Lapping A64 IHS and MSI K8N Neo2 Platinum Chipset (DONE, upto 16c Lower) *56k beware*

[FearTec]

This mini project was to check and see if my A64 IHS was concave in order to ensure the water block was in full contact with the processor.

Check Existing Thermal Paste

I remember the time before last when I applied the thermal paste I checked and saw that there was a gap in the AS5 in the centre of the processor (concave IHS or bad placement of the block?).

These photos show that last time I evenly spread the AS5 and made sure it covered all of the IHS as best I could.

Water Block 01


Water Block 01


A64 and AS5

[FearTec]

Lapping the A64 IHS

The only way to check and see if the IHS was concave was to get some 1200 grade silicone carbide (wet & dry) paper on a panel of glass and slowly move the IHS over in one direction and check for minute scratches in the IHS. If the scratches were even then the IHS is flat, if the scratches are random then this is showing the high points of the IHS.

The IHS showed signs of raised edges, I decided to lap the A64 IHS.

Using a piece of glass, anti static strap and 400, 600, 800 and 1200 grade silicone carbide wet and dry I carefully moved the A64 in circle patterns for 10 seconds, rotated the CPU 90 degrees (to prevent overdoing one corner) and repeated. I started from the rough 400 grit paper and repeated up to the fine 1200 grit.

Rotating A64 over 400, 600, 800 and 1200 silicone carbide (wet and dry) paper.

Finishing polish (light pressure from above).

[FearTec]

Lapped A64 IHS

During the lapping process I could see that the raised area about 4mm from the outside of the processor was being lapped. it took some lapping to get the processor to a smooth (mirror) finish. I know some people say mirror finishes are bad but it was either that or a rougher finish.

Mmmmm shiney and smooth (see where the high ridge was around the edge of the processor.


High ridge lapped


Reflection 01


Reflection 02


Reflection 03


Reflection 04

[FearTec]

Check MSI K8N Neo2 Platinum Chipset Thermal Paste Contact

While I had the main board out I decided to check the Chipset's thermal paste contact as many said MSI do a poor job. I have noticed before that my chipset temps have risen quite high when doing intensive tasks. It was a good thing that checked the chipset thermal paste as it was a crappy layer of white paste and it was thick and hard. I also noticed that there was an oily residue below the white paste, it was as if the white paste components separated.

It was lucky the chipset was being cooled at all.

Poor Stock Paste.


Little Paste on Stock Chipset Fan.


Little Paste.


NForce3 Chipset


Stingy MSI.

[FearTec]

Lapping the MSI K8N Neo2 Platinum Chipset Fan

The chipset fan was rougher than a bad concrete job so I lapped it also

Lapping the Chipset fan.


Large convex bump in the chipset fan.


All smooth and shiney.

[FearTec]

K8N Neo2 Platinum A64 CPU Backplate

fyi: Here is a photo of the original MSI K8N NEO2 PLATINUM CPU back plate.

[FearTec]

I have not booted the PC yet as I am still leak testing my new revised water-cooling pipes.

I will post results and temperature differences soon (if I have not zapped the chip)? Fingers crossed.

Also thanks to javascripterror for his research into this problem.

[-ShadowRunner-]

c'mon c'mon! I'll probably lap mine if you see a marked improvement. Scary about the chipset being so drastically cheated of thermal compound.

EDIT: Is the concavity to blame for the block sticking to the IHS? or simple lack of air between the two surfaces?

[uberbanzai]

You'll probably get a few .C difference. Best thing to do would be to actaully remove the shim off the top of the die. It's be done on dozens of P4 cpu's and I've seen it done on a few A64's before too.

Ryan

Edit:

Links: Removal of the A64 IHS
Google's links

[-ShadowRunner-]

Is the reason people get better clocks on phase change because it cools the whole die more effectively? whereas with water cooling the transistors towards the bottom of the die are further from the cooling source and still get hot but just conduct their heat to block through the die?

It's hard to explain what I'm trying to ask. Eg. someone will get 3.2 on a Venice with phase-change, but someone overclocking a venice on water, despite the fact that their load temps might be below an air-cooler's idle, they hit a wall long before the same clock. Irrespective of variations in fab etc.

[tHRiCe]

Sorry, last two posts by -ShadowRunner- were actually by me, silly goose left himself logged in on work comp :P

[FearTec]

Originally Posted by uberbanzai

Best thing to do would be to actaully remove the shim off the top of the die. It's be done on dozens of P4 cpu's and I've seen it done on a few A64's before too

I looked into this and found that once you remove the IHS the Water Block may not rest on the die as the new (lower) height of the processor is below the top of the raised bit on the actual ZIFF socket and I would have to grind back one side of the water block or grind the ziff socket down 3mm.

I may remove the IHS when it is spare and not my main processor.

Must air coolers have one side of the base cut away to prevent this but water blocks are generally 100% flat.

[uberbanzai]

I removed the IHS from my P4 and removed the HSF holding mechanism so I could make it sit flush again.

Ryan

[tHRiCe]

Originally Posted by FearTec

I looked into this and found that once you remove the IHS the Water Block may not rest on the die as the new (lower) height of the processor is below the top of the raised bit on the actual ZIFF socket and I would have to grind back one side of the water block or grind the ziff socket down 3mm.

I may remove the IHS when it is spare and not my main processor.

Must air coolers have one side of the base cut away to prevent this but water blocks are generally 100% flat.

oOOo ur here! TEMPS!?

[slobber]

thanks for the guide FearTec

I have done similar using a P4 and lost around 3' Celsius which was welcome but am I right that doing this would void your warranty ??

Intel told me it does