Thermal paste removal

Das_Saunamies · Post by **Das_Saunamies** » Sat Jun 16, 2007 1:52 pm

No wonder I had a hard time finding some...

I've been staying away from xylene after an incident I had with insufficient protection(rubber gloves, xylene paint solvent or some such, you can guess the rest)... but seems like it'll clean anything and everything. Thanks for the tip, got some black patches that I could do away with...

~El~Jefe~ · Post by **~El~Jefe~** » Sat Jun 16, 2007 6:29 pm

xylene is carcinogenic

~El~Jefe~ · Post by **~El~Jefe~** » Sat Jun 16, 2007 6:33 pm

Das_Saunamies wrote:"EipÃ¤ mitiÃ¤", rigid plastic cards are the best. The newer the better, as the edges will be sharper. Swapping chipset heatsinks is almost mandatory now, the quality of the factory heatsinks and their installation is just appaling. The chips can take a good deal of thermal abuse though, although I'm not sure if that bodes well for component life.

You should only apply the paste on one surface, preferrably the chip: this ensures you can observe the quantity better and it won't spill or drip onto non-contact surfaces.

I recommend cleaning chips and heatsinks regardless of packing status. It's not like they're oiled, but there is occasional residue here and there. It's not a big task and guarantees peace of mind.

I totally degree. those malaysian 10 year old kids with one hand missing (from being late 4 mins to work a year ago) normally do not put as much enthusiastic care gooping your north or southbridge as a spcr person might. Even a top of the line board is done without much adherance to full coating/high quality gooping standards. A simple alcohol scrub and A.S.5 re-gooping drops temps 4-5 degrees C I have noticed! slap on a zalman or another big Don King from an aftermarket company and like, blam, temps fall even further.

stevenkelby · Post by **stevenkelby** » Sat Jun 16, 2007 6:37 pm

How do you guys get the right thickness of AS5?

If the HS base is perfectly flat, do you just wipe some off then gently scrape it off, or scrape out a very thin layer?

~El~Jefe~ · Post by **~El~Jefe~** » Sat Jun 16, 2007 9:55 pm

perfectly flat just need a solid grey but really thin coating. sheer is bad as when you move the heatsink just abit, it pulls it into more areas vs others.

I have not found, EVER, a definitive explanation of how much goop. I think you cannot put too much so long as it doesnt ooze out and so long as you push it down real hard to even it out. Too little is ultra bad.

anyone ever try Ceramique? I always wondered how it was compared to AS5

stevenkelby · Post by **stevenkelby** » Sat Jun 16, 2007 10:01 pm

Thanks for that. I'll just put a thin layer on the NB. I read a guide that says to put a line down the center of the C2D core, and let it spread itself. Does that sound right or would you spread it out?

RaptorZX3 · Post by **RaptorZX3** » Sat Jun 16, 2007 10:47 pm

as i said, use the Zalman ZM-STG1, it's not thick and you can spread it evenly on the CPU or North Bridge with the included brush. And i heard this one is better or equal to the AS5

stevenkelby · Post by **stevenkelby** » Sat Jun 16, 2007 10:51 pm

Yeah but I already have AS5 and it is better than the Zalman.

http://www.arcticsilver.com/arctic_silv ... hesive.htm Greater than 7.5 W/mK

http://www.stonebridgecomputing.com.au/ ... 499d188ebd 4W/mK

RaptorZX3 · Post by **RaptorZX3** » Sat Jun 16, 2007 10:58 pm

i just hate the fact that the AS5 is thick and i don't know how much to put on it so it won't overflow. the ZM-STG1 is not thick and i can spread it evenly. i wish the AS5 would be like that

stevenkelby · Post by **stevenkelby** » Sat Jun 16, 2007 11:05 pm

Yeah I see what you mean.

I just realised I linked to the wrong product above, the glue, not AS5.

The Arctic page for AS5 doesn't give a figure in W/mK but I did read it somewhere, on the net or in a mag. I will try to find it. It was higher than 5 anyway.

I guess that getting a good application may be more important than which product you use anyway.

Das_Saunamies · Post by **Das_Saunamies** » Sun Jun 17, 2007 3:44 am

I'm not sure if they actually use child labour for board assembly, but it's obvious that costs and corners have been cut. Also, there is no gooping per se, but rather it's usually an industry-standard phase sticker or thermal pad in place.

Don't stress about the paste, it's not a life-or-death issue. Make sure there's an even layer on the chip and that it covers the markings. The markings are a good indicator as they are grooves that need filling: if you fill the big grooves, you'll have covered the small ones too.

To prepare, wipe chip and heatsink contact surfaces clean. Then start with a 1-2mm wide diagonal line of paste across the chip(I find the width my Arctic Silver 1 tube gives me adequate). Then spread the sides of the line towards opposing corners evenly with a sharp object(razor, credit card). Make layer as thin as you can, Arctic Silver should be able to cover the entire chip in one go. If the final layer is lacking in some areas, put a little drop on said area and smooth the entire surface out again. Lastly, apply heatsink evenly: no wiggle, no brutal force, just push it flat down and lock it in place. It's all really, truly simple.

- It's much easier to apply more than take some away. Start with as little as you can get away with, then add as needed until area is covered.
- The less paste the better the conductivity when surfaces are flat, but the less room for error in heatsink installation.
- The more paste the worse the conductivity when surfaces are flat, but heatsink installation is less prone to errors.
- An even spread is more important than exactly the right amount, so just get the basics right and you'll be fine.

I've never had a problem with Arctic Silver spilling or leaking. It's just the right viscosity for tweaking with small amounts.

stevenkelby · Post by **stevenkelby** » Sun Jun 17, 2007 4:14 am

Thanks for that great advice, I'm confident about it now!

~El~Jefe~ · Post by **~El~Jefe~** » Sun Jun 17, 2007 9:25 am

I just read about how as5 is supposed to be applied. 1 1/2 sizes of a piece of uncooked rice.

thats super small. You will possibly fry your chip doing that. there is a great website about people doing a spotty job of applying goop and how the chip is burned out. a total thin layer has always worked for me. the twist effect is interesting in eliminating air bubbles. next time I am going to try that.

Also, having read a lot about Ceramique, I think, without a doubt, it is the better substance. I always wondered why on earth it was sold as as5 is the preferred substance. Now it seems that ceramique is easier to apply which means that you do it correctly vs. not knowing if it is on right, it cleans off easier which is important, cures in 25 hours vs. 200, is completely non conductive and non capacitive vs as5, and is not a "grease" like as5 which voids warranties.

Interesting, right! it also is like 1/2 the price it seems. Not that I care about price at all, but it makes as5 look a bit less attractive. I hardly ever turn off my pc. I run a telnet/mud on it and other BBS stuff. I probably have not cured my goop properly. I could easily Ceramique my stuff, run stalker for a few hours, turn it off, let the watercooler run to cold, do that again, 2-3 times in the course of 2 days and I would be done with the process that is very important.

hmpf!

speedlever · Post by **speedlever** » Wed Jun 20, 2007 6:34 am

I used Ceramique on a couple of recent C2D builds. I'm happy with it. But read here before assuming that Ceramique is definitely better than AS/5.

Das_Saunamies · Post by **Das_Saunamies** » Wed Jun 20, 2007 12:32 pm

Thanks for the post speedlever, interesting read from an expert for once. I'm betting my chips on good old Silver, so to speak.

bkh · Post by **bkh** » Wed Jun 20, 2007 4:36 pm

>I just read about how as5 is supposed to be applied. 1 1/2 sizes of a piece of uncooked rice.
It depends on the chip type. The Arctic Silver web site has detailed instructions for several kinds of chips, and the instructions differ depending on the chip type. For an old AMD X2 it was the "grain of rice" blob in the middle. For a C2D it was one thin line in a certain direction. The goal is to get a thin layer with no air over the core --- the rest of the surface of the chip cover doesn't particularly matter.
With Arctic Silver 5 there is an element of art or skill. Too little and the core will run way too hot. Too much and the core will run somewhat too hot.
If your core temperature is below 60C under full load and with a satisfactory fan speed, then you did ok. If it is above 60, you may wish to take the heat sink off, clean off the old goop, and try again with more or less than you used last time.
Arctic Silver 3 was thinner and so a bit more forgiving, but it wasn't quite as good as a *well-applied* layer of AS5....

~El~Jefe~ · Post by **~El~Jefe~** » Wed Jun 20, 2007 6:12 pm

speedlever wrote:I used Ceramique on a couple of recent C2D builds. I'm happy with it. But read here before assuming that Ceramique is definitely better than AS/5.

eh.... the properties of ceramique truly sound that they are more foolproof than as5. I never thought I would switch but I am going to on my builds. As5 is good, but consistency is desired more than deciding which gives a 1 degree difference. Improperly applied goop that is still applied safely/adequately I have found to give 4-5 degree difference. I really hate to have to apply the goop more than once. I mean, i lock up my case, hope it works then run my mud and go to bed. I can't stand having to break the goop seal and reapply to get a lower temp.

speedlever · Post by **speedlever** » Wed Jun 20, 2007 6:37 pm

I tend to agree with you. Actually, it's more involved than just removing and replacing the goop when you consider the mobo likely has to come out of the case. At least, it would for me with the tower HSF I use.

Come to think of it, even with the stock HSF, I would want to remove the mobo to replace the goop with those darn clips... or whatever they're called that secure the HSF to the mobo. Unless I put some kind of support under the 4 snap locks for reinstallation.

~El~Jefe~ · Post by **~El~Jefe~** » Wed Jun 20, 2007 9:14 pm

lol yeah, alot of steps.

the worst part is when the standoff brass things strip the motherboard walls then spin with minimal effort. im really delicate but it happens

Bluefront · Post by **Bluefront** » Thu Jun 21, 2007 2:56 am

Over the years I read countless methods to apply TIM.....some sound as difficult as brain surgery. And the TIM itself.....anything from tooth paste to a mercury solution. So which method/TIM is the best? Good question.....no answer.

You don't need any TIM......if the surfaces are flat, 100% flat, and the attachment method assures even pressure across the contact area. No amount of TIM will make up for two surfaces that hit each other at an angle. My ideal heatsink attachment method involves placing a heatsink straight down on a CPU, holding even pressure on the top of the heatsink, and tightening down four spring-loaded screws on the corners evenly, like a turn at a time in a criss-cross manner till a pre-set tension is reached, and the screws bottom out.

And a back plate is a good idea. It prevents any flexing of the MB as the screws are tightened. Some boards flex quite a bit when a heatsink is tightened down.

That presumably ideal situation is almost never possible, and particularly so with large heatsinks that make the four-screw method almost impossible, and rarely used any longer. Some attachment methods are so bad they require you to latch two hooks on one side, and push down on the other side till you can get the other two hooks to attach. An XP-120 is like this. It is so difficult, no matter how careful you are, when finished with the attachment, your shiny new heatsink....will be scratched.

So what does the TIM do anyway? It only fills in slight air pockets between two surfaces. That's it....no amount of the stuff will make up for a poor attachment method. The recent reported Ninja problems are a good example of this. Is there a solution here....will a hi-priced TIM make up for a poor attachment? Probably not.....

IMHO.....spend your money on a heatsink with a good attachment method, maybe read up about other peoples experiences with that heatsink. The TIM itself is the least important part of heat transfer. If two flat metal surfaces touch each other correctly.....you don't even need TIM.

stevenkelby · Post by **stevenkelby** » Thu Jun 21, 2007 3:57 am

I just finished lapping my Ultra120Extreme and C2D. I wasn't going to but when I went to remount my little Zalman HS for the NB, I saw that only about a quarter of the AS5 had been touched, the Zalman wasn't flat. I checked it with a razor, 2 big ridges and a depression.

I lapped it with glass and 3 grades of wet paper, it was very unflat! I checked the Ultra120Extreme and C2D heatspreader as well and they were also very far from flat so I did them too. Took a couple hours but I took my time and did the best I could. Very happy with the result but wish I took before and after pics with the razor and light shining through the spaces.

Now I just have to apply the paste. Not sure which way is best so will try a line as per the Arctic site, press down then check. I will try with a thin layer too, and compare then go with the way I think is best for the final time.

speedlever · Post by **speedlever** » Thu Jun 21, 2007 4:18 am

You should be good to go. I think it's preferable to put on too little TIM than too much.

I suggest you use the TIM sparingly. Refer to Bluefront's post.

Post back when done and let us know what temps you're seeing.

stevenkelby · Post by **stevenkelby** » Thu Jun 21, 2007 4:26 am

OK I just tested and it looks like I can get a better result by spreading a thin layer with a razor. I actually used the back edge of a surgical scalpel, thin, long and straight.

The line left a lot of compound between the HS and CPU. I hear that it's best to have as little compound as possible between them.

The thin layer was thin enough that you can see the metal through it when removed, much less compund here than with the line. As both surfaces are as flat as possible there can be no spots where they don't touch. Thanks to all the advice I read here!

I am going to go with spreading a thin layer.

Also, the Ultra 120 extreme base was very curved in one direction, but straight in the other, like a section of a cylinder, not a shpere. It is curved in the way that means the C2D chip would probably touch all the way along if the HS is mounted on the mobo horizontially, but if the HS is mounted vertically, as I will, so the hot air blows out the back of the case, then the proccesor is only touching in it's middle, according to the pics on this page:

http://www.arcticsilver.com/pdf/appinst ... l_wcap.pdf

stevenkelby · Post by **stevenkelby** » Thu Jun 21, 2007 4:28 am

Posted at the same time Speedlever. Everyones posts have been very helpful, thanks.

I won't be up and running until the weekend at best (waiting for RAM and BigNg, also, 1st, build) but will report back with temps when I can.

stevenkelby · Post by **stevenkelby** » Thu Jun 21, 2007 4:31 am

Just a note, that Arcticlean stuff is amazing. The cleaner dissolves the grease instantly. Glad it was delivered today.

Das_Saunamies · Post by **Das_Saunamies** » Thu Jun 21, 2007 8:43 am

Good to see you got it in the first go(s), congrats!

I just had to redo mine. The Arctic Silver(the original!) had touched nicely across the A64 with an unlapped Zalman 7700B-AlCu, but there was so little that there were two vertical lines next to the core where it had dried out and stuck to the cooler. Still, temps below 50-55 at load so no worries there. "Too little" is just right, although I do tend to touch up the whole mating interface, not just the core part of it.

Forgot to mention that one definite sign of the right amount is a little pulling test that you can do when you place the cooler down: if it doesn't want to come off but starts to rotate, you've got it right(suction from squeezed paste). If it comes off easily, there's too much or too little(the paste tears up or doesn't seal enough ground). If it rotates and then pops off, there was a bubble. I've experienced the first and the third cases, but the second one is just logic.

I borrowed some AS5 from a mate once, but prefer my own, seemingly everlasting tube of the original stuff.

PS. A surgeon's blade you say...