New Discovery with Conroe IHS's!

[DrJay]

Another sharp edged retail here........its a 6400 though...

[Sadolf]

If the CPU wasn't in contact w/ the HSF, then it would throttle ALL the time. Not just at load. There just isn't enough metal in the IHS to dissipate the heat to keep it from throttling.

If you THINK your IHS is concave, take the bloody thing out and CHECK IT.

First thing I did when I got my E6600. That sucker is PERFECTLY FLAT.

You know of any good testing methods of flatness? Using a permanent marker pen to color the IHS, then drag it over a paper lying on a flat surface (like a glas table?) to se where the color wears of or not?

[derektm]

You know of any good testing methods of flatness? Using a permanent marker pen to color the IHS, then drag it over a paper lying on a flat surface (like a glas table?) to se where the color wears of or not?

Use a very flat object, like a ruler or a razor blade, put it from corner to corner, side to side, see if the center dips.

[Hyperformance]

I ordered a oem 6700 from clubit got it a week and a half ago and it had the old style 775 smoothed edge ihs. maybe they switched the design because they were having problems with so many of the ihs coming out concaved. sounds like the people whove gotten the raised ihs have all had relativly flat ihs on them.

[nettwerk2006]

The E6600 i ordered from ClubIT has sharp edges, but some of the edges are bent up pretty good too

My digital camera kinda sucks, but here are some pictures of the contact my waterblock makes with IHS after being screwed down as tight as possible.
The outside doesn't make very good contact, but the center seems ok. The bottom right corner I noticed was bent before I applied the TIM, and clearly shows its lack of contact afterwards.
heatsink-001.jpg

heatsink-002.jpg

[Stigma]

I just want to point out one common misconception:

Touching your heatsink cannot possibly tell you if your core has good contact with your heatsink or not. The heat-sink will allways be the same temperature no matter how much or little contact the heatsink has with the core.

Why? Because (assuming that the vast majority of the heat exits via the heatsink), the heat that the core generates has to escape. If there is poor contact, the heat in the core will increase until the balance point is achieved (at the point where heat generated = heat transfered away).

The balance point is allways achieved no matter if there is little or lots of contact, thus the heatsink will allways be the same temp because it is the "only" place where heat escapes from. in other words, if the CPU produces 60watts of heat, the heatsink has to radiate 60 watts of heat out. If it dosn't, then heat builds up until it reaches the point where it does. SO any way you turn it, the heatsink will radiate the same amount of energy away, and as long as the heatsink is the same, the math will be the same regardless of good contact. energy transfered to heatsink / dissapation capacity of heatsink = temperature of heatsink.

(Obviously, heat escapes from other palces too, like the back of the board, but in such small amounts that I don't consider them part of the equation for this example).

So, in conclusion: While touching the heatsink can get you a general idea of your CPUs temperature (relative to the power of your heatsink solution that is), it cannot tell you if there is good surface contact or not. The only way to tell something like that is to compare measurements directly from the core with the temps on the heatsink. The larger the delta, the worse thermal contact you have.

-Stigma

[Sadolf]

Credit card edge comes to mind.

All my credit cards are well used and has bumpy edges Might not be the best thing to measure with or?

[fornowagain]

All my credit cards are well used and has bumpy edges Might not be the best thing to measure with or?

Stanley blade?

[Eastcoasthandle]

The only option for everyone with C2D is to do a contact test
-tast some TC and spread it over the IHS leaving a thin layer ontop of the ISH
-Place your HS ontop and clamp it down as normal
-wait 2 minutes
-remove the HS carefully, and see what kind of imprint is left on the HS

Do you get full TC, square print on the HS contact surface or is part of that square imprint missing? If so, you can clearly see what portion is not making contact with the IHS.

The reason why I don't use the dollop method is because it's not 100% accurate when determining contact area. Sure it would spread but it doesn't always indicate if the gaps are caused by no contact or spreading of the TS.

[eva2000]

interesting finding all my intel lga775 cpus i've had are bevel not sharp

550
920D
955XE
2x E6600 B1 - flat as a board as i lapped them and no diif from before they were lapped in terms of ocs or temps heh

[Sadolf]

Well, it seems I have the "bevel edge" type of IHS, and using a credit card, it seems like it's flat and nice. But when I color it with a felt-pen, and then rub the IHS on a paper (actually a Post-It so it would lie still and not wrinkle) on a glass table, you see something else. The color is only been rubbed away on the edges of the IHS. So this I think explains my heat problems, bad contact between the IHS and the heatsink.

As I can see it i have 3 alternatives in order to fix this problem:
1 - use my warranty if possible, to get a new one.
2 - sanding it flat (but losing my warranty?)
3 - Fill the "void" with some super heat conductive cooling paste like Arctic Silver 5, and hope it does the work?

[party animal]

Yeah, that does loook pretty bad. I definately wouldnt choose number 3 as its just a waste of thermal paste and the thickness wouldnt be good for heat transfer. If you know you can lap it nicely, and take a chance that yours is a good clocker, then go for it. If not, its up to you then.

[Levish]

pictures of flatness should be level with the IHS as if you were looking into the horizon

credit cards, particularly used ones aren't flat

my e6400 SL959 - L624A851 has a sharp cornered ihs

and from Eyeballing it if anything it looks like the corners and edges are lower than the rest of the ihs so this one should have decent contact

[Kango]

I received my E6600 (Lot L626) yesterday. I examined it closely to see if it had been used (it's an OEM version and I checked the PCB pads) and whether it was flat or not (this thread made me worry a bit). My CPU had the sharp edges and corners and is very flat. I even applied AS5 evenly across the top and mounted the HSF. After about 5 minutes, I pulled the HSF and examined the spread of AS5. Every portion of the IHS was touched by the HSF and the AS5 was spread evenly. I'm very happy about this, of course. (Sorry, no pictures. My camera is elsewhere, right now.)

I am wondering if this sharp edge IHS is a recent change in the manufacturing process or not? Maybe this is the way Intel is letting us know (indirectly, of course) that there was a problem with IHS flatness and that it has been addressed. Did all this start during a particular work week (26 in my case)?

Kango

[SlackeR]

E6600 retail L624A816 - three numbers lower than derektm's
Sharp edges, concave? No. Convex if anything
Though i am waiting for an XBX and two stick of Team Xtreem in order to test it.

[an0nym0us]

so yeah this is a little off topic, but i just took my 820 out of its box and looked at it, i used a stick of ram to see if it was flat or not and it turns out my 820 is convex, not concave. are there a lot of intels out there with convex ihs's? all of my 775's are beveled as well, 520, 550, 820 and E6300.

[Sadolf]

Yeah, that does loook pretty bad. I definately wouldnt choose number 3 as its just a waste of thermal paste and the thickness wouldnt be good for heat transfer. If you know you can lap it nicely, and take a chance that yours is a good clocker, then go for it. If not, its up to you then.

Looks like I can have a new one on my warranty when they get more of them in stock... !!11

[DTU_XaVier]

If anything, convex is better than concave, to a point...
Intel probably just changed their manufacturing methods, probably have been listening to the customers again ()...
Will have my E6600 in about 2 weeks, should make for some interesting tests

Best Regards

[Revv23]

you guys dont think intel actually makes the IHS's do you?

They make cpu's, they dont manufacture peices of metal.

They go to the lowest bidder, and im sure they jump from company to company trying to get the cheapest ihs' possible... who would expect them not too?

Same as with AMD, they have serveral very different stock HSF designs, and im pretty sure intel does as well. (and im talking about hsf's for the same cpu, like i had 3 opteron 146, each with an alu hsf, and all 3 were different.)

[Eastcoasthandle]

neither convex or concave are good for IHS. You have to lap them if it's either one or the other.

[zoob]

Slightly convex >> any degree of concavity.

If it's convex, at least it guarantees that at least a portion of where the IHS contacts the core will also be in contact with the HS, eliminating any insulation layer (air or too much thermal paste)

[SlackeR]

Mine seems really good despite the slight convexity. Really flat at the middle and "sloping" half a milimeter perhaps, if that much at all towards the edge. If the tools i have used for looking are somewhat straight. A two days ld credit card (bad habit of wearing those out ) and a stick of Mushkin Lvl2 PC2700.

[Revv23]

i dont think ive ever had a heatspreader that was completly flat.

even my 1.8a wasnt flat, nor y 2.8c, and my 3x opterons and x23 3800, none were perfect, you want perfect, sand.

[fornowagain]

If the tools i have used for looking are somewhat straight. A two days ld credit card (bad habit of wearing those out ) and a stick of Mushkin Lvl2 PC2700.

Use a utility blade. Straighter than a credit card

[derektm]

Use a utility blade. Straighter than a credit card

This is a nice blade