Did I apply too much thermal paste?

[ChoGGi]

Mistakes are a part of learning

Besides he seems to be more cautious then anything after stripping a single screw, so your concern might be slightly misplaced?

 

[SomeGuy133]

Mistakes are a part of learning

Besides he seems to be more cautious then anything after stripping a single screw, so your concern might be slightly misplaced?

depends the backplate for the CPU. Some of them stripe really easily due to cheap backplate.

I had a couple streipe due to this cheap plastic design.

 

[dvsman]

Me: "I use the "few drops + credit card smear" technique. I go by the idea that the point of the thermal compound is to fill in the microgaps between the CPU and cooler, not act like the filling in an icecream sandwich."

Someguy133 "again thats a bad method for longevity because it causes air bubbles and the pasting will go bad after a month or two on a heavily overclocked rig and temps will go up 5-10C depending on the system. (very true for laptops or heavy voltaged OC rigs)"

Nope. My layers are even and smoothed out across the entire CPU/GPU/whatever I'm trying to put together. I put just enough on to ensure solid, 100% chip to metal contact. Never had a problem and my temps are stable until I decide to take it apart again to do some other build / experiment. I don't need any crazy OC, maybe 4.2ghz -ish so my temps are higher than stock but not crazy high. It all just depends on what TIM you use and how meticulous you are about putting it all together. If you half-ass it, no matter what the TIM, it's going to produce crappy results.

Plus, I don't need to ask anyone about how to do it :-D

 

[SomeGuy133]

Me: "I use the "few drops + credit card smear" technique. I go by the idea that the point of the thermal compound is to fill in the microgaps between the CPU and cooler, not act like the filling in an icecream sandwich."

Someguy133 "again thats a bad method for longevity because it causes air bubbles and the pasting will go bad after a month or two on a heavily overclocked rig and temps will go up 5-10C depending on the system. (very true for laptops or heavy voltaged OC rigs)"

Nope. My layers are even and smoothed out across the entire CPU/GPU/whatever I'm trying to put together. I put just enough on to ensure solid, 100% chip to metal contact. Never had a problem and my temps are stable until I decide to take it apart again to do some other build / experiment. I don't need any crazy OC, maybe 4.2ghz -ish so my temps are higher than stock but not crazy high. It all just depends on what TIM you use and how meticulous you are about putting it all together. If you half-ass it, no matter what the TIM, it's going to produce crappy results.

Plus, I don't need to ask anyone about how to do it :-D

again its been shown that this method causes more air bubbles and running high temps it degrades faster than pea method. Its common knowledge. 4.2Ghz your temps are low and you won't notice. Run rigs at 70-90C for months. that method lasts 1 month at best before temps start to go shit.

 

[ChoGGi]

I used to have an overclocked FX 8350. I think it maxed out around 80 something, I used the spread method on it fine for years.
I switched to doing the line method now, but that's a matter of laziness rather then temperature worries.

Hell I did the same with a 290x and an AIO with a bad pump (well something stuck in the lines), that thing regularly hit 94 or 95 till I tapped the pipes and it went away temporarily.

 

[pendragon1]

I used to have an overclocked FX 8350. I think it maxed out around 80 something, I used the spread method on it fine for years.
I switched to doing the line method now, but that's a matter of laziness rather then temperature worries.

Hell I did the same with a 290x and an AIO with a bad pump (well something stuck in the lines), that thing regularly hit 94 or 95 till I tapped the pipes and it went away temporarily.

umm isn't the max safe temp for the 8350 ~65C like all of the FX8xxx chips?? (I just googled and I guess nobody can agree what the max is, 62, 72, 78, or 90. so... nm.)
line method is fine too and thats kinda what I do. for the fx chips you look at the chip so you can read the etching right side up and put a line across the middle about 3/4 of the width of the chip. that's where the cores are, sitting side be side. I then put tiny little dabs in the open areas. ive never spread it as I always thought that could potentially introduce contaminants into the paste, dust, lint, hair, skin cells, etc.

that aio may not have a bad pump but a clog in the lines. I was given a h60 what wasn't working but one of the barbs was 75%+ clogged with idk what but it looked like calcium deposit. I cleaned it all out, put on new tubes and strapped it on to my 280x. works great!!

 

[ChoGGi]

umm isn't the max safe temp for the 8350 ~65C like all of the FX8xxx chips??

Doubtful it's a CPU, they always have high temp ratings. I always run stuff hot with slower fans (don't care for noise).

that aio may not have a bad pump but a clog in the lines.

Yeah it has something in the line, but I had another AIO of the same design sitting around that I fit onto the cooler structure. I don't use them (8350/290x) anymore though, switched to Intel last year (got tired of waiting).

 

[pendragon1]

Doubtful it's a CPU, they always have high temp ratings.

intel has high temp, 80+, amd has been MUCH lower for a while now. but nobody can agree on what. I know that by default my chip(fx-8120) was set to shutdown at 70c. so idk... I keep mine below 60c just in case.

I don't use them (8350/290x) anymore though, switched to Intel last year (got tired of waiting).

yeah ive seen that a lot! cant really blame anybody for switching with they way AMDs been for so long. if I had the cash...

 

[drklu]

I tried using a cc to apply the paste evenly across the cpu. In my experience this caused the paste in the center of the cpu to squish outward leaving a thin or lack or coverage completely in the middle. That is why I advocate using a pea sized dot in the center, that way it has plenty of paste to spread outward. YMMV

 

[ChoGGi]

intel has high temp, 80+, amd has been MUCH lower for a while now. but nobody can agree on what. I know that by default my chip(fx-8120) was set to shutdown at 70c. so idk... I keep mine below 60c just in case.

Shit it could be then, it has been awhile

 

[rgMekanic]

It doesn't matter how I apply it, I have the most PITA waterblock in the world to install, so as soon as I try to get it going, its fucked and smeared and lopsided anyway.

Temps still 36* under load.

 

[pendragon1]

this caused the paste in the center of the cpu to squish outward leaving a thin or lack or coverage completely in the middle

I'm pretty sure that that is normal and if you look at it after the pea method you'll still see the same thing. most chips with an IHS are now convex afaik...

 

[pendragon1]

Shit it could be then, it has been awhile

actually I think someone told me intels throttle temps is ~100c and ~105 for shutdown. most try to keep the 80c or lower. to me that's just insane temps!

 

[ChoGGi]

My 5930K does 78C at decent load, though the ambient temp is cooler now

 

[SomeGuy133]

Doubtful it's a CPU, they always have high temp ratings. I always run stuff hot with slower fans (don't care for noise).
Yeah it has something in the line, but I had another AIO of the same design sitting around that I fit onto the cooler structure. I don't use them (8350/290x) anymore though, switched to Intel last year (got tired of waiting).

P4 and before had a 60C rating IIRC. These 100C ratings have not always been around.

http://ark.intel.com/products/27478...g-HT-Technology-2M-Cache-3_00-GHz-800-MHz-FSB

It doesn't matter how I apply it, I have the most PITA waterblock in the world to install, so as soon as I try to get it going, its fucked and smeared and lopsided anyway.

Temps still 36* under load.

again at 36C load application method doesn't really matter but when you have 70-90C underload on 120-180 W TDP systems the method matters a lot for temps and longevity.

actually I think someone told me intels throttle temps is ~100c and ~105 for shutdown. most try to keep the 80c or lower. to me that's just insane temps!

100C or 105C depending on generation and ideally under 70/80C is good. I have run a 920XM at 90C for a year streight at 3.33GHz at 70 W TSP. Wanted to do 4GHz with PLL mod but was too lazy to learn how and make cooling better.

 

[doyll]

Many people think they need a layer of TIM between IHS and cooler / waterblock. This is not true.

Direct Metal to Metal, No TIM Layer In-between!
Be careful not to apply too much TIM. We want only enough to fill the voids in the crystalline structure of the metal where it does not give direct metal to metal contact. We do not what TIM to form a layer between the IHS and cooler base
In case you do not know, TIM heat transfer is not even close to what good metal to metal contact is, but it is much better than what air transfers when there is no TIM in the voids created by the pores of the metal structure.

Heat tranfer ability of different things:

Copper is 400 W/(m·K) heat transfer
Aluminum 210 W/(m·K))
Lead is 35 W/(m·K)
Typical TIM is rated 3.5-10.5 W/(m-K) (GC Extreme 8.5W/(m·K), Chill Factor III 3.5W(m-K) in use this is only 1-2c difference in CPU temp)*
Air is 0.024 W/(m·K)​

To put that into perspective, any way you look at it metal to metal is 50 times better than TIM:

Copper is 47 times better than Gelid GC Extreme TIM & 114 times better than Chill Factor III
TIM is 333 times better than air at sea level.​

*i don't know how accurate the W(m-K) ratings are onl TIM, but do know the temperature differences between top 30 or 40 TIMs are all within a degree or two, except for liquid metal variants .. which are a degree or two better.

 

[SomeGuy133]

Many people think they need a layer of TIM between IHS and cooler / waterblock. This is not true.

Direct Metal to Metal, No TIM Layer In-between!
Be careful not to apply too much TIM. We want only enough to fill the voids in the crystalline structure of the metal where it does not give direct metal to metal contact. We do not what TIM to form a layer between the IHS and cooler base
In case you do not know, TIM heat transfer is not even close to what good metal to metal contact is, but it is much better than what air transfers when there is no TIM in the voids created by the pores of the metal structure.

Heat tranfer ability of different things:

Copper is 400 W/(m·K) heat transfer
Aluminum 210 W/(m·K))
Lead is 35 W/(m·K)
Typical TIM is rated 3.5-10.5 W/(m-K) (GC Extreme 8.5W/(m·K), Chill Factor III 3.5W(m-K) in use this is only 1-2c difference in CPU temp)*
Air is 0.024 W/(m·K)​

To put that into perspective, any way you look at it metal to metal is 50 times better than TIM:

Copper is 47 times better than Gelid GC Extreme TIM & 114 times better than Chill Factor III
TIM is 333 times better than air at sea level.​

*i don't know how accurate the W(m-K) ratings are onl TIM, but do know the temperature differences between top 30 or 40 TIMs are all within a degree or two, except for liquid metal variants .. which are a degree or two better.

tl/dr
ignore him about too much TIM......

again....TIM will never....NEVER...BLOCK METAL TO METAL contact... If you honestly think metal to metal won't push TIM out of the way with just pressure alone let alone the added heat increasing liquidity you are seriously stupid.

if anyone claims to prove TIM prevents metal to metal contact they are totally full of shit. If they claim to have proof and aren't trying to bullshit you they simply didn't apply enough pressure....again 60-100PSI is the pressure you want and no metal to metal contact is going to be blocked by TIM

ROFLCOPTER

Also your 1-2 C difference is only realavent on low temp/low TDP systems. When you get into high TDP systems like 100-220 TDP TIM can make a 10-20C difference. AS5 is shit these days for those use cases and same with TEC systems. TIMs make a large difference.

Also depends on the heatsink. In laptops you can see a >10C difference due to very poor heatsink contact with the die and thick quality TIM is critical.

Anyone with experience with gaming laptops knows this.

 

[Susquehannock]

^^ Now you are just being childish. Stop trolling and do some reading on fluid dynamics. Then maybe you can understand Couette flow, and how solid particles within a viscous liquid can significantly affect drag force between two parallel plates.

 

[SomeGuy133]

^^ Now you are just being childish. Stop trolling and do some reading on fluid dynamics. Then maybe you can understand Couette flow, and how solid particles within a viscous liquid can significantly affect drag force between two parallel plates.

k show me how ICD at 80C and 100PSI is going to cause a problem with that equation. go do the math and prove it to me.

 

[doyll]

SomeGuy1332[H]4U
Please supply some links to test data supporting your claims.

Here is old test data clearly show showing how amount of TIM effects heat transfer.
http://www.hardwaresecrets.com/what-is-the-best-way-to-apply-thermal-grease-part-1/

 

[SomeGuy133]

SomeGuy1332[H]4U
Please supply some links to test data supporting your claims.

Here is old test data clearly show showing how amount of TIM effects heat transfer.
http://www.hardwaresecrets.com/what-is-the-best-way-to-apply-thermal-grease-part-1/

experience from testing Heat-Spring, ICD, MX-4, and AS5 on a slew of heatsinks.

Question. Did they say how much pressure was used? what TIMS and cure times? many TIMS take several days to fully cure. If they didn't allow cure times that study was a joke. Did you ask basic questions like this before posting said link and check for bullshit?

As i said, nearly everytime someone claims this they have a flawed study or failed to apply it right. indium corp and ICD parent company both agree that pea method is best (on squares...line method for rectangle like laptop direct die application) and countless reputable places agree on this. I am shocked this even comes up here and again i am talking about high TDP and high temp systems. All these review places use stock quads or very low overclocks. All my experience is with 80 W TDP laptops and desktops pushing 120-220 W TDP and these act very differently because they push things to the extreme.

If you ever post on laptop forums you will quickly see this is true. No one on here uses gaming laptops so maybe that is why everyone doesn't understand high temp/TDP systems. Any clevo/alienware owner should know line method is best (pea for squares)

Notebookreview F@H guys will tell you that MX4 is shit and lasts at best 2 months and ICD lasts at least 3-6 months depending on how good the laptop heatsink is.

TIM applicant method doesn't matter at stock levels because 50C v 54C is meaningless but when you are running at 80C or 90C...5 or 10C can mean extra MHz or bricking CPU.

Again longevity is a major part of this. you can probably not even use TIM of those rigs that run at 50C or if TIM craps out it won't matter but I have run F@H on many rigs for several years and when running at 90C any temp change is bad. MX4 would last 1-2 months. ICD would last 3-6 months and Heat-spring I have never ever replaced because it can never go bad.

My old m17 R4 still had heat-spring in it from 2012 lol

 

[doyll]

experience from testing Heat-Spring, ICD, MX-4, and AS5 on a slew of heatsinks.

Question. Did they say how much pressure was used? what TIMS and cure times? many TIMS take several days to fully cure. If they didn't allow cure times that study was a joke. Did you ask basic questions like this before posting said link and check for bullshit?

As i said, nearly everytime someone claims this they have a flawed study or failed to apply it right. indium corp and ICD parent company both agree that pea method is best (on squares...line method for rectangle like laptop direct die application) and countless reputable places agree on this. I am shocked this even comes up here and again i am talking about high TDP and high temp systems. All these review places use stock quads or very low overclocks. All my experience is with 80 W TDP laptops and desktops pushing 120-220 W TDP and these act very differently because they push things to the extreme.

If you ever post on laptop forums you will quickly see this is true. No one on here uses gaming laptops so maybe that is why everyone doesn't understand high temp/TDP systems. Any clevo/alienware owner should know line method is best (pea for squares)

Notebookreview F@H guys will tell you that MX4 is shit and lasts at best 2 months and ICD lasts at least 3-6 months depending on how good the laptop heatsink is.

TIM applicant method doesn't matter at stock levels because 50C v 54C is meaningless but when you are running at 80C or 90C...5 or 10C can mean extra MHz or bricking CPU.

Again longevity is a major part of this. you can probably not even use TIM of those rigs that run at 50C or if TIM craps out it won't matter but I have run F@H on many rigs for several years and when running at 90C any temp change is bad. MX4 would last 1-2 months. ICD would last 3-6 months and Heat-spring I have never ever replaced because it can never go bad.

My old m17 R4 still had heat-spring in it from 2012 lol

So you have no data, no pressure measurement, notTIMs, and no cure times to support your claims, but you ask me questions about an independent source instead of reading how it was done from link provided.

I have supplied supporting data.

You have supplied lots of talk .. most of which is irrelevant, but you have supplied noting to verify any of what you are claiming. I don't know you, but I'm not about to take your word base on claims posted in this thread.

 

[Nobu]

Arctic Ceramique, Dot in center and be done with it. If you aren't satisfied, do it again or lap the IHS and heatsink and mount with little or no TIM. If that's too [H]ard for you, gtfo.

 

[SomeGuy133]

So you have no data, no pressure measurement, notTIMs, and no cure times to support your claims, but you ask me questions about an independent source instead of reading how it was done from link provided.

I have supplied supporting data.

You have supplied lots of talk .. most of which is irrelevant, but you have supplied noting to verify any of what you are claiming. I don't know you, but I'm not about to take your word base on claims posted in this thread.

cure times are set by the TIM manufacturers....is that complicated? Go read OEM website. As i said I have used 100s of the fuji film pressure papers and consistent application is impossible due to how it works and how heatsinks are applied. If you don;t know this than you have zero experience in this area and need to stop talking.....period.

How about you learn what you are talking about first?

Anyone that has done this knows you want 60-100 PSI. Go talk to anyone with custom made TEC blocks at overclock.net.

You are just trolling now.

I never posted any of my pressure measurements due to the way fujipilm paper works. It is pointless to post it because it isn't reliable. unless you can 100% replicate the application. That means screwing all 4 screws in; at the same speed, at the same time, same pressure; leave for 10 seconds (IIRC) remove all 4 screws at the same speed, same time. It just isn't possible. So all all pressure paper results are complete trash.

Unless you are solely looking for complete contact and not accurate pressure results. That is the only thing it is good for.

Again for pressures look at indium corps website. They have tons of data on this. hold on...let me baby feed you.

http://www.indium.com/thermal-interface-materials/heat-spring/#products

If you have read or have any experience with TECs you know you want again min 60 PSI if not 100 PSI.

CPUs can handle 60-100 PSI perfectly fine.

OMG....look...common knowledge!!!!
http://www.innovationcooling.com/data_sheet.html

Cure time of 2 hours if greater than 50 PSI and it works best at >60 PSI!!!! AMAZING!!!!!

Again the source you posted is complete trash because it didn't state PSIs, cure times, or anything!

The fact you don;t know this means you never bothered to even read the directions that comes with whatever TIM you have used....if you ever used one.

Also as i said if you have any experience with a lot of systems TIMs act very very differently when you use 100-220W TDP systems or worse.....800 W TEC systems. The differences in TIMs are substantial. Plus higher the temp the more different they act.......basic......

EDIT: oh and your irrelevant claim is hogwash. I am trying to teach you to understand TIMs function substantially different in different envirorments, low and high pressure applications (look at H-S and ICD data vs other TIMS. ICD/H-S require high pressure to be good)

Higher the temp or more heat going through the TIM makes substantial difference. Some TIMS break down, others bleed, others are more prone to air bubbles, some can't keep up with the amount of heat while others can.

This is where you see certain high quality TIMS shine. When you run 80-90C and 200 W TDP cheap/low quality TIMS are trash but at 50 W TDP and 50C the TIM you use is meaningless. Same with the application.

Application is critical when you are running high powered systems. Applications don't just affect how well it cools but greatly affects its longevity. Do you want to repaste a server in a data center every 2 months because of a crappy TIM or application? Obviously not.

 

[doyll]

Please 'baby-feed' me some more and show me the flaws in my data source versus yours.
Plase remember I'm a 'baby' and need very detailed and simple explainations.

 

[doyll]

Arctic Ceramique, Dot in center and be done with it. If you aren't satisfied, do it again or lap the IHS and heatsink and mount with little or no TIM. If that's too [H]ard for you, gtfo.

While a little TIM is probably enough, no TIM does not work .. unless you solder the block directly to CPU.

 

[Nobu]

While a little TIM is probably enough, no TIM does not work .. unless you solder the block directly to CPU.

It works if you do it right, though it may not be optimal due to imperfections and expansion of differing metals. There were some videos on this thread, but they've bit the dust.

 

[doyll]

It works if you do it right, though it may not be optimal due to imperfections and expansion of differing metals. There were some videos on this thread, but they've bit the dust.

The surface of metal is not smooth .. never! ever! This is becuase metal crystaline in structure, so no matter how smooth it is polished it still has miccroscopic voids in the finish. When we put 2 of these surface together these microscopic voids create open spaces between the two surfaces. TIM is what is used to fill this spaces.

The problem is many people use way more TIM than they should and because there is not a great deal of pressure pushing the cooler block onto the base themore TIM that is used (and the thicker consistency the TIM is) the more stays between the two metal surfaces. As I posted a ways back the thermal transfer rate of TIM is only 5-8 W/(m·K) compared to copper being 400 W/(m·K) and aluminum 212 W/(m·K) .. kinda obvious how much better metal to metal is, but air is only 0.024 W/(m·K) so TIM is way, way better than nothing (air).

I tried uploading an image to show what I'm talking about but it won't load. Here is a link to more info .

 

[Nobu]

I know all of this. Despite the facts, it still works, just not optimally.

 

[doyll]

Fact is it does not even come close to working even acceptably.
More like it manages to marginally work in spite of all it's limitations.

 

[bigdogchris]

In the past I used the line method because that's what Arctic Silver tells you to do (I imagine they know the right way to do it). Now I use the pea method.

 

[RazorWind]

I have a be quiet dark rock pro tf and I installed it on my board with an i7 6800k yesterday. I looked up close and it is not squishing out, but I can see a little bit where the cooler comes in contact with the CPU. I took a picture after I applied the thermal paste before I put the cooler on.
View attachment 11557

That's more than you really need, but well within the bounds of "acceptable."

 

[RazorWind]

I read online that it would make the thermals bad if I put on too much. I guess they were referring to the clip on coolers. I was in doubt because there are so many methods of applying thermal paste. I feel like people here are attacking me like sharks. Can't you just drop it?

That's theoretically true, but too much is like, the whole tube. You'd almost have to try to use an amount that is actually too much.

 

[CarpCharacin]

I did pretty much use most if the tube.

 

[Susquehannock]

Wonder how Dell applies their TIM. Looked pretty tired in this T3500 workstation. Pressure and heat definitely did NOT squeeze it out after 5.5 years.

Thin layer of Ceramique dropped full load temps from 94 to 76 C.

And yeah ... the AS instructions recommend a "surface spread" method on Xeon family server CPU.

 

[Krenum]

Here is a video to put your mind at ease.

 

[doyll]

The above video is another 'talking head' trying to become popular on YouTube by spouting things that are obviously not accurate or true to people who do have knowledge on the TIM application subject. Sorry if that sounds egotistical, but I can find too many obvious errors in his testing procedure to consider his finding anything but a waste of time.

If you look at 6:43 into video, the print of 'too much thermalpaste' is not good. The thinnest area of print is not over the CPU chip, but offset.

Compare the 'too much' print with the 'dot' print at 2:54 and we can see how the

His increase in temps with 'too little' only shows the application at 5:16 and no print. The applicatiion lines look like more than enough TIM, but notice it is not applied in center of IHS, so not over the CPU ciip .. so of course it will not transfer heat as well.


There is so much out there in cyberspace that it is near impossible to sort truthful and accurate data from what is not. It is very hard for someone without knowledge to determine what is true and what isn't. This is compounded by the fact search engines bring up results based on words with sites having the most hits a site gets on a page/video/artical, not on how accurate the information is. In other words the more popular a site / page is, the more hits it gets, the closer to the top of a search list it will be.