Why does my paste keep failing on my r9 290

[JasonM]

I find all the so called higher end TIM's hard to work with as they are very dry coming out tube but no doubt that is how they can preform better IF you can apply it well.

A new tube of IC Diamond it's not hard at all, it's more like the composition of butter at room temperature. IC Diamond however does start to harden once the tube is opened, it should still last 1-2 years once the plastic tube seal is re-fitted however.

I do have some AS5 thats stored in same location of the ICD, AS5 appears to last many many years in tube however.

Question on AS5, how do you find the durability of AS5 over long term once it's applied? Also what is best method of applying AS5 to CPU/GPU?

 

[b00merang69]

I have never had to change over AS5 (Mobo Chipset's/CPU/GPU) but for storing long term they do say "tip down" and ideally in a fridge and after long time storage due to separation a little oil comes out which i discard.

Spreading is the best method and most consistent if you are good at it, the Pea method IMO was brought about to let nearly anyone do it but again its down to the size of said pea and a bit of luck as I have never got the same results twice with that method.

 

[JasonM]

I have never had to change over AS5 (Mobo Chipset's/CPU/GPU) but for storing long term they do say "tip down" and ideally in a fridge and after long time storage due to separation a little oil comes out which i discard.

Spreading is the best method and most consistent if you are good at it, the Pea method IMO was brought about to let nearly anyone do it but again its down to the size of said pea and a bit of luck as I have never got the same results twice with that method.

Thanks for info, I might try it one day, would be something that did not matter to much at least to try at first.

Going back to IC Diamond, it will be difficult for me not to use that paste. To give idea of performance, IC Diamond once took 12c from a graphics card, that was compared to the poor Nvidia paste but even so.

 

[b00merang69]

Stock TIM (very hard and dry) needs to be there forever so its not as good as aftermarket TIM that may well need changed out every so often (drys out or pump out) though some claim IC Diamond does not seem to.

 

[Phil2008]

I always use the spread technique with my finger in a food bag, it works really nicely and then your 100% sure all of the cpu/gpu is covered, plus you dont get any oozing off the cpu/gpu if spread thinly enough.

 

[Armageus]

I always use the spread technique with my finger in a food bag, it works really nicely and then your 100% sure all of the cpu/gpu is covered, plus you dont get any oozing off the cpu/gpu if spread thinly enough.

Which if you have inadequate clamping pressure for any reason, you won't know but will be spreading it too thin. A blob would at least spread a minimum amount and still make decent contact.


EDIT:
Realistically there isn't a huge science to thermal paste and most people are overthinking it. If even toothpaste can work as thermal paste, with little thermal difference, then anything can.
Between different pastes or application methods the results are often within a couple of degree (or the margin of error).

 

[Phil2008]

Which if you have inadequate clamping pressure for any reason, you won't know but will be spreading it too thin. A blob would at least spread a minimum amount and still make decent contact.

I have put the paste on both the gpu and heatsink, so double the thickness. So I would have thought it would have evened it out if the heatsink wasn't making good contact with the gpu?

 

[Rroff]

Which if you have inadequate clamping pressure for any reason, you won't know but will be spreading it too thin. A blob would at least spread a minimum amount and still make decent contact.


EDIT:
Realistically there isn't a huge science to thermal paste and most people are overthinking it. If even toothpaste can work as thermal paste, with little thermal difference, then anything can.
Between different pastes or application methods the results are often within a couple of degree (or the margin of error).

IIRC in a couple of scenarios Gamers Nexus found hand spreading gave very poor results while in several cases going for a huge blob gave the best results by sometimes as much as 6-7C if the other methods were inadequate (i.e. over a big area like Threadripper).

When I'm back from work will try and find the relevant videos.

 

[mmj_uk]

I spreed a thin layer so it doesnt ooze(as you need to be careful with AS5) but thick enough so you cant see the gpu surface. Basically the same way as I do my cpu's and the paste lasts the life time of the cpu normally

I'm probably about the tenth person to say this since I'm late to the thread and can't be bothered to read but you need to use the blob method that way air gets pushed out the sides when you apply the heatsink, when you spread and then put the heatsink down it creates pockets of air which expand when heated (hence performance gets gradually worse). I used to spread all of the time and had the same problem it's basically bad application. It's better to have excess TIM hanging over the sides (providing that it's non conductive) than to have air pockets between core and heatsink but even so if you judge the amount needed right there shouldn't be too much excess anyway.

 

[b00merang69]

That may sound right and makes sense but it is not the case if you can spread well and this can be seen by better temps for many including myself and when you remove the cooler and see a perfect impression with no little missing dots.

In that case the manufactures would tell you to put a big drop of Liquid Silver in the middle of IHS instead of spreading it with supplied Q-Tip (not taking shorting into scenario as you would have painted over those parts to insulate them).

 

[Rroff]

That may sound right and makes sense but it is not the case if you can spread well and this can be seen by better temps for many including myself and when you remove the cooler and see a perfect impression with no little missing dots.

In that case the manufactures would tell you to put a big drop of Liquid Silver in the middle of IHS instead of spreading it with supplied Q-Tip (not taking shorting into scenario as you would have painted over those parts to insulate them).

There are pastes that work better/worse with hand spreading a lot of older pastes required hand spreading which seems to be where the practice came from but a lot of modern pastes don't require it or in the case of AS5 they reccomend other methods for most scenarios. Gamers Nexus has a load of in-depth videos which are worth watching.

 

[mmj_uk]

It doesn't matter how well you can spread it's basically impossible to put the heatsink down on a layer of TIM without pockets of air becoming trapped. I'm not sure about Liquid metal it's a different composition (extremely fluid when applied) and highly conductive right? trapped air might not affect it once it's set so maybe they weighed it up and decided spreading was less risky.

Lots of videos on Youtube showing how the air pockets form when using spread method:

Then over time those air pockets further expand due to the air being heated.

 

[b00merang69]

I bed to differ and for me I can get far better results spreading and I will stick to that.

 

[gpuerrilla]

Much better than the factory, they seem to cake the stuff on and get it spilled over.

 

[JasonM]

While on the subject of applying paste, does anyone use the Arctic surface purifier?

One of the things I notice is even a brand new CPU and cooler is not 100% clean. The CPU coolers have plastic on the underside that leaves a mild trade of glue when removed.

So what i do is fit the CPU (into the motherboard), then I polish the CPU surface with the Arctic Pure, the metal starts to shine up more. I then do the same with the cooler surface. I do this immediately before applying the paste and then fitting the cooler, this is to minimise dust/air particles settling on the two surfaces.

Now unsure if this additional step of using the purifier makes any real difference, but the surfaces are more reflective before applying the paste.

And there's more... this is my method for fitting a CPU cooler.

1) The motherboard fitted in case, and case flat (board faced up). I then screw the cooler down side by side in other to spread tension.

2) I then move the tower case on it's side, I then check tension again, checking the top screw first as this side is more under pressure from gravity.

3) Then I turn the computer on, and let it run for maybe 20-25 mins under no load. This lets any initial expansion happen, but more important lets the paste warm up.

4) Then I re torque the CPU cooler screws, often you can tighten a very slight amount more, and final stage is compacting the thermal paste more and giving better spread.

5) The final thing I do, or more don't do, I won't load the CPU for at least 24 hours. I will install the OS, but not excess heat. For at least 24 hours I let the paste start to cure, then I will put load into the CPU. Again unsure if loading a CPU after paste is recently applied can have a negative effect, however I don't do this. New computers I take them easy for at least 24 hours.

Does anyone else use the two elaborate processes i'm doing above?

 

[Armageus]

Does anyone else use the two elaborate processes i'm doing above?

Nope life is too short.

Blob of paste and move on

 

[JasonM]

Nope life is too short.

Blob of paste and move on

Your probably correct, but what are your thoughts on this video.

https://youtu.be/ZBumREI0PEg?t=163

He's saying paste won't spread correct without it being heated up. As mentioned I already re torque the CPU cooler after computer has been initially on, and I can always tighten the screws a very little more (maybe 1/3 of a turn) compared to when the cooler is initially fitted.

 

[Phil2008]

I'm probably about the tenth person to say this since I'm late to the thread and can't be bothered to read but you need to use the blob method that way air gets pushed out the sides when you apply the heatsink, when you spread and then put the heatsink down it creates pockets of air which expand when heated (hence performance gets gradually worse). I used to spread all of the time and had the same problem it's basically bad application. It's better to have excess TIM hanging over the sides (providing that it's non conductive) than to have air pockets between core and heatsink but even so if you judge the amount needed right there shouldn't be too much excess anyway.

It doesn't explain why I get good temps for a few hrs after I have applied the paste, then the temps go down hill....Im thinking about buying a different cooler for the 290?

 

[Len Dananey]

It makes zero difference how paste is applied, aslong as the chip is covered enough and the paste is good. Plenty of tests that prove that time and time again.

 

[Rroff]

It makes zero difference how paste is applied, aslong as the chip is covered enough and the paste is good. Plenty of tests that prove that time and time again.

That is the key thing - some application methods will provide better coverage or better long term stability than others.