Applying MX-2 on an exposed die.

[daehxxiD]

Hey everyone;

I'm at it again replacing the TIM on all my devices. First I used the Liquid Metal Pads instead of the pre-installed paste and needless to say, it performed a lot better; but I've had a friends PS3 die because a bit of the pad dropped on the Mobo causing a short (after well 6 months of usage! Burn-in should've happened months ago when we brought his ps3 to turn itself off because of the heat)...

Long story short, I'm striving to change the TIM on all my devices now, done it on PS3 (with the positive result of having even better cooling with MX-2 [well I even lapped the heatsink of that beast]), now I want to move on to my PC and 360. As you may see from the systemspecs my PC is a laptop; it's been running the pads fine for 1 year now, but the expectation of maybe even better cooling through MX-2 and and a safer system drives me to open this baby yet again... One thing is still bothering me though; how would I apply this stuff?

I know that with AS5 you spread it yourself using a creditcard until it's nice and flat on the die, but for some reason the spreading with a creditcard doesn't work to well for MX-2; I had a much better result with the "blob"-application method on PS3 anyways (mind you it's Chips have IHS'). So what do I do on an exposed core? Should I still try to spread it myself, or would it work the same way as with an IHS and a (smaller) blob, squished and rotatet with the heastink would do the job?

Please let me know, preferably from experience rather than "common knowledge", as although it's not rocket sience, it is still pretty amazing how many diffrent and often totally wrong "rules of thumb" circulate the web about paste application.

P.S.: I still have some AS5 left too, if you believe this is better suited for direct on die application; just let me know.

Thanks for the help.

 

[Solaris17]

MX-2 all the way it has better thermal transfer properties..now for spreading what you can do is drop a small blob on it and spread it while having a ziplock or sandswitch bag over your hand...alternatively you can squirt a little bit (about a rice grain and a half) on the core and do a test mount to spread it out.

 

[dark2099]

MX-2 all the way it has better thermal transfer properties..now for spreading what you can do is drop a small blob on it and spread it while having a ziplock or sandswitch bag over your hand...alternatively you can squirt a little bit (about a rice grain and a half) on the core and do a test mount to spread it out.

I do small dabs in each corner of the cpu, and then like half a grain of rice in the middle, maybe 2/3's of a grain, but that usually does well enough.

 

[Solaris17]

I do small dabs in each corner of the cpu, and then like half a grain of rice in the middle, maybe 2/3's of a grain, but that usually does well enough.

sounds good basically you just want to put a good enough amount were it spreads nice and doesnt leave any bald spots....their are usually "rules of thumb" about pea size and rice size etc..but in all honesty just apply until it looks good surface smoothness of the core and HS matter alot and contribute to how much TIM to use and this can very significantly so basically you want about 2 sheets of paper thickness of TIM on your core.

 

[rangerone766]

i've tried most of the compounds out there, and they all seem to work best using this method.

http://www.innovationcooling.com/applicationisnstructions.htm

also, the ic diamond has beat out mx2 by 1-2c for me. altho the diamond is pretty expensive.

 

[newtekie1]

I always use the small blob, even on exposed die CPUs. Has always worked in the past.

 

[daehxxiD]

Hey,

thanks for the super fast answer guys.

@Solaris: I will try that with the test-spreading by putting a blob on the die and have the heatsink do the job;
the thing is, through testing I noticed that MX-2 has the tendency to strech itself out a bit through several thermal cycles resulting in a perfectly even and very thin layer of TIM (giving another 1-2°C of extra cooling), whereas it still retains it's "blobby" ( ) form in the center of the IHS on the "cold"-test.
Also, it spread alot better when I lapped my heatsink on the ps3 than with the rough surface.
Alternatively I will use your advice and do it with the plastic bag on my finger, I guess it is true then, that the creditcard doesn't really work for MX-2.

@dark2099: Yeah, that is exactly what I did on the RSX for the PS3 (it has it's 4 on-chip Ram chips on the corners of chip and the RSX-die right in the middle); however right now I'm looking for methods of application on an exposed die, i.e. no Integreated Heatspreader.
Help still appreciated though.

Edit: The instructions of the diamond page are awesome, they pretty much mirror what I have observed up until now; I've read about the compound and read that you can even use it to give your lapping the final shining touch; will definatly have to try this one out once; but for now MX-2 will be good enough for me, got this 30g tube lying around and feel like I have to get my moneys worth out of it

 

[WarEagleAU]

I would think CoolIT Laboratories Pro liquid pads and metal compound would be better than anything, including the diamond. Very odd.

Any plans on writing a Guide on how to change (and lap ) the TIM on the PS3?????? I would be grateful (even though I could lap the sucker )

 

[daehxxiD]

I would think CoolIT Laboratories Pro liquid pads and metal compound would be better than anything, including the diamond. Very odd.

Any plans on writing a Guide on how to change (and lap ) the TIM on the PS3?????? I would be grateful (even though I could lap the sucker )

I might just write a guide; X-mas holidays coming up, so I might have a lot of free time on my hands; gotta fix up another PS3 (60GB US-Version) anyhow, so I might just take a few pictures of the process and write a tutorial to it. I really believe it pays off, as the Heatsink in my PS3 was so rough and concave, it did not even make contact with the center of the IHS; leaving the whole thermal transfer to the TIM...

 

[Solaris17]

I would think CoolIT Laboratories Pro liquid pads and metal compound would be better than anything, including the diamond. Very odd.

Any plans on writing a Guide on how to change (and lap ) the TIM on the PS3?????? I would be grateful (even though I could lap the sucker )

it is but it destroys AL

 

[WarEagleAU]

But if you have a copper Core it should be alright then right? I thought about one point in time trying the pads (or even the liquid) instead of my MX-2. However, Im a scaredy cat because if the liquid drops onto the board, its pretty much a goner.

 

[daehxxiD]

it is but it destroys AL

Thankfully the Pads don't, but I don't think they are as effective as the liquid metal. The pads really stay there even after melting; there is no more direct contact between HS and CPU/GPU; while it may yield better cooling on most systems because pre-applied past did basically the same but performed much worse, it doesn't come close to a properly applied TIM and a nicely lapped HS.

After all it "only" has 84W/mK (AS5 has something like 9W/mk) or something like that of thermal transfer, whilst Aluminium has around 200W/mK and Copper something like 300W/mK if I'm not totally mistaken.

Edit: @WarEagleAU: PS3 is using Aluminium Heatsinks, so better not try using the Liquid Metal "Paste" on that; as I said the Pads won't cause any harm unless the drop onto the board as it happened to my friend (strangely enough). Personally I'm rather investing a bit more time in preparing the HS and TIM for optimal transfer than going with the Metal Pads, as judging from experience, they really aren't magic, other than easy to apply.

 

[WarEagleAU]

Gotcha. Ill just wait and see if you do a "Guide" Thanks Dixx and Sol!

 

[daehxxiD]

Ok, I replaced the paste now and may I just say one thing?

I love you all guys, thanks for your help! It's just wow. CPU Load temps went from 63°C (I repeat sixty-three degrees Celsius) to a whoopin 55°C (fifty-five degree Celsius); if that is not an improvement I'll shit bricks.

Go, MX-2, go. (btw. I checked and the Liquid metal pads were placed correctly and melted properly, but as I presmumed there was no direct contact between die and heatsink because the LMPs did not move but actually stayed there [thank god, or my laptop would have been a goner ).

Edit: oh and I used the blob method; although some of the paste always squirted out of the sides of the die (couldn't really get less out of the 30g tube), it seems to work just fine, finer than expected. Thanks again.

Oh and GPU idles away at 26°C, first it was 31°C.

 

[Solaris17]

awsomme best part is when the mx-2 settles it should drop a little more.