DIY Quick Guide: TIM replacement in 15 minutes. (excellent results)

[KingFatty]

I think that's an irrefutable point: nobody can make a blob/line of thermal compound get squeezed into a perfectly square/rectangle shape by simply squishing it between the heatsink and chip.

So if you use the grain of rice/blob/line method, you have to accept a compromise of the shape mismatch between the flattened blob and chip, which means 1) there will be uncovered sections that aren't reached, or 2) you'll squeeze out over the edges.

Pre-smearing the paste can ensure you never fall under problem 1), but I feel it's also impossible to avoid problem 2). No matter how thinly you smear the blob onto the surface of the chip, some will still be squeezed out the sides when you compress, simply because you've pre-applied it right up to the edges of the chip surface.

That is why I'm a fan of the blob and wiggle technique. I know I'll get some squeezing at the sides, but by wiggling the heatsink after it's on, you ensure the spread is very thin. And you get the hang of how small of a blob to use. Every time I remove the heatsink I have very nice coverage with no missing parts, and very minimal squeeze out at the sides.

 

[wand3r3r]

Every time I remove the heatsink I have very nice coverage with no missing parts, and very minimal squeeze out at the sides.

That is the critical point regardless of the strategy.

There have been a number of good articles which I've read several years ago and they would be beneficial for those new to it.

 

[crashtech]

This is ludicrous, I have no idea where you got that from.

From the wiki, not that that is where I got my information, but it will suffice to school the ignorant:

Isopropyl rubbing alcohol USP/B.P. contains 68–99% of isopropanol (isopropyl alcohol) by volume, the remainder consisting of water, with or without color additives, suitable stabilisers, and perfume oils.

If you want alcohol with a very low water percentage, use denatured alcohol from the hardware store, not isopropyl from the drug store.

 

[KaRLiToS]

I've taken apart CPU heatsinks many times and I've tried beads vs. lines vs. spreading it out, after reading a lot of articles about it and seeing the results. Personally I prefer to spread it out on the CPU since I usually put a very thin layer which will just be enough to suffice without being excessive. I've opened my CPUs and GPUs after to confirm the spread and I'm just doing what I've personally figured out how to do it. Everyone has an opinion, go read some articles and form your own.

I think cleaning it is important, but it was very clean and shiny when I was done.

The GPU has bottoming out screws, you cannot put the cooler on as tight as you want. If you plan to squeeze the drop in the middle all the way to all the edges, you have to apply it liberally. I prefer to ensure it makes it everywhere and is sufficient. I have opened GPUs up after applying it to see what's happening, I suggest you do too when you are doing this for the first time(s).

When I put the GPU heatsink back on I line up the screw holes carefully and put it directly on. If you only use a grain of rice style you have to ensure it spreads (whether you twist it a little or try roll it etc.). Having opened up grain of rice type applications and seeing portions of the heatsink cover missing TIM has made me wary. To each their own.

Everyone has a point, and everyone has an opinion. It's best to try make the most informed decision possible since you can potentially wreck something. This was just one way to do it and a demonstration of how easy it can be.

Well said mate, you have PM. (freebie)

 

[brag.yondide]

Strikes me that the die of a CPU occupies only a portion of the area under an IHS. So the rice grain or line method, which aims to apply TIM to the area of the IHS over the die, makes sense (if you believe that the unoccupied area has minimal effect in heat transfer).

In contrast, a GPU does not have an IHS, so applying TIM to the whole die seems consistent with the CPU rice or line method. Were you to apply the rice grain or line to the GPU, there is a danger of not covering the whole die thereby leaving potential heat producing surfaces uncoupled to the heat sink.

Whether application by spreading can avoid air gaps is another question.

 

[MrK6]

From the wiki, not that that is where I got my information, but it will suffice to school the ignorant.

You can scrub your PCB with soap and water and it will be fine. The little bit of water in isopropyl alcohol vs. any other means little. Honestly, have a clue before you try to be condescending.

 

[crashtech]

You can scrub your PCB with soap and water and it will be fine. The little bit of water in isopropyl alcohol vs. any other means little. Honestly, have a clue before you try to be condescending.

So you didn't know store bought alcohol had water in it. Glad I could be of help. What kind of soap should I use on my motherboard? Is Tide OK? Should I use gentle cycle?

 

[TerryMathews]

So you didn't know store bought alcohol had water in it. Glad I could be of help. What kind of soap should I use on my motherboard? Is Tide OK? Should I use gentle cycle?

Make sure you use cold water, don't want your CPU to shrink.

 

[crashtech]

Seriously, let me try and be more clear about what I was trying to say:

1) Store bought isopropyl alcohol contains varying amounts of water.

2) There's no compelling need to use solvents to remove most TIMS satisfactorily.

3) The chance that some of the water component of the store-bought alcohol might remain under a surface-mounted part is non-zero if someone is not careful.

 

[KaRLiToS]

Seriously, let me try and be more clear about what I was trying to say:

1) Store bought isopropyl alcohol contains varying amounts of water.

2) There's no compelling need to use solvents to remove most TIMS satisfactorily.

3) The chance that some of the water component of the store-bought alcohol might remain under a surface-mounted part is non-zero if someone is not careful.

No power = no problem. Everything is based on water (90% alcohol = 10% water).

Just let it dry before and anyway, by the time you remount the block or cooler and put the power on, the water has plenty of time to dry.

By the way, what do you use to clean paste? Sand paper?
Being meticulous is a must when repasting.

About your last post. YES it needs to be perfectly clean, even finger grease will cause some bad thermal transfer, even a small particle of dead skin is bad.

So yes, alcohol and lint free cloth is important. By the way, I dropped some water on my components very often but always let them dry. Never had any issue. I even drop some water in a CPU socket couple of times and after letting it dry, it was fine.

Like I said, if you drop water, as long as you let it dry before powering.

 

[MrK6]

Seriously, let me try and be more clear about what I was trying to say:

1) Store bought isopropyl alcohol contains varying amounts of water.

2) There's no compelling need to use solvents to remove most TIMS satisfactorily.

3) The chance that some of the water component of the store-bought alcohol might remain under a surface-mounted part is non-zero if someone is not careful.

The problem is you came to a (seemingly) instructional thread spouting all kinds of nonsense and now you're upset you got called on your ignorance. Everyone who's taken chemistry past middle school knows there's water in isopropyl alcohol, and no one cares because it doesn't matter. Except for you, evidently you thought this was some big secret and the fact that you're trying to blow it up into something shows how little knowledge you have on this topic. Furthermore, the fact that you believe "there's no compelling need to use solvents to remove most TIM's satisfactorily" shows you again don't have the slightest clue what you're talking about. It's comical at this point and you need to stop replying.

People reading this thread hoping for some knowledge: TIM's exist to enhance heat transfer to the heatsink, everything from air, to skin oils, to dead skin cells in between those surfaces matter, and YES you need a solvent to remove them properly. Isopropyl alcohol works, acetone works, anything that is of decent purity and evaporates quickly for convenience.

 

[KaRLiToS]

the problem is you came to a (seemingly) instructional thread spouting all kinds of nonsense and now you're upset you got called on your ignorance. Everyone who's taken chemistry past middle school knows there's water in isopropyl alcohol, and no one cares because it doesn't matter. Except for you, evidently you thought this was some big secret and the fact that you're trying to blow it up into something shows how little knowledge you have on this topic. Furthermore, the fact that you believe "there's no compelling need to use solvents to remove most tim's satisfactorily" shows you again don't have the slightest clue what you're talking about. It's comical at this point and you need to stop replying.

People reading this thread hoping for some knowledge: Tim's exist to enhance heat transfer to the heatsink, everything from air, to skin oils, to dead skin cells in between those surfaces matter, and yes you need a solvent to remove them properly. Isopropyl alcohol works, acetone works, anything that is of decent purity and evaporates quickly for convenience.

amen

+100

 

[crashtech]

I'm sure you guys would never want to believe you are actually wasting your time using solvents when 99% of used TIMs can be buffed off completely with clean paper towels, but it's true, and especially for noobs, it's less risky. There's no evidence to support the assertion that solvent cleaning improves thermal transfer. It just makes you guys feel better because it makes everything bright and shiny.

Now I will concede that if there is gunk that won't come off, of course a dab of solvent on the towel will make short work of it. But it just doesn't need to be SOP. It's a waste of time.

 

[crashtech]

Something else to note about this mostly pointless argument: Once the horrible OEM thermal pad is scraped off and replaced with decent TIM, pretty much everything else matters more than the TIM. The size of the heatsink, the mounting pressure, and the surface finish of the interface all matter more than squabbling over whether there is a fingerprint or a speck of dust in there.

In some cases, the mounting setup is designed for a thermal pad of much greater thickness than the TIM that is intended to be applied. This is especially true in laptops, but some GPUs have shouldered mounting screws that may not allow a thin layer of TIM to be properly compressed. If this is suspected, it's advisable to pull the assembly back apart and view the contact pattern of the TIM.

 

[KingFatty]

It's not fully accurate to describe various techniques as "wasting time."

However, I think you would be fully accurate to say that a particular technique, whether used or not, will result in zero measurable temperature difference.

For example, consider fingerprints. You could say that leaving a fingerprint (typically caused by human finger oil) on the GPU or heatsink will not result in any measurable temperature difference, so it's fine if you put thermal compound on top of a fingerprint.

People do things to relax their worry/minds and have fun. I can see applying thermal paste being something like this, like computer yoga. It's just kind of neat to do it, and if you have a particular routine/technique, it's fine to do that even if it's just for the mental relaxation and feeling of safety, even if there is no measurable difference. I mean, you'd still wipe your bottom even though you had a perfect bowel movement, just to maintain the routine and preserve mental sanity.

 

[crashtech]

I can agree with that. Perhaps this hobby attracts a disproportionate number of OCD sufferers. I tend to be overly meticulous about certain things, but TIM placement is not one of them.

 

[Schmide]

Dude. I use windex, clean up wipes, alcohol, eyeglass cleaner, etc. If it cleans, I'll use it. Never had an issue. All these items evaporate very fast and are generally just reactive with oils.

Kirkland Cleanup Wipes are by my standards the best for removing material.

 

[Tristor]

It's not fully accurate to describe various techniques as "wasting time."

However, I think you would be fully accurate to say that a particular technique, whether used or not, will result in zero measurable temperature difference.

For example, consider fingerprints. You could say that leaving a fingerprint (typically caused by human finger oil) on the GPU or heatsink will not result in any measurable temperature difference, so it's fine if you put thermal compound on top of a fingerprint.

People do things to relax their worry/minds and have fun. I can see applying thermal paste being something like this, like computer yoga. It's just kind of neat to do it, and if you have a particular routine/technique, it's fine to do that even if it's just for the mental relaxation and feeling of safety, even if there is no measurable difference. I mean, you'd still wipe your bottom even though you had a perfect bowel movement, just to maintain the routine and preserve mental sanity.

That's all fine and dandy, except that there /is/ a measurable difference when a fingerprint gets on the surface of the CPU beforehand. There's far worse things that could happen than a fingerprint getting on there, but it does make a difference and since there's minimal effort required to ensure that these things aren't a problem it's just simple common sense to clean the two mating surfaces properly before applying TIM. Doing anything less is just laziness.

I don't need to be Monk in order to take 5 seconds out of my busy schedule to do whatever project I'm working on correctly. It's like building furniture. Do you have to round off the edges with sandpaper after you're done? No. But do you want to cut yourself the next time you go to pick it up and move it? Well, then take the time to do it right.

 

[tolis626]

I don't know about you guys,but I use 90-95% ethanol and a paper towel to wipe off the majority of the TIM,and then use a lint-free cloth (Mainly the ones for cleaning glasses) again with ethanol and wipe off any residues,dust and fingerprints.Then I do the rice method on the CPU.I've never replaced the TIM on a GPU though...

 

[KingFatty]

That's all fine and dandy, except that there /is/ a measurable difference when a fingerprint gets on the surface of the CPU beforehand. There's far worse things that could happen than a fingerprint getting on there, but it does make a difference and since there's minimal effort required to ensure that these things aren't a problem it's just simple common sense to clean the two mating surfaces properly before applying TIM. Doing anything less is just laziness.

I don't need to be Monk in order to take 5 seconds out of my busy schedule to do whatever project I'm working on correctly. It's like building furniture. Do you have to round off the edges with sandpaper after you're done? No. But do you want to cut yourself the next time you go to pick it up and move it? Well, then take the time to do it right.

You are correct that a fancy enough temperature probe could measure the difference in a lab setting, but for our purposes/discussion, the built-in temperature measuring sensors within the chip/motherboard etc. are unable to measure the difference. You simply won't see it, even if you use your own temperature probe that is average/for mere mortals and not professional labs.

But another way to approach this, is how do you think the temperature difference caused by a fingerprint would manifest? Would you guesstimate it might cause the chip to be 0.00000001 degree C hotter? I think a fancy temperature probe could detect that, but we don't have access to that and the computer won't see it.

Would this minuscule difference cause your fan spin faster? No, the fan controller cannot change the RPM based on that unmeasurable difference. Would you feel more heat in the room? No, your skin is not that sensitive. I cannot think of any way that the unmeasurable temperature difference would manifest in a computer.

But I would still do my routine and make sure there are no fingerprints on my chip. It just ruins my mental vision of how I want my computer equipment to be installed. I just want to be realistic about something like a fingerprint. I guess every effort should be balanced with the temperature effect.

For example, people are willing to polish/lap the surface of their chip. It has a measurable difference, and many people feel the payoff of slightly lower temps is not worth the effort. Same for de-lidding. You balance the effort vs benefit.

There is a lot more data for quantifying the benefits of lapping and delidding, but not as much for fingerprints. But I think a user Idontcare did a fascinating analysis of thermal compounds and variations (I think he even used toothpaste) so I'm hoping I recall his data correctly regarding how fingerprints did not matter. Same with wiping off with a paper towel only, vs using a solvent and lint-free cloth (I think it doesn't matter whether you use solvent/lint-free and doesn't show a temperature difference that is measurable, just like fingerprints, but I'd have to re-read the analysis).

 

[lavaheadache]

I see that this thread has spiraled into people blowing the whole TIM application process out of proportion yet again. This subject seems more controversial than abortion.

 

[Kalessian]

http://www.youtube.com/watch?v=EyXLu1Ms-q4

 

[crashtech]

None of our assertions have been backed up by data as yet. As KingFatty mentions, I remember something about IDC doing some testing on the subject, but it will take a while to wade through that thread. I think at this point some links to data would be more beneficial than continuing to argue fruitlessly. Any reader should take ALL the currently expressed opinions with a grain of salt, including and perhaps especially mine.

 

[Torn Mind]

Here's my anecdote

I replaced the TIM on my Powercolor 7970 with that awesome Noctua past. It can now stay under 75 degrees Celsius without the need for a big fan blowing air into the already open case. The fans used to run at full blast at lower ambient room temps, such as 58 degrees Farenheit.

Heatsink was "flat", so no worries about grooves or anything.

I used a "triangle pea" method to apply my TIM, and I didn't even fully clean the heatsink TIM off(residue was still on the copper), although the GPU was much more easy to make "shiny". I left some of the old "pumped out" TIM on since it wasn't touching the die.

The heatpipes were capable of burning skin prior to the TIM change. I guess they are even more capable now.

 

[KingFatty]

Another issue:
The surface of the heatsink itself (or CPU's heat spreader) can be concave.

So you will have poor contact between them due to the concavity, no matter how clean they are, you'll have that concave gap. You could lap it, but if you don't lap it, I would think applying enough thermal compound to "fill the gap" formed by the concavity would be better than leaving an air gap in the concavity. But that's really hard to quantify how much extra you'd need, because how can we calculate the volume of a concavity like that? I'd rather err on the side of caution and put just enough to squish out the sides, to make sure I've filled that concavity without leaving an air gap. Same goes for having them being convex/bulging out - the extra paste could help with thermal transfer and avoid gaps around the edges.