perhaps not unlike how the cpu is secured to the HSF, only permanently?
It depends on the CPU. Some are soldered to the chip itself, and some are glued there with some compound to help transfer the heat from the chip.
Let the air flow!
The reason I asked was I wonder if people have also lapped the chip to the spreader to improve the *overall* thermal efficiency...
I've used the white, generic stuff for a long time. Switched to Thermaltake TG-1 after I saw the breakage of oils and solids (hard crust on the middle, gel-like substance on the borders of the heatsink) while cleaning a CPU. Nowadays I use MX-2 and MX-4. Nice temps and not hard to apply correctly. Also not electric conductive, so it's safe to be used even on exposed dies.
Yep if we remove the IHS to clean the old paste, apply fresh one and then put the IHS back, it makes sense to clean and lightly sand the underside of the heat spreader while we're at it.
No need to lap the die, they come already well flat and polished and any lapping would probably destroy it.
Let the air flow!
if you take the IHS off, might as well leave it off, unless your going to scrape off all the old epoxy so the IHS sits 100% against the die again for sure. Its pefectly safe to not use the IHS, we didn't have them on P3's, Athlon's or Athlon XP's remember.
That was fine with the older sockets as the had some wiggle room that could help preventing cracking the core from to much pressure. Mainly because they had clip on posts that were much closer together as the CPU package was smaller. Cracking the core still happened but only to overzealous handling of the cooler or from people who had never installed one before. Nowadays we use the motherboard to get extra leverage over a much larger area than the old socket system and its designed with the IHS in mind. If its removed, then shimms will be needed or the chance to crack the core will be sky high.
wrote:... we didn't have them on P3's, Athlon's or Athlon XP's remember.
Yes, I forgot about that.
wrote:That was fine with the older sockets as the had some wiggle room that could help preventing cracking the core from to much pressure. Mainly because they had clip on posts that were much closer together as the CPU package was smaller. Cracking the core still happened but only to overzealous handling of the cooler or from people who had never installed one before. Nowadays we use the motherboard to get extra leverage over a much larger area than the old socket system and its designed with the IHS in mind. If its removed, then shimms will be needed or the chance to crack the core will be sky high.
last chip I delidded was an i5 2500k and then i put my hyper212+ back on it without an issue, no cracked core at all.
I didn't say/imply that it would crack it every time. I pointed out that times have changed and the risk is far greater than it used to be because the current mounting methods require far greater force. Going off the top of my head, the amount of force used for old athlons and P3's was somewhere between 15-20 ft lbs. Now days its double that. Adding shimms help aid block/heatsink installation by preventing some of that force from going in the wrong direction if you manage to slip a few degrees when pushing down. It not at all hard to put a wee bit of pressure on one corner when pushing down and crack a core of a de-lided CPU. Just because you managed to do once doesn't mean you can do it again and again without any risk. Lots and lots of people started breaking their later gen athlons when the coolers and sockets got bigger. Many others managed to install the cooler but missed good contact with the core and fried the CPU. That's why AMD started putting IHS's back on their trayed CPU's.
