IHS removal and a Fusion WB

[Exahertz]

I have an e2160 that id like to push just a little harder, I figured since I'm going to be changing the liquid in my rig that I might just go ahead and remove the IHS while I’m in there.
I was wondering how fragile the die is and if the Fusion WB with its convex shape would cause any problems? I have a nice machine at my job that could polish the fusion block perfectly flat (w/ a mirror finish).
Is the die brittle?
And how tight should I press the block to the die?

It’s already running a 3.0GHz and I’m hoping to get to 3.2 or more. I’m not too worried about this CPU as it’s only worth $40 and the warranty is already long gone.


CPU's Die Under 230lbs Of Pressure (http://www.youtube.com/watch?v=iNd8ddu1Z58)

[stangracin2]

i'd say go for it

[Exahertz]

hmm... well i am pretty sure i can get the IHS off and expose the die, I know already that the die is not soldered to the IHS but what im really asking is:


when the die is exposed, can i damage it by compressing the water block too tightly on to the face of the die?
and since the water block is bowed won't it exert too much pressure on a small spot of the die?
lastly, wouldnt a flat WB be better than a bowed one for contacting the die?

[mattkosem]

when the die is exposed, can i damage it by compressing the water block too tightly on to the face of the die?
and since the water block is bowed won't it exert too much pressure on a small spot of the die?
lastly, wouldnt a flat WB be better than a bowed one for contacting the die?


Lots of CPUs used to be sold with exposed dies. Intel and AMD both used that design on multiple cpus.

1. Yes you can but, with a sane amount of pressure, it's only really likely to cause damage if the block isn't perfectly flat against the die.
2. Potentially. You could always lap the block while you're at it.
3) yes, probably not by that horribly much though.

--Matt

[DeathWalking]

Don't tighten the block down as much as you would normally...if you apply 60ftlbs of pressure, which is on the low end of good on an IHS, you're going to crack the die.
I would apply a little bit of thermal compound, and try a test mount, and see how good of a spread pattern you get. If it's bad, lap the block. If it's good, then don't bother to change anything.

[Exahertz]

ok i got the IHS off but i noticed that the die's face sits lower than the socket 775 retainer clamp (that big clamp that holds the cpu tightly in the socket with a lever).

I guess i have to remove the clamp, right?

[smee]

Yes you'll have to remove the clamp. :up:

But, as for waterblock and no IHS...
You'll more than likely get better results with the IHS on, because the heat will be more spread out on the IHS which will help the Waterblock pick up the heat quicker and get rid of it faster.
With the Die being in one small spot, the waterblock will be getting very hot in that one spot on the block and you probably won't get any better temps, possibly worse.

And you also risk cracking the die if you get a bowed block.
I'd go for a block that is flat, like the Fuzion V2. :up:

Keep us updated, let us know how it goes. :)

[Exahertz]

Ok, Im going to lap the block flat then

Thanks for the suggestions.

I was looking at the TIM intel used on the die/IHS and i saw i gaping air bubble in there :eek: so im really hoping this helps...

[nikhsub1]

But, as for waterblock and no IHS...
You'll more than likely get better results with the IHS on, because the heat will be more spread out on the IHS which will help the Waterblock pick up the heat quicker and get rid of it faster.
With the Die being in one small spot, the waterblock will be getting very hot in that one spot on the block and you probably won't get any better temps, possibly worse.
Totally incorrect. It's like saying that you could get your hand colder holding ice with gloves on vs. without gloves.

[_G_]

Ok, Im going to lap the block flat then

Thanks for the suggestions.

Instead of lapping that block why not try one of these?
http://www.jab-tech.com/Swiftech-Storm-Universal-Waterblock-Rev-2-pr-3022.html

[Exahertz]

Ok i spent most of the night lapping my fusion, but i couldn't get it perfectly flat... when i mated the two surfaces (Die to Water block) and held it in between a bright light i could still see a sliver of light as i rocked the die back and forth (when the two where sitting parallel you could see maybe a 1-3 thousands of an inch gap on each end).

I figured that gap would easily be filled with some arctic silver, and proceeded to install the CPU.

I tried my best to install and tighten the water block evenly. then i changed the water in my loop.

after running the pump all night to check for leaks and extract the micro bubbles i reconnected the PC and powered it on. I immediately jumped into the bios to monitor the temperature of the CPU and noticed 5*c drop in idle temps.

So i booted into windows, ran speed fan and some short bursts of SuperPI. I noticed about an 8*c drop in load temps there.

Then i ran Sandra's CPU arithmetic test. An I did notice i drop in load temps EXCEPT, There was a spike on Core0 that seemed unusually high for a short second or two it hit about 65*c, while Core1 never went above 50*c and did exhibit the same spike but not nearly as much.

Could this indicate that the water block is mispositioned or not parallel?

[Exahertz]

I just ran Prime95 Full Load Torture test. Core0 @ 57*c and Core1 @ 54*c

so I'm not sure what that spike was but these are the full load temps of an Intel E2160 overclocked from 1.8Ghz to 3.01Ghz

Definitely a drop in temps, though i cant say exactly how much...

[Bobly]

But, as for waterblock and no IHS...
You'll more than likely get better results with the IHS on, because the heat will be more spread out on the IHS which will help the Waterblock pick up the heat quicker and get rid of it faster.
With the Die being in one small spot, the waterblock will be getting very hot in that one spot on the block and you probably won't get any better temps, possibly worse.



Correct me if I'm wrong but this is how I see it:
The HIS is a heat spreader right? And while I doubt it's made of gold or silver, it sure ain't copper, and next in terms of thermal conductivity is aluminum which is probably what it's made of.

Now remove the HIS and apply a copper heat spreader (aka the base of your water block) which is being cooled DIRECTLY by water as opposed to having the heat go through an aluminum head spreader, thermal paste and copper, isn't that going to cool it better?

[Exahertz]

The IHS is not aluminum, Its very heavy. When you lap the surface of the IHS you get into a copper layer. so it may just be solid copper that is coated with something.

Eithway, my temps did improve! :D

[bryan_d]

Exahertz,

Congrats on the achievement, but post some pics of the mount please! :)

I already took off the IHS on my e4300 and am running it @ 3.2GHz, but I intend to WC the chip and am hoping for 3.6GHz. My Sunbeam (AC Freezer 7 Pro wannabe) is sitting nicely with the retention bracket removed, and connected with CD-ROM screws to the Thermalright backplate.

http://i5.photobucket.com/albums/y196/bryan_d/IMG_0266.jpg

And for those thinking it foolish to remove the IHS, unless you have an efficient means of heat transfer between IHS and die, the IHS does more to hamper thermal efficiency than good; This is aggrevated with crappy TIM used for the E2/4/5/7000 series. I know the TIM application for my e4300 was aweful.

Congrats again, and post some pics!

Bryan D.

[stangracin2]

if i ever get an E8x00 i'll take the ihs off of my 5200

[Exahertz]

http://i44.photobucket.com/albums/f9/laserskys/100_2787.jpg

this is the best pic i could take, its too crowded with tubing to get my camera any closer.

[stangracin2]

nice

[Sparky]

FWIW, I used a fuzion v1 in stock form (with that slight bow it has) on my IHS-less opteron 165, cranked down as hard as I could. Never cracked the die, cooled well, and that opteron is still chugging along just fine in a friend's PC now :D

[bryan_d]

Awesome job again,

Here is a sneak peak of what is to become of my E4300 in the coming week:

http://i5.photobucket.com/albums/y196/bryan_d/100_1446.jpg

I know it will not be top performer, but given my ambient temps I am sure it will be enough to reach lower temperatures.

- Question for you hurtz, does tightening down the waterblock further produce lower temperatures by any chance?

Nice job again,
Bryan D.

[Eternalightwith]

Core0 temp increased over the others due to the very slight imperfections you couldn't get out with lapping?

[Exahertz]

- Question for you hurtz, does tightening down the waterblock further produce lower temperatures by any chance?

Nice job again,
Bryan D.

Thanks for the complement ;)

At the moment the WB is not very tight. I was skeptic of going too tight so as not to damage the die. but tomorrow during my lunch break ill take an old Pentium III and see just how much pressure the die can take before it cracks.

should be interesting...

[Exahertz]

Here it is!

CPU under 230lbs (http://www.youtube.com/watch?v=iNd8ddu1Z58)

[stangracin2]

did the core crack

[Exahertz]

It was a junk CPU that i used just to test how strong the silicon wafer is.

The core did not crack. i was unable to find ANY cosmetic damage at all. but i am not able to test that CPU as i dont have the correct motherboard...

[Exahertz]

I just tightened down the water block (not even that much) and really brought the temps down on core0

http://i44.photobucket.com/albums/f9/laserskys/TightBlock.jpg

Time to keep on tightening!

[NaeKuh]

Here it is!

CPU under 230lbs (http://www.youtube.com/watch?v=iNd8ddu1Z58)

omg it survived... uhh wait.. so what happened to all the resistors under the die?

You know the things that look like this which is annoying when you lap your CPU:
http://i125.photobucket.com/albums/p73/aigomorla/IMG_1144.jpg

OMG I CALL SHINS IF HE DIDNT CRUSH THEM....

Wait.. gotta be nice..

And... that pressure system was FLAT... not bowed...

The fusion is bowed.... so your putting pressure at 1 point and not the entire surface area... does this make sense??
At that kind of pressure, the die will crack if its not flat...

My recomendation goes ocz block as its not bowed, but stepped...

then you can safely drop that 230lb pressure on the block, because the motherboard has a hole for those resistors not to get crushed..

no seriously exahertz did u do that??? i want to see the underside of that cpu...

[crazy1323]

Don't tighten the block down as much as you would normally...if you apply 60ftlbs of pressure, which is on the low end of good on an IHS, you're going to crack the die.
I would apply a little bit of thermal compound, and try a test mount, and see how good of a spread pattern you get. If it's bad, lap the block. If it's good, then don't bother to change anything.

Ft Lbs is a measure or torque not pressure. If you apply 60 lbs/ft to the screw for a heat sink you will for sure crack something. To put this into perspective I attach the wheel to my car with 78 ft/lbs of torque.

[Nanometer]

lmao torquing a cpu bolt to 60 foot pounds rotf, id say more about 5 pounds

[Exahertz]

omg it survived... uhh wait.. so what happened to all the resistors under the die?

Nothing...
http://i44.photobucket.com/albums/f9/laserskys/downsized_1223081243.jpg

dont mind the jumper wires that you see bonded to some of the pins. this was a test CPU from our SQUID testing lab

[NaeKuh]

Nothing...
http://i44.photobucket.com/albums/f9/laserskys/downsized_1223081243.jpg

dont mind the jumper wires that you see bonded to some of the pins. this was a test CPU from our SQUID testing lab

OMG AND IT HAD PINS!!

:rofl:

Exahertz seriously EPIC!

:up: