Repaste issues

I recently ordered some arctic cooling mx-4 thermal paste for my desktop and had a bunch left over. So I decided to repaste my 6970m and my temps are on average 5*C higher.... So I figured I did it wrong and repasted it 2 more times. Same result using different methods. I tried the pea method, the credit card method, and the saran wrap on the finger method. What am I doing wrong?

 

Isn't there a significant settling time associated with the TIM?

 

Yes, if I remember right with Artic Silver you need to give it a week or so to cure before you'll see your optimal temps.

 

MX-4 is marketed as not having any curing time though. You might be thinking of the similar Arctic Silver.

To OP: It sounds like you did the correct application method (spreading it on)

 

So is IC Diamond really just that amazing? This MX-4 is supposed to be one of the best on the market, but this temp difference is crazy... Close to 8*C temp difference under heavy load...

 

I'm not sure, from what I've seen though- these "premium" pastes all do roughly the same in benchmarks. They tend to vary by only a degree or two in most tests. An 8C difference is unexpected, frankly.

Arctic Cooling MX-4 Thermal Compound - Testing and Results :: TweakTown USA Edition

 

This is a bit off topic, but does anyone know where I can buy some IC diamond? I live in Canada.

 

MX-4 supposedly does not require burn in time like Arctic Silver does.

 

I thought NCIX carried it but I can't find it on their site anymore. You can always get it from Hardware Canucks! Come joing the forums there if you haven't already.

Hardware Canucks - IC Diamond

Mega Computer.ca - IC Diamond

 

Did you have IC Diamond on there previously or was it stock thermal compound. If it was IC Diamond before that would be why you are seeing higher temps now.

 

I bought IC7 off Amazon(No prime though). Many people have seen temps fall 3-6c using IC7 compared to Artic Silver.

 

IC Diamond was there before. But a ton of it was applied. It was a mess actually.

 

I have ran some tests on my current laptop some time ago when I had temp problems and I found that spreading it on gives a higher temperature on average than with the dot method (in this case the chip is not square so you have to make a very small line so all the paste is evenly far from the edge of the chip).

this is, I think because spreading will go wrong fast as you can get a lot of stuff mixed in the paste by accident. also u will use 2 much 2 fast and you can create air bubbles.

use a creditcard (or similar) which is clean and dont use your finger when using spreading method (although I recommend dot/line method).


I have already done a lot of repastes so I can assure you that I know what I'm talking about.




edit: also dont forget to clean the chip and the coolingblock first! and dont use toilet paper as the fibers may make scratches on the chip and the cooling block instead use for example underwear which is 2 be thrown away or fiber cloths.

 

If it was ICD on before this is your reason, like someone from Xotic said. I've gotten the same drop repasting from a "stock" paste to ICD that you've seen as an increase leaving ICD.

If by a "mess" you mean it was pushed out the sides of where contact is made, this is a correct installation. They want to guarantee you get full, no-air bubble coverage, so a good amount is applied. When you put enough to reach all the corners of a square, and your paste is spreading like a circle, it pushes out.

If you are REALLY particular, paste the ICD down, note how much you squeezed, press it down, then remove the heatsink and check the spread. You'll then have to wipe off everything, and then apply your "more appropriate" amount that you want. I don't think though you'll see much difference in temperatures of a push out vs. "hardly any" ( I can't say "none", since like I said, by nature of a proper installation, it is going to push out, and look funny compared to our methods of old where you thinly applied TIM with a credit card and never had any push out).

Check out Innovation Cooling to see what I mean, under the application area.

 

Alright I pasted it once more using the IC method even though I'm using MX-4. No change in temps. But maybe I just need to run a very long burn in to make sure it spreads properly? Something I also noticed the first time I went to repaste; after I cleaned off all the IC Diamond I noticed many small scratches on the chip. I used a lintfree cloth and very gently removed the paste so I'm pretty sure it was not my doing. Maybe that's causing some air pockets and thus increasing my temps?

 

I have also researched on this as well. It has been officially responded to by Innovation Cooling that yes, the simple use and removal of their product will scratch the top of your chips.

However, even though their compound is comprised of diamond (hardest of all), other compounds are also abrasive to the tops of the chips also.

The micro-scratches you're referring to any way are one of the reasons that the pastes exist... fills 'em up and smooths 'em out, allowing for better contact.

At this point, if you're wanting to use MX-4 (I think it's design is not to have a break-in like IC Diamond) just stick with it, as simply applying and reapplying as much as you have been will wear away at the chip, no matter what compound you're using.

Otherwise, if you really want those degrees back, do the IC Diamond. In either case, see if you can get someone to post ICD temps of your GPU, in a test you come up with ( 15 minute run of FurMark Only ) that way you know for sure you want to go back to IC Diamond. (Again, just looking out for you not having to repaste again, if not needed).

I'm mid-post right now and forget, I think you had a 6970 or 90, so my GTX560 won't offer help, otherwise I'd be happy to.

 

I appreciate all the help. +1 I suppose I'll put in an order for some more IC Diamond paste.

 

Make sure you remove the black plastic sheet since it traps residue cleaners and paste.

And a 5C difference only means either you put the paste on wrong or cleaned wrong. An IC7 won't correct that.

 

Compared to stock, I got a 7 degree drop in Furmark with IC7.


I don't know though how IC7 compares to MX-4, specifically in his particular system, on this card... but he has repasted so many times and so many ways Rich, (Plus he originally had IC7 with his desired temperature) I think it's safe to say IC7.

If nothing else, it's worth a $10 try if he's curious. Not like he's scared of the process

 

Yeah no kidding. I put some on order yesterday. It may get here by Friday. *Fingers crossed*

 

Do you mean remove it, clean it and under it, and the replace it? Or can it just stay off?

 

Cool, (hopefully right -- pun intended) did you read anywhere what people were getting with ICD and the 6970m to compare before ordering?

I peeled mine up in the middle, made sure to clean the last bit of old paste out and just put it back down. Never unstuck it fully, it was on there tight and I wasn't sure. So, my temps dropped well, like they did while still leaving it attached. We should search around, I'm kinda curious as to the real purpose now that Rich has made the good point--- it could be "cool" to remove it entirely.