Thermal Paste application for delidded CPU?

[mattspace]

Hi,

Just a pre-install question while I wait for the arrival of my 4,1's brain transplant - has anyone got tips / links to a tutorial for the correct application of thermal paste for a delidded cpu?

I've found numerous tutorials that suggest a single pea-sized blob in the centre is the best strategy, but they're all with CPUs that have the IHS on. Is there any difference of technique required?

Thanks.

 

[h9826790]

In general, no, but just require much less thermal paste (with IHS, the recommended amount is 0.2cc, so 50% less without the IHS). However, if you want to know the "official way", here it is.

(This is from the Apple Mac Pro 2009 model Technician Guide)

 

[SoyCapitanSoyCapitan]

Lidded or delidded, most of these debates are just geeks who have nothing better to do than have an Internet argument. The only thing that matters is correct quantity and having no air bubbles.

If you're interested there was a test between different application methods and there was no significant difference. Even if there was a 1 degree difference it would still be tens of degrees below your CPU's safety limits.

 

[Dubadai]

Just make sure you get something like arctic cooling mx-4, which is non-conductive. Otherwise you can damage the processor since its delided.

 

[mattspace]

Just make sure you get something like arctic cooling mx-4, which is non-conductive. Otherwise you can damage the processor since its delided.

They come as a (no doubt overpriced to some - I just want turnkey) kit with Artic Silver 5, which claims to be non-conductive, though it does state it's mildly capacitive. Not sure what to think about that.

 

[Dubadai]

They come as a (no doubt overpriced to some - I just want turnkey) kit with Artic Silver 5, which claims to be non-conductive, though it does state it's mildly capacitive. Not sure what to think about that.

Sorry, my bad. I meant capacitive!

Taken from their website:
Arctic Silver 5 was formulated to conduct heat, not electricity.
(While much safer than electrically conductive silver and copper greases, Arctic Silver 5 should be kept away from electrical traces, pins, and leads. While it is not electrically conductive, the compound is very slightly capacitive and could potentially cause problems if it bridges two close-proximity electrical paths.)

I wouldn't use it on a delided cpu, graphics card etc.

 

[mattspace]

Sorry, my bad. I meant capacitive!

Taken from their website:
Arctic Silver 5 was formulated to conduct heat, not electricity.
(While much safer than electrically conductive silver and copper greases, Arctic Silver 5 should be kept away from electrical traces, pins, and leads. While it is not electrically conductive, the compound is very slightly capacitive and could potentially cause problems if it bridges two close-proximity electrical paths.)

I wouldn't use it on a delided cpu, graphics card etc.

Quick search... God, thermal paste is a freaking rabbithole to get lost down...
[doublepost=1463324707][/doublepost]

In general, no, but just require much less thermal paste (with IHS, the recommended amount is 0.2cc, so 50% less without the IHS). However, if you want to know the "official way", here it is.

(This is from the Apple Mac Pro 2009 model Technician Guide)
View attachment 631333

That application pattern really looks like its at odds with recommendations to avoid air pockets - surely its just going to create 2 voids as the paste locks the air in them?

 

[h9826790]

Quick search... God, thermal paste is a freaking rabbithole to get lost down...
[doublepost=1463324707][/doublepost]
That application pattern really looks like its at odds with recommendations to avoid air pockets - surely its just going to create 2 voids as the paste locks the air in them?

May be depends on which paste you use. However, in my own experience, the biggest main factor is just the amount, as long as you get it right, the temperature will be fine.

I even put the heat sink in after thermal paste application, and then remove it to check if the paste can evenly distributed, and then add / remove paste as require. This should create a lot of air pocket theoretically. However, I can tell you that this is not a problem at all (in fact, work very well on my CPU and GPU). Air is bad if it practically insulate the heatsink and the CPU. I think, as long as the paste is eventually disturbed, the heat can find it's way to reach the heatsink. Heat is not light, no need to travel in straight path, don't need line of sight.

However, this is ONLY true if using AS5. I only test this on AS5. Other thermal paste can be very difference (viscosity may make huge difference in this matter).

I did try traditional method few times (one application, installation the heat sink without removal), and the "adjustable amount method" few times. Both gives me more or less the same temperature. However, with the adjustable method. I never have to worry about the the paste distribution (sometime it doesn't fully cover the CPU because I apply too less).

I know it's little bit contradict to what generally says on the internet. But it's my own experience, and again. ONLY for AS5.

 

[AidenShaw]

May be depends on which paste you use. However, in my own experience, the biggest main factor is just the amount, as long as you get it right, the temperature will be fine.

I even put the heat sink in after thermal paste application, and then remove it to check if the paste can evenly distributed, and then add / remove paste as require.

I use a very similar method.

• apply the paste
• set the sink down
• "wiggle" the sink to spread the paste (most heatsinks can rotate a tiny bit - rotate back and forth a few times)
• pull the sink and look for gaps (tiny gaps not important)
  • note that the paste can stick to either the sink, the chip, or both
  • if there's a gap on the CPU, see if there's paste on the sink in the same place
• set the sink down in the same orientation
• wiggle it a bit more
• tighten the holddowns

 

[mattspace]

so as a total newb to cpu upgrades - going with a non-capacative paste like the Arctic MX-4 over the Arctic Silver 5 effectively means that if I put too much on and it overflows onto the CPU die, I'm not going to damage anything.

And, with a non-lidded CPU that's more likely because the lack of IHS means you're putting grease on a smaller target, and there's more stuff on the die exposed that would normally be under the IHS?

 

[AidenShaw]

so as a total newb to cpu upgrades - going with a non-capacative paste like the Arctic MX-4 over the Arctic Silver 5 effectively means that if I put too much on and it overflows onto the CPU die, I'm not going to damage anything.

And, with a non-lidded CPU that's more likely because the lack of IHS means you're putting grease on a smaller target, and there's more stuff on the die exposed that would normally be under the IHS?

Whichever paste you use, don't put too much on so that it oozes onto anything but the mating surface.

In essence, you're asking "if I'm going to apply it badly, which paste is better".

 

[mattspace]

In essence, you're asking "if I'm going to apply it badly, which paste is better".

well more a case of "If I'm doing something I've never done before, based only on diagrams and youtube videos, which option removes the possibility of a catastrophic screwup."

As it stands, it seems Arctic MX-4 is basically unobtainable (listed as discontinued) in Australia, and for most of the international stores, $7 for the goo, and $20+ for shipping.

 

[h9826790]

No need to make it too complicated.

AS5 should be widely available. It's not the best one for sure, but it's reliable and 100% guarantee can do the job. By getting another better thermal paste may allow the CPU to run 1-2 degree cooler. That's it, not significant at all.

GC EXtreme is another good choice on the PC world. It's not that common in this forum, but few users report has good result on this one as well.

Anyway, the non conductive paste should not hurt anything if you accidentally make them overflow a bit. However, as we pointed out, the cooling efficiency is mainly base on the correct amount of the thermal paste. If you did that wrong, and overflow it, your CPU most likely will run at a much higher temperature then it should be. On the 4,1, with the high quality Xeon, that generally won't kill anything, but gives you unnecessary high fan noise.

I still believe that at your experence level. It's much better to apply just tiny bit of thermal paste to the CPU, put the heatsink back, wiggle it, then remove the heatsink and check the coverage.

As long as you can see that the CPU is fully covered by the thermal paste, that should be good to go. If not, then add a little bit on the dry area and try again.

This is better then apply too much and have to remove the excessive paste.

 

[mattspace]

No need to make it too complicated.

heh - i wet clean the sensor of my DSLR (still a clenchiningly tense experience), so I'm sure I'll be fine, once I know the parameters for the risks involved.

I guess the difficulty from my perspective is trying to figure out what is likely to cause a short with AS5 (which I already have) - if some squidges out onto the green CPU die, is that going to cause a short, or are we talking people who get it all over the pins underneath etc? Is the gold border around the CPU (X5675) core decorative, or is that a Running Man-esque line of death for AS5?

The AS5 application guide seems quite different from most of the methods recommended for processors with the IHS on, it's very specific about covering the entire cpu core with a thin layer, as opposed to the put a bead on and let the cooler squish it out, but is that just because it's an older product and the instructions are a little outdated? I guess the beadsquish method makes a lot of sense when the IHS is much larger than the core.

If I wanted to get Arctic MX-4, I'd have to order it overseas - noone here sells it any more. It seems Cooler Master is pretty much the only name brand anyone stocks now.

*edit* managed to find MX-4 on Amazon for what seems to be the same price inc. shipping that places were selling it for locally. I'll just have to look at, and occasionally fondle, my new processors while I wait.

 

[DPUser]

Careful with that fondling. We've already had enough oozing talk 'round here.

 

[h9826790]

The AS5 application guide seems quite different from most of the methods recommended for processors with the IHS on, it's very specific about covering the entire cpu core with a thin layer, as opposed to the put a bead on and let the cooler squish it out, but is that just because it's an older product and the instructions are a little outdated? I guess the beadsquish method makes a lot of sense when the IHS is much larger than the core.

http://www.arcticsilver.com/pdf/appmeth/int/vl/intel_app_method_vertical_line_v1.1.pdf

This is directly from AS, and particularly for X56xx / W36xx. On page 5, the last sentence says
"Do NOT spread the line of thermal compound out. When you place the heatsink on the top of the metal cap, the line of thermal compound will spread out like the pink oval pattern on top of the metal cap shown in photo QP4. "

 

[mattspace]

This is directly from AS, and particularly for X56xx / W36xx. On page 5, the last sentence says
"Do NOT spread the line of thermal compound out. When you place the heatsink on the top of the metal cap, the line of thermal compound will spread out like the pink oval pattern on top of the metal cap shown in photo QP4. "

That's the guide for lidded CPUs, the guide for de-lidded CPUs has the opposite instructions, suggesting both a direct application over the surface of the heatsink - "tinting", AND covering the entire surface of the CPU core.

That's the crux of my uncertainty in the whole matter - there are very few resources that I've been able to find for de-lidded CPUs

 

[h9826790]

That's the guide for lidded CPUs, the guide for de-lidded CPUs has the opposite instructions, suggesting both a direct application over the surface of the heatsink - "tinting", AND covering the entire surface of the CPU core.

That's the crux of my uncertainty in the whole matter - there are very few resources that I've been able to find for de-lidded CPUs

I see, you mean this document.

http://www.arcticsilver.com/pdf/appmeth/int/ss/intel_app_method_surface_spread_v1.1.pdf

To be honest, no matter which method, the idea just spread the thermal paste across the CPU die, and make it as thin as possible. I guess because the die size is so different in difference CPU, and even the die location can be different. So AS recommend to use the spread method.

TBH, what me and AidenShaw recommended is basically the spread method, but just use your heatsink to spread. The advantage of use the actual heatsink to spread (and then perform a visual check) is even thought the surface is uneven, you can still guarantee the die is fully covered. If you spread it evenly, but the heatsink copper surface is not even / flat. Some of the area may has thermal paste covered but no actual contact with the heatsink.

Anyway, again, no need to make it too complicated. You can try that a few times, and then realise almost all method give the same result (as long as the amount is right).

 

[mattspace]

yeah, I tink that's what it'll come down to - my concern with using the capacitive AS5 and letting the heatsink spread it out was because the core itself is so small and narrow, that I can see it being very easy to go from not enough to too much, and overflowing the sides onto the die. With the MX-4 on order, that worry is reduced.

 

[MacVidCards]

Always remember that heat transfer of paste is always inferior to that of metal.

You are just trying to fill microscopic pits in surface, ideally bulk of surface is metal to metal.

Should not be "floating" or surfing on paste.

 

[mattspace]

Always remember that heat transfer of paste is always inferior to that of metal.

You are just trying to fill microscopic pits in surface, ideally bulk of surface is metal to metal.

Should not be "floating" or surfing on paste.

I'll bear it i mind