Removing the Prescott Heatspreader

[fafnir]

we have the method,

with thanks and gratitude to EmineM, for teaching us that PATIENCE is needed:


<before you start, you'll need both the chip and your room to be at room temperature, ~20-30 celcius, the procedure must not be performed on a still hot chip>

you'll need razor blades (the retro blade type, not safety razors, e.g. not mach3's), angled cutters, water and sandpaper

Originally posted by EmineM


1. cut the rubber seal around the heatspreader


2. using four razor blades, gently elevate the heatspreader;

a. start from one side, and GO SLOW
b. elevate the ~0.1mm by stacking another razor blade in between, and WAIT
c. wait 5 minutes and rotate 90 degrees and procede to the next corner
d. every time you wedge another razor blade in, the total resistence (of force going up/down) should be no more than 5lbs/inch^2
e. when all four sides have been lifted 0.5mm, you should be able to feel the bond break, DO NOT USE FORCE, DO THIS SLOW

p.s. you'll need ~15-20 razor blades for the job, and extreme time and patience


3. lap the remaining epoxy layer on top of the core clean with 800 grit wet sandpaper

a. procede to 1000 grit and then 1500 grit for the mirror shine, use no more than 5lbs of pressure or force

_

[Karnivore]

Leave it on.....


If I had to absolutely remove one I would remove it slowly as if lapping it, seems youv'e tried most reasonable possibilities...

[freecableguy]

I tried this when I first got my M0 stepping 2.4C....many, many months ago. Needless to say, I killed that chip before I had a chance to even test the OC. Oh, well. Anyway, these new IHS DO NOT come off. That's all I can tell you. You might want to go find out from the person that removed the IHS from this "naked" Prescott that you saw....

[(907)FAN-BOY]

I thought all new P4's had the IHS expoxied (theral epoxy) on.. so that'd mean it take the core, or a large chunk of it with it.

[charlie]

The only "correct way" to do it would be to put the cpu on a mill and machine off the IHS, doen to the core surface, then remove the perimeter of the remaining IHS with a razor blade

C

[(907)FAN-BOY]

Cause it's epoxied right (newer ones)? Or is that just the only easy way to do it?

[skate2snow]

I know how to remove it, and you have tried some harder stuff then what you had to.

Find the tinyest and more "solid" knife you can(it as to be very tiny and very solid...). and then, you that the corner that is opposing the "cutted corner", and you just take the knife blade between the IHS and actual CPU, and you do like a saw(Sorry, I dont know how to spell the manual wood cutter(?)), and after you got as deep as 0.5CM, you do the lever trick(it mulplies your actual force by about 10x), and it should come off.

BTW: The epoxy will resist to about -60C.

[Karnivore]

Originally posted by skate2snow
I know how to remove it, and you have tried some harder stuff then what you had to.

Find the tinyest and more "solid" knife you can(it as to be very tiny and very solid...). and then, you that the corner that is opposing the "cutted corner", and you just take the knife blade between the IHS and actual CPU, and you do like a saw(Sorry, I dont know how to spell the manual wood cutter(?)), and after you got as deep as 0.5CM, you do the lever trick(it mulplies your actual force by about 10x), and it should come off.

BTW: The epoxy will resist to about -60C.

Skate you've missed the point....

the CORE IS EPOXIED, so when one does what you suggest, the core shatters, cracks, chips, etc....

Originally posted by charlie
The only "correct way" to do it would be to put the cpu on a mill and machine off the IHS

Same thing as I suggested, only difference is your suggestion is done mechanicaly, and with less time...

[Lithan]

Strip naked and start playing with yourself, then compliment the prescott on what a "Hot little chip" it is and ask it to join you.

[freecableguy]

hey freecableguy, how did you try to remove your M0 heatspreader? you didn't just cut and then pry really hard did you?

and also, did your chip core come clean off the pcb or was there more carnage?

My M0 2.4C has now been asked to star in "Comando II"....YES, there was much carnage...poor little guy, He never knew it was coming (the chips shattered into MANY pieces).

[HKPolice]

I doubt even -60C would rip it off since they've done cascade runs at -110C? The IHS remained intact

Originally posted by fafnir
skate2snow, are you confident on the -60 celcius part?

i'm hitting -40 and its with all the dry ice i can get

so what can i get -60 celcius with?

[snowwie]

first off, why you concerned about the IHS?

the added resistance in thermal conductivity from the TIM (epoxy) is so negligible that the advantages of having the IHS by far outweigh those of without the IHS with respect to heat management. With the IHS, the clamping force allowed is at least doubled, and the larger area further ensures a more even mount. Better clamping force is of course better for heat management, and if you are really concerned about the added thermal resistance from the thickness of the the copper IHS, then you can always lap it down, but even that has a very valid purpose, (from intel datasheet #30255301)

The primary function of the IHS is to transfer the non-uniform heat distribution from the die to the top of the IHS, out of which the heat flux is more uniform and spread over a larger surface area (not the entire IHS area).

but whatever, if you want to...

just keep in mind that intel would most likely use the most robust thermal epoxy they could find

[snowwie]

I think that any gain in lowering temps is either within margin of error or placebo effect or both

My serious suggestion would be to lap it down as far as you could.

[Doomeous]

I removed my prescott 2.8E the day I received it from fed-ex. I remove ALL my p4 spreaders, it makes them all come down in temp, and slightly improves overclocking. Anyways, back to the prescott, I used 2 razor blades, just like the standard P4's, just gently work under the metal starting at one corner and slowly let it slice its way around, once all the way around, it would appear to be glued to the chip, but it's not, I grabbed another razor blade and did opposite corners on a flat table while gently prying upwords. Poof! it popped off and there ya have it, a perfectly flat Athlon-like core ready to be cooled a bit better. My 2.8E on my max3 motherboard will boot at 3.92ghz, but not run stable, run up to windows at 3.85ghz but crash, get into windows and run at 3.73ghz all night long at work in the cold of the night, but have to clock down to 3.65 for this damn summer heat that goes all night and day ~8) ... I suppose liquid cooling or vapor cooling would easily run this baby up to 4ghz +.. but I have this machine custom configured into a portable bag that I pack to work.. looks like I have a large lunch is all.. then I plug it into a work monitor and game all night eheh.. So I'm stuck with air cooling for now.

[enzoR]

Originally posted by fafnir
yeah, but he's says its a "secret" how he got the chip, and from the way the chip looks, it doesn't seem there was a heatspreader there to begin with


looks clean, too clean, doesn't quite seem right, but its definitely a prescott (tried it in 5 different board to make sure i wasn't seeing stuff), and a 3.2E


hey freecableguy, how did you try to remove your M0 heatspreader? you didn't just cut and then pry really hard did you?

and also, did your chip core come clean off the pcb or was there more carnage?

DUDE thats a ES chip for factory testing. they took it off the production line before the HS was bonded on. the ES sent to reviewers and stuff has the multis unlocked aswell but the heatspreader is bonded on.

[Çhrist0ph]

my god this is sad. Its only a few degrees. Since when did a few degrees get you a sizable max overclocking increase. Sure it helps, but not much. Not enough to warrant 5 plus dead cpu's

[skate2snow]

Originally posted by HKPolice
I doubt even -60C would rip it off since they've done cascade runs at -110C? The IHS remained intact

I searched Epoxy resistance on google, And got a PDF, it was saying a very high temp of resistance(I dont remember what it was, and I cant find the PDF back ). And I remember that it was around 60C the cold resistance. The reson they have no problem at -110C is becoze the cascade is putting a pressure. I hope you warant thinking that cascade was holding in the air

[snowwie]

yeah, the reason why it doesn't come off at -110C is because it is under compressive stress, no big deal; try twisting or prying the IHS at those temps and it's a diff story

[Perc]

Originally posted by freecableguy
My M0 2.4C has now been asked to star in "Comando II"....YES, there was much carnage...poor little guy, He never knew it was coming (the chips shattered into MANY pieces).

probly a little off topic by now but had to reply

man i did the same thing. from the mail box to the trash can in 60 seconds!! i will never never remove another freakin ich again! the only MO steppin i ever got my hands on and i trashed it in a minute flat.

peace perc,

[MentholMoose]

Originally posted by Çhrist0ph
my god this is sad. Its only a few degrees. Since when did a few degrees get you a sizable max overclocking increase. Sure it helps, but not much. Not enough to warrant 5 plus dead cpu's

Well this IS xtremesystems.org...

[lalPOOO]

Originally posted by Doomeous
I removed my prescott 2.8E the day I received it from fed-ex. I remove ALL my p4 spreaders, it makes them all come down in temp, and slightly improves overclocking. Anyways, back to the prescott, I used 2 razor blades, just like the standard P4's, just gently work under the metal starting at one corner and slowly let it slice its way around, once all the way around, it would appear to be glued to the chip, but it's not, I grabbed another razor blade and did opposite corners on a flat table while gently prying upwords. Poof! it popped off and there ya have it, a perfectly flat Athlon-like core ready to be cooled a bit better. My 2.8E on my max3 motherboard will boot at 3.92ghz, but not run stable, run up to windows at 3.85ghz but crash, get into windows and run at 3.73ghz all night long at work in the cold of the night, but have to clock down to 3.65 for this damn summer heat that goes all night and day ~8) ... I suppose liquid cooling or vapor cooling would easily run this baby up to 4ghz +.. but I have this machine custom configured into a portable bag that I pack to work.. looks like I have a large lunch is all.. then I plug it into a work monitor and game all night eheh.. So I'm stuck with air cooling for now.

Post more info about your chip please (stepping, where it was bought, where it from...) Its possible that not all M0/Prescott cpus have the ihs epoxied on, although I'm pretty sure they do...

[Çhrist0ph]

Originally posted by MentholMoose
Well this IS xtremesystems.org...

yeah, say that after you've slaughtered 5 cpu's in a row...

[froudeg]

I've attempted to remove the HS off a recent P4 3.2 northwood, using same way i removed one from my old 2.4 NW - but its a no go, it wont shift.

Like you, i tried freezing, and controlled heating....initial theory was it was soft soldered on - nope...upto 250 celcius and it still didnt budge an inch.

I froze it to -50c - no effect.

I bent the substrate significantly with a large flat blade screw driver prized under the HS, trying to pry it off - no chance at all.

They are epoxied on solid, and the epoxy thats holding it on is stronger than the epoxy that holds the die to the substrate - so it will never come off in this way - the only way is to sand it down.

Btw, some very early prescotts from a certain plant wernt epoxied on....and from what i can tell ALL chips coming from ALL plants, be it northwoods, extremes or prescotts are now using epoxied on HS's.

Theres prolly still a lot of old stock out there for northwoods that arn't epoxied - but dont think northwoods arn't epoxied - as mines proof they are....btw this was a northwood without as many caps on the back as the newer ones - so you cant judge by that either.

[froudeg]

Ok heres a thought ive just had.

There is something that might work (bar sanding it down) - an epoxy dissolver...they do exist!

Only problem is it might dissolve the stuff thats used to hold the die to the substrate - but its worth a shot.


See if you can get hold of this stuff:




http://www.fiber-tensioners.com/BIO-SAFE1025specs.htm

[freecableguy]

i would try that epoxy dissolver...except don't just throw it in there. Rather, sand it all the way down and then use the solvent on a q-tip to apply the dissolver exactly where is want it....to the top of the die. Clean the rubber when the IHS seals to the substrate, then lightly lap the die. You should be done.