Intel IHS removal... Any reason boiling water couldn't be used?

[celemine1Gig]

as far as an alternate method for exposing the cpu, could you use a circular sander disc with fairly high grit sand paper (1000) or so ... and sand down until your completely through the IHS and even a tad into the silicon of the cpu (to clean it of any solder)?

would that leave too much of the IHS around the sides and perhaps cause shorts?

With 1000 grit sandpaper you'd wait hours for it to sand through the whole IHS. I'd say that this method doesn't really sound too good. Better get rid of the whole IHS at once. But that's just my .

[alexio]

Nice work

I have a hotplate/magnetic stirrer combi with 0-350C temperature control. Guess I found a good use for it now

[Supertim0r]

haha, crazy

[Qkjhfhaiguihfma]

since this seems to be an active thread, i'll post my question here. i removed my ihs but didn't get the results i wanted, still a very large 10C+ difference between cores. is it possible the core itself is warped? the as5 footprint seems like it's touching all over, but the pressure isn't even i suppose.

[s1rrah]

With 1000 grit sandpaper you'd wait hours for it to sand through the whole IHS. I'd say that this method doesn't really sound too good. Better get rid of the whole IHS at once. But that's just my .

i mean while using a high powered circular sander that i'd apply the IHS to. perhaps would have to move back to 800 or something ... but using a powered disc, the work could go fairly quick.

seems like there would be too much scrap metal left hanging on the sides, though ...

[Methylphenidate]

I hate to do this to you guys, but I do not have my idle/load numbers from before my experiment. I can guesstimate that I got about 5C off my load numbers and almost nothing off my idle numbers as read by core temp. Overall whether you're better off lapping or removing IHS is totally up to you. I honestly don't think this is that much riskier, and they both void your warranty. This gives slightly better results by a couple degrees (I'm using an APOGEE, under better water, the difference might be more pronounced), but the choice is up to each individual user. I'd love to see other people try this and report their results though. I have an X6800, but I don't know if I can remove the IHS on this one because it might need to go back where it came from.


It looks like this was moved to help and support... Weird...

[zerolise]

I would suggest using distilled water just to make sure no minerals dry on your chip to reduce possible damage. That's if you don't use the pot method.

If you are worrying about minerals, you would not use the pot - but what else is there to use? . By the way, you should use de-ionised water to start with (those they use in the fab), distilled water is not good enough.

[zerolise]

Even when the IHS is out, there is still significant thermal resistance between the core and the heat sink (or cooling block). Thinking extreme and being irrational, has anyone tried welding the cores (or the IHS) onto the heat sink (or the cooling block) to improve heat conduction, in this way, there is no need for any freaking thermal compound ... LOL.

[Methylphenidate]

I hope you mean solder, and not weld... You would still have to contend with the poor thermal conductivity of solder, which would likely be worse than thermal compound.

[celemine1Gig]

If you are worried about the two cores not being of the same height, thus causing "bad" contact to the heatsink, you could always do the following:

WARNING! THIS IS NOT RECOMMENDED, SO DO THIS AT YOUR OWN RISK!

You could take for example 1mm thick steel and fabricate a narrow frame, which runs along the edges of the CPU (not the cores, but the green package ).
Now if you lap the whole CPU with the frame in place, you'll get both cores down to nearly exactly the same height and flatness. The hand lapping will lead to some rounding on the edges of the steel frame, but it's OK that way. This ensures that the middle, where the cores are located is slightly higher than the edges, hence giving perfect contact between the cores and the heatsink, instead of only having the whole pressure on the edges.

Just one idea. There sure are other ways. It's only how I would try to improve contact, as you don't seem to be happy with what you got ATM.

[bazx]

lol you crazy soab

[zerolise]

I hope you mean solder, and not weld... You would still have to contend with the poor thermal conductivity of solder, which would likely be worse than thermal compound.

No. I really meant weld... it produces the best thermal conductivity, but I reckon that this is a crazy idea, unless the heatsink (or water block) is part of the wafer substrate to begin with - that can only happen (if ever) in a fab.

By the way, if soldered properly, a joint would product good thermal conductivity too. 65% Sn and 35% Pb is all metal and definitely has lower thermal resistance than thermal compound, which is either silicon based or at best silicon loaded with a small amount of silver (and sold for lots of $$$)

[celemine1Gig]

lol you crazy soab

I'd prefer creative soab.

[Grinch]

wow...waiting to see some results...

[Methylphenidate]

Alright I've got some results for you guys, but I don't have anything but approximate numbers from before removal, so hopefully someone with an APOGEE and a similar water loop can give us some load numbers with IHS

CPU 1.512V as read by eVGA bios on 680i board. Supposedly it's one of the more accurate boards out there in this regard

core temp reads cores 1&2 as 75C and cores 3&4 as 71C

nVidia monitor reads CPU temp as 62C

CPU is clocked at 12x277 for 3.33GHz.

[ziddey]

No. I really meant weld... it produces the best thermal conductivity, but I reckon that this is a crazy idea, unless the heatsink (or water block) is part of the wafer substrate to begin with - that can only happen (if ever) in a fab.

By the way, if soldered properly, a joint would product good thermal conductivity too. 65% Sn and 35% Pb is all metal and definitely has lower thermal resistance than thermal compound, which is either silicon based or at best silicon loaded with a small amount of silver (and sold for lots of $$$)

lol I think the point being made is the temps of welding!!

[Methylphenidate]

I think part of what's making my chip run so hot is the fact that it's a relatively early engineering sample. I have an ES Q6600 coming that I might have been able to get better results on, except that my board is worthless for Kentsfield FSB overclocking (680i reference)

[BulldogPO]

Well I might do that on my E6600

[vudoodoodoo]

Nice thread. Neat guide here.

[Tuppen^^]

If i'm gonna remover a ihs, then i wil use that guide..

[pumbertot]

I hate to do this to you guys, but I do not have my idle/load numbers from before my experiment. I can guesstimate that I got about 5C off my load numbers and almost nothing off my idle numbers as read by core temp. Overall whether you're better off lapping or removing IHS is totally up to you. I honestly don't think this is that much riskier, and they both void your warranty. This gives slightly better results by a couple degrees (I'm using an APOGEE, under better water, the difference might be more pronounced), but the choice is up to each individual user. I'd love to see other people try this and report their results though. I have an X6800, but I don't know if I can remove the IHS on this one because it might need to go back where it came from.


It looks like this was moved to help and support... Weird...

no this is better than lapping and no voided warranty as with lapping when you stick IHS back on with epoxy. they will never check it. I say well done and also say big thumbs down to all who said blah blah you were warned etc. if someone wants to try delidding cpus etc then they should only be congratulated not put down by those of you who wont risk it. dont like it, dont do it but dont post negative remarks for those of us with balls.
my 10 cents.

and btw B1s are hot steppings like B0s. B3 ES and retail usually run a little cooler.

[the highlander]

Its easy to say things like that after all goes well... After you've seen the cat's balls, saying that he is male... its an easy thing.

[pumbertot]

Its easy to say things like that after all goes well... After you've seen the cat's balls, saying that he is male... its an easy thing.

nope. i delidded my 8800gtx without fear. then killed it for another reason, ramsink shorting components. only reason im not doing the qx6700 is becuase the TT wont reach the cores and dont really have stuff to mod it. might delid it once my prommy is regassed and ready though.

[Roger_D25]

and btw B1s are hot steppings like B0s. B3 ES and retail usually run a little cooler.

Do you have any info on B2 steppings, like how they compare temperature wise to B0's, B1's, and B3's? Thanks for your help!

[SharkerSk]

Could you post some photos .... with naked cpu in mobo ... something like this