Arctic Silver - Shelf Life?

[ik9000]

I get the strong impression from the various vids that if you get the paste wrong it's detrimental to heat conduction and therefore cooling. Also, if there are surface imperfections it's supposed to fill the spaces and exclude the notoriously insulating air pockets.

However, there never seems to be any right way to apply it - a blob in the middle, one or a few parallel lines, an even coating like you're buttering toast, a little or a lot. Have none showing round the edges or have it squeezed out all the way around. What's right, if anything?

you're buying a ryzen 3xxx right? Cribbed from a post I gave elsewhere:

ok see under the lid here and note the corner with the arrow:



you want to get a line across those two core dies and the larger IO chiplet. This is the pattern I used based on a guide I found online but now can't find.
So it looks like a big line down through the N of Ryzen and a small line focussed under the Y and Z with a blob either side (references to the LID markings, not the image above).





For the avoidance of doubt you only need to put it on the lid - you do not need to de-lid the CPU. I'm just trying to show how the TIM application relates to the stuff below the lid.

What then happens is the stuff spreads out when you put the cooler on, but it helps avoid no dead-spots directly above the bits that matter. Since Ryzen CPUs have soldered lids this helps give you the best chance of decent temps. Do not overdo the paste. Less is more, provided you don't skimp too far.

Generally I would dress the cooler by smearing a blob of paste over the cooler mating surface using my finger in a latex glove. Then I'd wipe off as much as possible using an old credit card at 30 degrees or so. This helps by leaving paste residue in any microvoids. Then apply the paste to the CPU lid as shown above. Mount the cooler and let it do its thing.

[ik9000]

re watercooled comment about hitting the heatpipes, that vertical line crossed mine nicely. I might have made it a tad longer than drawn above, but it is performing well. I disagree about letting it spill over the edge: I'd rather not have that. I don't want thermal paste anywhere near the CPU socket.

[watercooled]

Spilling over the edges was only about bare dies, not processors with a heatspreader.

[neojay]

As it was a salvaged cooler it had been given a good clean before install. Its probably more to do that the old volcanic beast coming to the end of its life.

[watercooled]

That shouldn't be the case though. If it's not the original cooler, I suspect it might not be mounted properly.

[Xlucine]

If the TIM is covering the whole heatspreader after affixing the cooler, then you're good - it's as simple as that. This could be from one blob in the middle, or a fancy pattern, or manual spreading - either way, the result is the same. Thermal cycling causes any excess of paste to migrate out in a short time during use, and any bubbles will similarly make their way to the exit

[watercooled]

The 290 doesn't have a heatspreader, so would be particularly sensitive to application: https://www.techpowerup.com/review/amd-r9-290/3.html

[ik9000]

The 290 doesn't have a heatspreader, so would be particularly sensitive to application: https://www.techpowerup.com/review/amd-r9-290/3.html

Hence my question! With the heatsink factory applied to the chip what does he hope to gain by removing it and using standard TIM?

[watercooled]

More detail would be useful for diagnosing, I agree. Factory TIM can degrade over time, but it's hard to know whether that's the case here without more details.

[DemonAndrew]

I've used several year old thermal paste several times and never seen any noticeable difference in temps compared to fresh thermal paste. For reference, I use the "pea" method.

[Froger007]

Depends mostly if it is a gel or paste. Gel based compound is normally good for 10 years or even longer. Paste in storage would probably still be ok in your case but if installed on CPU would be hard and end of life.