Removed the IHS from an Ivy Bridge i5 3570k

M

Moogoos

Limp Gawd
Joined
Apr 30, 2012
Messages
334
I decided to mess around with the newly discovered TIM bound IHS on the Ivy Bridge CPUs. First step is removing it:

This tool works VERY well. Others that are thicker don't. Get it if you can at Home Depot:

IMG_5297.jpg


IMG_5298.jpg


Place the blade on each corner and firmly press into the corner. Don't saw! It should just slip in with nice even pressure.
Be sure to have the CPU on a matted surface to handle any top pressure you may apply:

IMG_5299.jpg


Get each corner to this depth and then you'll feel the IHS get loose:

IMG_5300.jpg


And it's ready to clean:
IMG_5302.jpg
 
Nice work. Make sure to give us before/after temps, and which TIM used. Also, I wonder if one can actually mount a HSF without the IHS like the old days?
 
nice man. what are u gonna clean it with? and what are u gonna put in place of the TIM?
 
senakun said:
nice man. what are u gonna clean it with? and what are u gonna put in place of the TIM?
Click to expand...

I am going to try rubbing alcohol to clean it.

I am first going to try shin-etsu TIM, then I have a couple other TIMs coming in to test out as well.

This chip looks like the old P3 flip chips when the IHS is removed. I wish I could do away with the IHS and go direct contact. Maybe there's a way.

ShuttleLuv said:
We were doing this 10 years ago guy but good job and post results. :)
Click to expand...

Yeah I was there with ya 10 years ago. :p
 
Snap!! i mean coooolll... :cool:

I've read ppl mounting the heatsink straight on it but they gotta remove lid from the mobo ...
and without a shim gotta be careful how much pressure you apply on there....

Intel's bringing back the old school!
 
I think I can go direct. I found that an allen wrench can remove the motherboard top plate.

DSC03305.jpg


Removed:

DSC03309.jpg



I also use an H80 cooler which comes with 4 mounts at height depending on which CPU one has. The AMD mounts are lower so if I can remove the plate I should be in pretty good shape for more direct contact. The capacitors might be a problem though. We'll find out in about an hour.

Crap, I'm gonna be up all night. But tomorrow is Diablo III release and I've been waiting 10 years so this better work. :p
 
Last edited:
Just make sure theres enough pressure for it to keep the cpu pin in place (thats what the top plate did) and not too much as to cruch it.
Also the first ppl that tried the heatsink straight on the die reported higher temps but that was because they didn't have the proper mount and most likely it wasn t making proper contact, but from what you've just wrote you seem to know more than them!

God speed , we're behind you....
 
Nice! I was waiting someone with the balls to rip open their Ivy here on [H] and give us some testing results! I'll be sticking around to see how this turns out, thanks Moogoos!
 
Got it all nice and shiny! I may have scored the layer a little when removing the IHS. We'll see.

DSC03310.jpg


Thumbs up! so far.....

DSC03312.jpg
 
Nice, can't wait to see the results!

So are you essentially going to rely on the tention from the water block to keep the CPU in place?
 
RamonGTP said:
Nice, can't wait to see the results!

So are you essentially going to rely on the tention from the water block to keep the CPU in place?
Click to expand...

I don't see a problem with it as long as it's tight and secure. Lok-Tite it on there :D
 
Yeah, after having posted that, I went back and looked at images and.....not so much alike.
 
I'd be worried about crushing it with direct contact without the socket bracket. Any way to make a metal shim that goes around the edges but leave the die surface area open?

Using the socket bracket helps ensure proper pin contact without over doing it.


I am interested in seeing how direct contact with die works out but at the same time I don't want to see you destroy your cpu.

Good luck.
 
brings back memories. I remember getting a mobile amd64 processor and rigging it to work in my desktop. Took some trial and error since desktop heatsinks were made a little higher due to the ihs.
 
Another thing that would come in handy would be a perfectly sized spacer to keep the block from squeezing up too much when you tighten it. Basicly just screwdown till it hits the spacer and done.
 
SonDa5 said:
Another thing that would come in handy would be a perfectly sized spacer to keep the block from squeezing up too much when you tighten it. Basicly just screwdown till it hits the spacer and done.
Click to expand...

Like those little rubber bumpers the Athlons used to have. Maybe just cut a small piece of rubber or foam to put on the corners - help keep the heatsink flat also. Most people that cracked the dies before cracked corners off when the heatsink was crooked - as long as it is flat it should take quite a bit of pressure.
 
I would use a small torque wrench unless you have good "feel" for tightening the screws down evenly. Possibly torque it down minimally, just to contact, and turn on the computer to run tests to see what kind of temps you get. Then if the temps are bad, try a little bit more torque evenly on all four screws. Repeat as needed.
 
Well after a couple setbacks I was able to get the H80 block on the CPU. It was supported by the AMD screws that come with the H80 but I had to elevate them slightly with a nut as they were too low. I also had to widen the support bracket holes to take the AMD screw heads.

I was able to then torque the screws down and mount it for what felt to me firmly. So far temperatures are about the same. 30° idle and 57 under prime95 stock. I'm going to load an overclocking profile and see if it's any better. I might adjust the TIM too as I think I may have put too much on. Hard to tell right now. The good news is it still works. :)
 
You won't need much TIM AT ALL. The core should have a very smooth surface without defects. The surface of your H80 will be the only surface that would need TIM to fill in the "cracks."
edit: I might actually consider lapping the surface of that H80 to a mirror finish and running it TIMless!
 
I loaded my 4.5 profile that I ran stable before at 69º and it doesn't seem much better. In fact I think it's worse currently. Core 1 and Core 4 run at 70º while Core 2 and Core 3 run at 80º. I think I'll have to reapply the TIM again tomorrow and check how well this first round seated on the die.
 
One of the other guys (maybe the first) that tried running it lidless also got temps that were about the same. Might need to increase the pressure a little - but that die is so small it might not matter much.
 
Crazy Chuckster said:
Have you thought about putting on some good TIM and putting the lid back on to see if it improves temps?

See this here http://www.eteknix.com/news/ivy-bridge-thermal-grease-is-the-culprit-for-heat-problems-after-all/
Click to expand...

I second this. Having the lid on greatly reduces the chance that you're going to torque the cooler down too much (while still allowing you to apply enough pressure to ensure good contact with cooler and good pin contact). Something about having the cooler directly on the die just makes me nervous. And then there are the few forum posts that have already been mentioned that talk about people getting higher temps when they direct mount.

I also found an article talking about someone testing the thermal paste and not getting any better temps when he replaced it, which leads me to wonder if possibly just certain batches of Ivy Bridge got the shoddy paste job...

http://www.decryptedtech.com/index....used-by-thermal-interface-material&Itemid=138

Anyways... Good luck and I still look forward to seeing some more results from your testing!
 
I think you will get it cooler...good job and thanks for posting
 
good work. but everyone should have realistic expections.

no one can beat intel at intel's game.
 
since you might not be getting even pressure with your water cooler, building a shim is the obvious next step. you could also rig up a smaller heatsink that is of more manageable size, to test and compare. if it worked better you might guess the contact was off on the other one. example: Dynatron R15: 435g copper vapor chamber, 90.0 x 90.0 x 27.0 mm, rated 150W for socket 2011. vapor chambers are very effective at spreading heat out.
 
if it takes away 10C then its worth it... but if it takes away 3C or less, then it is not worth it to void your warrenty for 3C less ..
 
That's a very clean looking chip surface.... The die that size looks like people might even have simpler opportunity to do direct-die water cooling....
 
You must log in or register to reply here.
Back
Top