FLATNESS - anyone interested ?

[Joe]

I am actually sorta amazed this is a debate... I thought everyone knew that the letters are laser engraved.

[BillA]

my bone of contention is the plastic deformation of copper by a CPU

my (admittedly ignorant ?) numbers say it can't happen

and if it were happening, it would be quite easy to pick up when re-lapping

- and someone would be saying so . . . .

but I'm not hearing that

[bigben2k]

Originally posted by unregistered
my bone of contention is the plastic deformation of copper by a CPU

my (admittedly ignorant ?) numbers say it can't happen

and if it were happening, it would be quite easy to pick up when re-lapping

- and someone would be saying so . . . .

but I'm not hearing that

I think you've got your answer here, between myv65, gmat and Cathar.

More importantly, Thoroughbreds don't have any etching on the die, so what's the point of all this?

[myv65]

Yeah, I think everyone knows the letters are laser engraved and I think everyone knows the vast majority of the letters' area lies below the surface. The only question in my mind is if there is a "ridge" around the periphery of the letters. If not, what is the image seen on so many heat sinks?

As a footnote for Bill, I know of one source for plastic deformation on the copper: My brother (also an engineer ). He bought himself an Asus mobo, XP2100, and Thermaltake heat sink. He didn't use compound ("I thought that was only for the retail heat sink") and slid the heat sink after fastening the clip. It was so bad that he sheared a couple of the rubber pads off the chip. When he brought the system to me to diagnose, the slivers of copper and missing rubber made it pretty easy. He actually had a couple of furrows in the base of the heat sink from the edges of the die. Definitely a little plastic deformation going on there.

[BillA]

Dave, are we talking about the same thing ?

yes, the silicon die is MUCH harder than the copper
yes, the edges of the die and of the lettering are quite sharp
yes, transverse movement (with some angularity - the sheared pads ?) will cut into the copper

there could be a ridge, I don't - with a sample of 2 only - have enough data to say
but I have not seen it (yet ?)

but this is not my question; to restate as clearly as I am able:

I do not understand how the compressive load even approximating the AMD spec, or rather well above it, can result in the yielding of the copper in contact with the die face.

yes Ben
I can understand staining/discoloration, I too see that all the time
and it will also vary with the grease used AND the time interval

-> I also see hydraulic edge-cutting from the influx of grease when the closely pressed substrates are separated, sucking in the 'bead' of grease around the die perimeter previously extruded
(this seems to be a characteristic of Cooling Flow having boron nitride particles, and quite a number of mountings are needed before the little erosion channels begin to appear - yes, they lap out gradually, they are real)

no desire for contention here, lets not get all het up
just workin' on 'plastic deformation'

[gmat]

Just a theory: say, if the core surface is not perfect, it acts like a lot of little 'needles' that apply a very localized pressure against copper. That means, both surfaces get more or less 'into' each other, to a point which amounts to the sum of irregularities heights divided by more than 2 (ie surface irregularities of HS and those of CPU core).
If as i suspect the CPU core material is much harder than copper (it's more like glass), and since most people (including myself) wiggle the HS/waterblock just a bit *after* mounting to squeeze more TIM out, i think then that the CPU core makes its way *into* the copper base... (for a short distance indeed)

And yes a little polishing makes that imprint disappear.

(edit) ah Bill our posts crossed each other

[nicozeg]

I managed to take this picture of a dead Duron I have.

It still dont answer what is carved or rised, but at least it can be seen the laser spot size and edge irregularity.

[BillA]

gmat
I believe you are quite correct,
and I suspect (as stated) that the AMD die surface is expressly designed to promote substrate-to-substrate contact
(as does that Intel spire - in a much cruder fashon, sitting on a sharp post !)

perhaps my focus is misplaced ?
there should, with a FLAT bp, be some very localized deformation over the entire die contact area
- lotsa little tiny indentations (anyone got a good 'scope to look at this - N8 ??)

MY handicap in this area is that I work with 2 polished and flat substrates having the same hardness
so only on my computers are the 'real' conditions apparent

some loose terminology gettin' in the way here

[bigben2k]

I found an interesting link:
http://info.lu.farmingdale.edu/depts...formation.html

It leads me to think that the deformation is a slip "since this requires the least amount of energy".

Here's a 3 page PDF:
http://alspubs.lbl.gov/AbstractManag...oads/01114.pdf

on grain direction under stress, and the slip effect.

I haven't been able to find any references (online) to a measurement of the necessary forces, but expert Googling should do it.

[BillA]

no Ben, that is not an issue here

incidently, to try to preclude misunderstanding
I am using yield and deformation similarly
(like head loss and pressure drop)

[myv65]

@Bill,

Yeah, I'm pretty sure we're talking about the same thing (with the exception of my brother's moronic, ham-fisted heat sink mounting story). That picture of nicozeg's is pretty darn good and I don't think it takes much of a leap to "hunch" that the ridges resulting from the laser are significantly larger than the Ra of the remaining die face.

I'm sticking to my original conclusion. Namely that the effective surface area amounts to little more than the "outline" of the letters on such a die until those letters begin working their imprint into the copper. Only then does the peak pressure drop below the yield strength of copper. Granted, we're talking on the order of microns here (perhaps less), so the indentations in the copper should be easily removable via lapping.

If this is indeed the case, one would expect a central spring clip to be slightly more prone to indenting than a pure four-bolt heat sink due to the deflected shape of the baseplate. And in addition to the earlier time aspect I mentioned, there is also the factor of extruding excess TIM before the ridges could indent the copper.

[Cathar]

Am about to jump on a plane. Just some anecdotal experience to report. With regards to the imprint. My experience is that it is removable by using a jeweller's rouge (ala ~6000-10000 grit equivalent). I haven't been able to remove it using something that's completely non-abrasive. I believe myv65/Dave is correct about the ridged edges, but the issue here is that of size. To take just a jeweller's rouge to remove the imprint in the space of seconds, we're talking some extremely small imprints going on here (sub-micron's?)

[deeppow]

BillA, interesting! I was wondering how you plan on using the information you gather.

I understand that flatness is one effect that influences contact resistance between the cpu and hs but there are several factors that are typically recognized including surface roughness and contact pressure. I'm interested in if and how you see intergrating your measurements into a possible representation of contact behavior that will include the influence of a TIM material.

A good paper that developed a lot of the basic theory for contact resistance between two surfaces is by A. Mujumdar and C. L. Tien, “Fractal Network Model for Contact Conductance,” ASME J. Heat Transfer Vol. 113, pp. 516-525 (1991). There are others also.

Keep up the good work!

[BillA]

deeppow
for the moment just gathering and looking
then I'll hope to achieve some understanding (of what I'm 'looking' at)
then - hopefully - to an assessment of influence
etc etc
but I will always be handicapped by a lack of means to quantify roughness

I am used to N8's superlative stuff, he works in the field and has the knowledge and equip
but today I saw a pip, a prototype wb



the CPU contact area is about that darker patch in the center

truly FLAT, amazing for a commercial product (in 'our' industry)
sorry. no name available - yet

[UnloadeD]

Looks nice and flat, did you ask how they did it? Also do you check them after you've mounted and tested them? I'm really curious if a block needs to be relapped after every use. Also I have a question about your comments re:Cooling Flow. Did you mean there is some abrasive characteristics to it, that does a bit of eroding to the baseplate after removing multiple times? Thats been my compound of choice for a while, I've got AS and Shin Etsu but I prefer the CF for its ease of apllication and removal, and all seem to give similar results for me.

peace.
unloaded