Round 4: Why the goo failed.. BONUS: core2 QX lapping!

[cauzomb]

It turns out that the core of the CPU and the bottom of the heatsink are both concave!!

The cpu was cupped bigtime; the edges were raised higher than the core; and the base of the heatsink was cupped slightly, so it compounded the gap between the core and the base of the heatsink.

Alot of sites recommend applying a line of thermal goop down the core of the cpu; or a small blob of cpu goo about (uncooked grain of rice) sized, right in the middle of the core, but I've had best results using the spatula/spreading technique, except this time.. So now I know why they recommend applying a glob in the center for core2's without spreading, the heat spreader is warped up like a taco'd rim.. I tried both methods of the glob in the center technique, both rounds of applications didn't do anything for my temps, I was getting the ocassional game crashing back into windows; hence this thing starting with round 3.. In round three I decided to spread a thin layer across the whole cpu; I'm not a fan of using alot of cpu Goo as it tends to act like an insulator when you get too much on there.. I thought the previous two attempts were hot due to using too much cpu goo... Guess I was wrong... Round three's application got me an actual BSOD

So!! On to round 4... I cleaned off the CPU and heatsink really well, with alcohol and a few tissues.. Taking a look down the top of the cpu while I held my straight edge across the top, holding it up to a light, I could see light shining through some pretty big gaps in the middle and right near the edges, but no gaps at the edges.. Sorry no pics of that... The unflat top really needed lapping!

I'm supprised that intel would let something like this ship! especially considering the price tag, it makes me wonder how many more expensive cpu's are running around out there that need a good lapping.

I gathered up my lapping junk and started working on the heatsink first. Since the mounting tabs hang down past the base of the heatsink, I can't just put the thing on a flat lap and sand away at it.. So I'll be moving the sand paper over the heatsink base, instead of the other way around.. I used a "flat-lapped" celly core as my sand paper backing

So I went on lapping away at the heatsink untill I had scratched up the whole surface, IE, there were no low spots left where sandpaper wasn't touching... "might sound noobish" but I do a pretty good job at lapping....

Moving up to 1000grit and finishing with 2000 grit "you don't really need to go over 800 grit but I have 2000 grit sand paper, so I used it.

Finished off the lapping job by wiping the surface down with some denatured and some lint free tissue's. Cleaned up a little bit and started on the CPU..

I'm including pics of these so you can see how bad it was.. it's not like an old mmx heatsink with the big cut marks n stuff, but it was pretty bad...

If you got high temps on your core2 or quad core; it could be due to the surface not being flat enough to get the heatsink properly interfaced..

Even if you got enough goo in there to close the gap, the goo is only supposed to fill "tiny" voids and surface scratches. Not an entire valley!

[cauzomb]

More pics of the lapped core:

last two pics are showing the mirror like finish as a result of using 1000 grit and 2000 grit sand paper.

[cauzomb]

Round 4 trial fit, 90 percent better fit than last time........ If it were a no compramize situation, I'd be lapping the CPU installed in a motherboard, so that it is flat when clamped down!!!! sort of like cylinder honing torque plates. There's visible deflection in the heatspreader where it get's pressed on by the lga775 socket's clamp.. I can't get away with an ultra thin application of as5 the way it is, but I did re-work the heatsink base very slightly on the edges that had the most squish-out.

I applied a thicker layer of as5 this time..

It's been idling at 49c with windows xp runing; used to idle at 54+ in windows.. Idles at 37c in the bios; used to idle at 48 to 50c in the bios.

Gotta do some load testing... to be continued.......

[Gizmokid2005]

Thats ridiculous! Intel should be shot for that...just another good reason to stick with amd

[cauzomb]

I can't say I've had any problems with my 939 chips; or the stock coolers...., It's been so long that I don't remember what my 2500+ looks like, I think it's a bare chip? but it runs at 3200+ with stock everything.. I run my 3400+
IHS 939 chip at 2."something" ghz on the stock heatsink, stock voltages; without lapping. I can't even run stocker's without crashing.. I'm contemplating RMA'ing the board and CPU and maybe filing consumer report's about the crappy performance, flatness and possible damage from bending, it was bending before I lapped it, I could tell by the contact patch, lapping it reduced the effect's but it's still NOT a propper interface.. I don't care if they think it's "lapped" or voided waranty, I'm UBER IRATE.. I'm not going to have to pay for a computer for 3 years and not get to use it.. The lga775 board may have damaged the chip internally by "bending" the substrate that the chip is fixed to. This is the worst possible CPU mounting solution that I've ever seen. With how things are going and how processors are "obsolete" in 5 months I'd rather not have my chip permanently soldered to the motherboard, but the lga removable clip from [edit] "ECS" on their 680i boards is total crap... I've seen three of their 775 boards with unflat mountings and now this one is bending a 900 dollar processor.... [edit] sorry foxconn, it's actually an ECS board, Stay away from ECS/elitegroup 775lga boards untill they own up to their socket retainer issue. Although I think it might be a board MFG'd by nvidia cause it seems to be identical to a couple other MFG's boards. It's the black 680i board

[Red Squirrel]

Good thing the new cpus are flat now though, they used to expose the registers and stuff, was harder to put paste then and sometimes you'd end up with a mess.

But yeah the concave thing is not too good lol.

[cauzomb]

Good thing the new cpus are flat now though, they used to expose the registers and stuff, was harder to put paste then and sometimes you'd end up with a mess.

But yeah the concave thing is not too good lol.

CONCAVE=not good for sure.... The heat spreader is less troublesome than direct die cooling solutions... Although, I've never "fried" a chip, I do remember crunching an athlon 1200 under an alpha PAL heatsink. I wonder what's hapening to my silicon under the heatspreader though! is the substrate bending under the pressure of the socket CPU retainer? or just the edges of the IHS deflecting over some kind of softer sealant?