Applying thermal compound on 1155 socket CPUs - dot, line, cross, or multi dot?

[Fjodor2001]

Hi,

I just wonder what method you think is best for applying the thermal compound on 1155 socket CPUs?

According to the manufacturer of Arctic Silver 5, it should be applied in a single line as shown in this PDF (see page 5):
http://www.arcticsilver.com/pdf/appmeth/int/vl/intel_app_method_vertical_line_v1.1.pdf

However, when looking at the Intel stock CPU cooler it has pre-applied thermal compound in three evenly thick "patches". It looks like this:
http://img.tomshardware.com/us/2008...ooler_charts_2008_part_iii___inel_box_tim.jpg

Then finally, this very interesting video shows how the thermal compound spreads under pressure, when different application methods are used:
http://www.youtube.com/watch?feature=endscreen&NR=1&v=EyXLu1Ms-q4

According to that video, the "cross" method seems to be the best. But then how come e.g. Arctic Silver and Intel do not use that method?

Any opinions?

 

[SithSolo1]

Intel is just doing whatever is cheap, mass productable, and works. The line, dot, and cross method will all work about the same. You can also spread it out yourself with a razor or an old credit card.

 

[_Rick_]

as long as you cover roughly the area of the die, and don't get air bubbles, and apply enough pressure, it doesn't really matter.
A line is slightly trickier to do right, as you have to make sure you don't use too much material, and also make sure you squash the line out evenly. If you can, make a thin line slightly offset, along the die-axis, and then use the heatsink's base to squash the line across the die.

A dot is easier to get right though, which in the matters more. Just apply a blob, and give it a decent squish, it should normally cover enough of the heat spreader for there to be no worry.

It also depends somewhat on the compound's viscosity, how well each method ends up working.

 

[Fjodor2001]

You can also spread it out yourself with a razor or an old credit card.

Check out the video I linked to in my original post. Spreading out the thermal compound using e.g. a credit card seems to be the worst method, since it creates air bubbles. Or?

 

[Fjodor2001]

A line is slightly trickier to do right, as you have to make sure you don't use too much material, and also make sure you squash the line out evenly.
[...]
A dot is easier to get right though, which in the matters more. Just apply a blob, and give it a decent squish, it should normally cover enough of the heat spreader for there to be no worry.

Good points. Applying completely even pressure when mounting the CPU cooler is not always so easy.

So a method that is not so sensitive to uneven pressure being applied is of course very good.

 

[bononos]

Hi,

I just wonder what method you think is best for applying the thermal compound on 1155 socket CPUs?

According to the manufacturer of Arctic Silver 5, it should be applied in a single line as shown in this PDF (see page 5):
http://www.arcticsilver.com/pdf/appmeth/int/vl/intel_app_method_vertical_line_v1.1.pdf

.........
Then finally, this very interesting video shows how the thermal compound spreads under pressure, when different application methods are used:
http://www.youtube.com/watch?feature=endscreen&NR=1&v=EyXLu1Ms-q4

According to that video, the "cross" method seems to be the best. But then how come e.g. Arctic Silver and Intel do not use that method?

Any opinions?

All I saw from the video was that the cross method covered a large area of the heatsink but for optimum performance the layer of tim should also be as thin as possible which didn't seem to be the case in the video.

AS5 is a very thick paste that is a little difficult to spread out by just pressing on the heatsink so AS advise is to tint and use the line method.

 

[SheHateMe]

The smiley method was cute!

 

[VirtualLarry]

AS5 is a very thick paste that is a little difficult to spread out by just pressing on the heatsink so AS advise is to tint and use the line method.

I've never known that to be true. Perhaps you had some old dried-out AS5? It spreads fine for me by using the rice-grain/pea-sized dot in the middle, and then using the pressure of the heatsink to squish it out.

 

[Fjodor2001]

All I saw from the video was that the cross method covered a large area of the heatsink but for optimum performance the layer of tim should also be as thin as possible which didn't seem to be the case in the video.

But if there is excess TIM it will be squeezed out on the edges anyway, right? That is of course not desired, but shouldn't hurt either, unless it is so excessive that it drops down on some part of the motherboard and causes a short circuit.

AS5 is a very thick paste that is a little difficult to spread out by just pressing on the heatsink so AS advise is to tint and use the line method.

Actually, I get the impression that the "tint" is not needed. The AS documentation I linked to says:

"Why tint the heatsink and metal cap? Simply put, it will lessen the break-in period. If the break-in period is reduced you will achieve maximum performance in less time"

So, it mainly shortens the break in period as I understand it.

Also, I'm a bit sceptical to tining the CPU cooler/heatsink. The AS documentation described how to perform the tinting:

"Squeeze enough thermal compound onto the center of this area to create a small mound. By working the plastic tool (old credit card) back and forth in all directions (See green symbol in photo HS1) you will smooth out the compound and work it into heatsink. This will ensure optimum filling of the microscopic valleys in the metal where the CPU cores will contact the heatsink."

However, as far as I'm aware spreading out the thermal compound using a credit card risks creating air bubbles when the CPU cooler later is mounted under pressure (as was shown in the video I linked to). So I don't know if that is such a good idea after all? Especially if is not needed, and only is done to reduce the break in period?

 

[Denithor]

I've always just cleaned both faces with 91% rubbing alcohol and then applied the dot in the center. Been mounting these things for 15 years and never had a single problem.

 

[Fjodor2001]

I've always just cleaned both faces with 91% rubbing alcohol and then applied the dot in the center. Been mounting these things for 15 years and never had a single problem.

How do you know that? If you would have used another method, you might have been able to achieve lower temperatures.

 

[Phynaz]

However, as far as I'm aware spreading out the thermal compound using a credit card risks creating air bubbles when the CPU cooler later is mounted under pressure (as was shown in the video I linked to). So I don't know if that is such a good idea after all? Especially if is not needed, and only is done to reduce the break in period?

Next step is to remove the compound, so you have a clean flat surface, with all the microscopic areas filled in.

 

[Phynaz]

How do you know that? If you would have used another method, you might have been able to achieve lower temperatures.

To what benefit? What's a degree or two? I suppose it may matter for IB cooling, but up to this point we haven't been thermally limited.

 

[Fjodor2001]

To what benefit?

Better OC ability. Also, lower fan speed needed => lower noise.

 

[sm625]

According to the manufacturer of Arctic Silver 5, it should be applied in a single line as shown in this PDF (see page 5):
http://www.arcticsilver.com/pdf/appmeth/int/vl/intel_app_method_vertical_line_v1.1.pdf

I would say they are crackheads because you can clearly see in the image labeled QP3 that they've got you running a line of paste right off the edge of the cpu die.

 

[Fjodor2001]

I would say they are crackheads because you can clearly see in the image labeled QP3 that they've got you running a line of paste right off the edge of the cpu die.

Not sure what you mean? The line stretches across the center of the CPU die. Why wouldn't that be desired?

 

[Zap]

There is no ONE CORRECT METHOD for applying thermal compound.

The goal is to have as thin of a layer as possible, without air bubbles, without too much being squished out (wasteful and messy), and covering all portions of the CPU and cooler base that make "contact" with each other.

What "method" works for one compound and set of hardware may not work for another. For instance some coolers notoriously do not have perfectly flat bases, and so you may have to compensate for that. Some compounds are thicker than others, and so you may have to compensate for that. Some coolers have "direct touch heatpipes" with gaps between them, and so you may have to compensate for that.

 

[Phynaz]

Better OC ability. Also, lower fan speed needed => lower noise.

1c or 2c isn't going to make a bit of difference in either of those.

 

[bononos]

But if there is excess TIM it will be squeezed out on the edges anyway, right? That is of course not desired, but shouldn't hurt either, unless it is so excessive that it drops down on some part of the motherboard and causes a short circuit.

Slopping on so much tim until the interface layer is thick and even overflows to the sides is bad since it seems to defeat the purpose of having the tim. Thats just not smart.

Actually, I get the impression that the "tint" is not needed. The AS documentation I linked to says:
.....
However, as far as I'm aware spreading out the thermal compound using a credit card risks creating air bubbles when the CPU cooler later is mounted under pressure (as was shown in the video I linked to). So I don't know if that is such a good idea after all? Especially if is not needed, and only is done to reduce the break in period?

You're not reading the instructions fully- reread all of it and you will see that tinting is creating a hazy layer to fill in the microscopic holes and ruts, which will also work to reduce air bubbles.

Again, that video doesn't confirm the best method since it doesn't measure how thin the tim layer is and doesn't take into account concave/convex shapes or deep ruts on the heatsink base like the popular evo models as Zap mentioned above.

 

[Theguynextdoor]

Honestly it's not really taht big of a deal we're talking 1-2F at most difference.

If you want to do the super thin coat you got to make sure that both the processor and the HS Base as perfectly flat as possible. Check out the thread in my sig to see what I mean.

Again it's a lot of work for very little gain but if you are that anal might as well go all the way.

 

[billyb0b]

it's ok if you put a little too much and is squeezes out of the side some, as long as the paste is non-conductive

like others have said, there is no one single correct way to apply TIM

 

[Fjodor2001]

Slopping on so much tim until the interface layer is thick and even overflows to the sides is bad since it seems to defeat the purpose of having the tim. Thats just not smart.

If the excess TIM will be squeezed out on the edges anyway, does it matter? In that case, applying more TIM than necessary will not increase the TIM thickness.

You're not reading the instructions fully- reread all of it and you will see that tinting is creating a hazy layer to fill in the microscopic holes and ruts, which will also work to reduce air bubbles.

You're right, I missed that the TIM should be swiped off after having applied it on the CPU heatsink when tinting it. Now when reading it again I get the purpose of tinting the CPU heatsink.

 

[Fjodor2001]

1c or 2c isn't going to make a bit of difference in either of those.

You're assuming it only makes 1-2 C difference. It may be more than that. Just the choice of thermal compound itself has been shown in tests to result in 10+ degrees difference...

 

[Fjodor2001]

There is no ONE CORRECT METHOD for applying thermal compound.

The goal is to have as thin of a layer as possible, without air bubbles, without too much being squished out (wasteful and messy), and covering all portions of the CPU and cooler base that make "contact" with each other.

What "method" works for one compound and set of hardware may not work for another. For instance some coolers notoriously do not have perfectly flat bases, and so you may have to compensate for that. Some compounds are thicker than others, and so you may have to compensate for that. Some coolers have "direct touch heatpipes" with gaps between them, and so you may have to compensate for that.

The question is what method is best for socket 1155 CPUs (i.e. Ivy Bridge), as stated in the thread subject.

Also:

* In what way would you compensate for a thermal compound that is thick? I guess you're implying that care has to be taken so that it spreads out properly, but how should that be done? And if there is a way to make the compound spreads more easily, why not use that regardless of compound thickness? Good spread is always desired, right?

* Also, regarding coolers without "flat bases", how would you compensate for that?

Finally, under what circumstance would a "dot" or "spread using credit card" method be better than the "line" method on socket 1155 CPUs (i.e. Ivy Bridge), and why?

 

[KingFatty]

Did the massive analysis done by one of the members here (I think it was idontcare?) put to rest any of these disagreements? If I remember correctly, he tested things like fingerprints, toothpaste, all crazy stuff in addition to very helpful technical/scientific analysis on legitimate products/techniques.