Gouge marks on WB from IHS!

[jman1310]

i'm using a brand new DD TDX and an Intel 3.2 P4
the CPU has been in several systems under a Swiftech MCX4000 and one other WB
i just took it off to see if i could get a better mount and found that the 4 corners of the IHS gouged the base of the TDX
the 2nd pic is a selection of the 1st blown up 200%
using a metal straight edge the IHS and the TDX are nearly flat, minimal light is visible
with no TIM joint if the CPU is placed on the TDX, there is a slight bit of stickyness (can't remember what it is called - similiar to surface tension in water)
should i lap the IHS? maybe just round the corners a little bit?
i'm thinking that the IHS may not be fully contacting the CPU and when pressure is applied the IHS deforms slightly
or
i applied too much pressure when mounting the TDX but i don't think so as i only compressed the springs about 60~70%

thanks for your thoughts

[JamesAvery22]

I went through a fair amount of xeons (just bought different steppings didnt fry em or anything ) and all of em tended to leave marks on the stock coolers. The aluminum bases of the stock heatsinks tended to get marks from ther corners of the IHS just like your DX got. Some were more concave than others. They never marked up my mcw5002s though. Used the same spring clips so it was the same amount of force. Guess the aluminum was just softer? Guess your TDX springs are pretty stiff.

If you are going to remove the IHS and lap it why not just remove it and make a shim so you can mount the wb directly to the die?

[Delirious]

It shouldnt really hurt performance as the marks are on the edges of the ihs and not over the core.

[bigben2k]

That's puzzling.

The IHS is nickel plated copper (last I checked), so the softest part of the contact is the block's copper baseplate. Copper is "relatively" soft, so I would expect the gouging.

You mentionned that the base and block stick well together (I know the effect you're talking about, no, can't remember the name either). Did you try turning the block around? (especially 90 degrees). That would reveal any warpage.

DangerDen's site says to compress the springs down to 10mm; if you went any further, I'd expect the gouging.

Overall I think you'll be ok.

[JamesAvery22]

Originally Posted by bigben2k

That's puzzling.

The IHS is nickel plated copper (last I checked), so the softest part of the contact is the block's copper baseplate. Copper is "relatively" soft, so I would expect the gouging.

You mentionned that the base and block stick well together (I know the effect you're talking about, no, can't remember the name either). Did you try turning the block around? (especially 90 degrees). That would reveal any warpage.

DangerDen's site says to compress the springs down to 10mm; if you went any further, I'd expect the gouging.

Overall I think you'll be ok.

Now that I think about it I remember one of the IHS looking like the corner was raised. Kind of like when you cut/shear sheet metal and it raises the cut edge.

[Joe Camel]

sure you can lap it (but youll loose the stepping)

only thing i see a problem here is: if it were FLAT, why would only the 4 corners leave marks :shrug:

here is my 3200 Winnie after a few minutes of lapping:




and here it is after quite a few more minutes:
yep, there is copper under there.



the term is wring

** wring:
To measure with a gauge block you simply select a combination of blocks (using the fewest in combination you can, in a process called stacking) which adds up to the length you want, "wring" the blocks by rubbing the appropriate ends against each other to squeeze the air out from between them, and compare the length to the target. One interesting feature of gauge blocks is that the surfaces are so smooth and flat that once they are mated in this way they stay together unless considerable effort is used to part them. The combination of the adhesive action of the ultra-thin film of preservative oil or moisture on the blocks and the molecular attraction, or bonding, between the very flat and parallel mating surfaces, will actually hold them together.

[nikhsub1]

A little lesson in the Intel IHS's; they are CONCAVE by design. Leave it alone. The corners are supposed to be raised a bit, this ensures a good mount - ie; pressure on IHS to core.

[Joe Camel]

oops... INTEL not AMD! sorry

Nik, im @ a bit of a loss as to how a concave IHS can help contact to the core.

o wait, to the CORE....ok ill give ya that. but what about the contact from the IHS to the WB? anything other than flat is a "give & take" ie better contact between core to IHS but worse contact between IHS and HSF/WB.

[nikhsub1]

Joe, under the Intel mounting pressure specs (don't recall offhand), the IHS flattens out; read the IHS becomes flat when an HSF or WB is mounted with the recommended force.

[Joe Camel]

huh, the "glue" must "give" and let the IHS edges "go down" under pressure.

good info (as usual) Nik!

nice to know (if i ever turn to the "dark side" )

[Roscal]

Originally Posted by nikhsub1

A little lesson in the Intel IHS's; they are CONCAVE by design. Leave it alone. The corners are supposed to be raised a bit, this ensures a good mount - ie; pressure on IHS to core.

No, they are concav (or convex on some processors) because of a bad process and poor quality control ! IHS MUST be flat, datasheets are very explicit about that. We don't need to use elasticity and softness of the copper to make a good contact, this is an urban legend. If all parts are flat (core, IHS and base) and if a good pressure mounting is applied properly, all parts remains flat and contact will be *perfect* (core->IHS and IHS->base because vertical load is applied to whole system) ! Make a diagram and you'll understand, an numerical mechanical simulation could also show interesting visual results.

The reason to flatten out and polish an IHS is true and provide real benefits about temperatures (some degrees under load, easy to see under controlled situations). Sure, a concav IHS flattens out a bit under pressure, but it's always worst than a true flat one. The IHS center is the most important thing and if the HFS/WB don't touch it in a correct manner (too much thermal paste or air gap), perfs will be bad, even if WB touch correctly the 4 corners/edges... A non flat IHS always remains concav under load (we talk about small distances <<0.0x mm ) and the 4 marks under the WB are an evidence.

[nikhsub1]

Roscal do you have a linky?

[Roscal]

Geometrical and mechanical datasheet of pentium4 on their website. From memory, flatness should be less than 0.05mm all over the IHS area, some IHS are far from this requirement.

EDIT : Found it in my docs, see the attachment. First symbol refers to flatness (real surface must be enclosed between 2 parallel surfaces separated by 0.05 mm), second symbol is parallelism condition with the lower surface A.

[Jag]

Rosco,

Don't we have to take in consideration the effect of warpage on thermal interface, as shown in this picture:

[Roscal]

Thermal stress in ball array always exists (epoxy encapsulant between solder bumps reduces it a bit) and is annoying when there's a big temperature difference like in the article on Electronics Cooling where tests are made between 20°C and 150°C. Dilatation effects are very small (some microns or less, depends of length and temperature, look at their tests with a ~5µm per fringe) which is at least one order magnitude less than a geometrical defect on a concave IHS. Don't forget that on their tests, there's no IHS on the die, so no pressure load to constraint the dilatation effect.