Removing the Prescott Heatspreader

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

Quote:

---

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

---

_

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...

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....

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.

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

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

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.

Quote:

---

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....

Quote:

---

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...

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.

Quote:

---

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, :owned: He never knew it was coming (the chips shattered into MANY pieces).

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

Quote:

---

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?

---

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)

Quote:

---

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

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.

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.

Quote:

---

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.

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 :eek:

Quote:

---

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:confused: ). 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;)

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

Quote:

---

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, :owned: 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,

Quote:

---

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 :eek:

---

Well this IS xtremesystems.org... ;)

Quote:

---

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...

Quote:

---

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

---

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

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.

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

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.

Quote:

---

Originally posted by 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...

---

yeah, thats the only way i thougt of as well. you should have access to a lapping machine, no? just put in the cpu upside down and use very rough sand paper. the thickness of the ihs is well documented so you know when you should switch to a thinner grid.

the prescott without ihs you saw was most likely a mobile prescott. they were released some weeks ao and should be available soon. they come without ihs like all mobiles.

IF somebody managed to remove the heatspreader of a prescott or gallatin core then he/she was VERY lucky, and there was probably bad applied thermal epoxy.

the only guy i heard of that took off the heatspreader of a prescott was on vr-zone, and he ripped of the entire cpu of the package. i also heard of a guy who managed to remove the ihs of his m0 chip with gallatin core wich was epoxied... WITHOUT killing it... and the gallatin core is significantly larger than the prescott core... but i think that was just good luck, i wouldnt try to reproduce that.

as i said, lapping seems to be the only safe way.

the pic i have is from vr-zone. and its not actually a pic of the prescott without heatspreader, its a pic of the prescott packaging without heatspreader and without core :D

they ripped off the heatspreader and the core came off with it :D

http://www.vr-zone.com/?i=308&s=1

i think its the same stuff they used on the newer northwood/gallatin cores and on all prescotts. i think they tried it first when people started removing the IHS of their M0 chips and then used it on all chips.

heres the final version btw :)

I have a question for you

those A64 packages there, do they apply for s940/939 cpus as well? does the FX have the same die as the clawhammer? the s939 cpus have the same newcastle die? or do those pics only refer to the s754 chips?

atm every 1Mb chip from amd is clawhammer and all 512kb chips are newcastle, no matter what package.

soon there will be newcastle 512kb L2 chips and oakville(?) 1mb 90nm chips.

i've done this to 4 pentium 4s now. 2 willys, and 2 northwood "c" chips.

I used a box cutter razor blade for three of the four and twine to floss out the compound on another

I haven't tried any prescotts yet. If epoxied, they present a problem.

EDIT: I guess epoxy would give you problems. Perhaps the one you saw was an early production of prescott

welcome to Xtremeystems! :toast:

here fafnir, theres a guy from austria who managed to remove the ihs of his prescott :)

http://www.xtremesystems.org/forums/...threadid=38814

http://www.ocforums.com/showthread.php?t=312344

um?

Ahhhh! It's my first 2.4C m0! Yep, I killed that one good. I think that was the first DIRECT proof that the new m0 steeping were failed Gallatin cores...

Yes, they do look like Gallatin cores, look at the width of them. Note to self: don't remove the heatspreader on ANY m0 stepping processors that I aquire.

Quote:

---

Originally posted by freecableguy
Ahhhh! It's my first 2.4C m0! Yep, I killed that one good. I think that was the first DIRECT proof that the new m0 steeping were failed Gallatin cores...

---

no, but it was the first direct proof that the latest m0s were all epoxied to the ihs :)

keep lapping fafnir, will take ya looong, but should be worth it. 4°C temp drop at least :)

what is up with the thread at ocers?

i think i have success here

this is a 3.0c

i used a razor blade and went around the edge of all sides of the HS

through doing this i was clearly breaking the epoxy seal, so i felt somewhat successful, and kept on going.

after all 4 sides were done the HS just kinda popped off.

it appears as if I got lucky and this was bonded by nothing more than some intel issue thermal compound.

i was a bit worried, but i have plans lined up for a 3.4ee, so i wasn't too worried.

all is well now,

my biggest concern is that from going around the HS with the blade, slight scratches have been made in the green outer coating of the chip, and I'm certainly hoping that these scratches did not kill the chip.

How I removed the S775 Heatspreader off the 3 ghz prescott.

I gently cut all edges so that I have some space to play with.

now I force the sharp (stanley knife) knife a bit deeper. You should be scared you will have to put some force on.

Ok now since you have cut all for edges but not the in between bits you should make your way slowely from one edge to another.
After completing that leave the CPU for 1-2 mintes ( mental pause ).
Now get some zahn seide?? ( string you use to clean the in between bits from your teeth )
And try to pull it through the HS. It will most likely break where the core is. Now you know the HS is completly sticking on the core.
Now get a very thin screwdriver used to turn VR's and other things and wrap a thin layer of toilet paper on the tip of it.
Now carefully put the screwdriver on one edge where you cut.
DO NOT TRY TO LEVER. Because you feel a great amount of force.
Now leave it for 5 minutes. Air will go underneath the heatspreader.
Now do the same for all four edges.
After this is done you can try to lift of the HS by a little force. Try doing it bit by bit otherwise you will most likely rip the core off.
I had some Silver epoxy or so on my prescott and the core didnt go off so its possible.
Its just a feeling where you know stop using force and leave it for a couple of minutes and try again.

Try each side of the cpu do not try to pop it off just from one side. You can also try using two thin screw drivers, insert them below the HS at opposite sides and slowly use force.

I hope it helped some of you guys, you must know these things dont like force so best thing is you just concentrate yourself and do everything slowly.

EmineM - pics please.

fafnir posted the pic a couple of posts above but here it is again

as this is what we're looking for, the seperation of the epoxy (marked in red) and not the seperation of the die, as you can see from freecableguy's first M0

http://www.xtremesystems.org/forums/...threadid=39084

heres the new thread :)