Core 2 Duo heatspreaders

[lopri]

Originally posted by: neuralnut
Had a lot of temperature problems with my E6400, I'd reseated the HSF a couple of times and got large variations in temps. Finally pulled the CPU out and had a close look at it, basically a straightedge touched the IHS at the edges but not in the centre.

I lapped it for a short time, and as you can see from the enclosed photo, the lapping hardly touched the center of the IHS - all the meat was taken off near the edges.

Put it back in and it ran 8 degrees C cooler under load. So the lesson is, if your conroe runs hot, check the IHS, some are not nearly flat enough.

That'd the weirdest IHS I've ever seen. Do you have a picture of the CPU before IHS removed?

 

[Baked]

Originally posted by: cmdrdredd
snip

The right side looks concaved still. Damn, you must have really Fed up IHS. Nice temp improvement.

 

[CupCak3]

Originally posted by: cmdrdredd
Ok I'm all done. fully lapped C2D. Pictures below with my system specs and before+After temps. I included a screenshot of CPU-Z and core-temp too

Antec P180 case with the intake and rear exhaust fan on high top exhaust on low
Seasonic S12 600watt PSU
Asus P5B Deluxe at 400Mhz FSB
Corsair XMS2 6400 at 5-5-5-12 w/2.1v
C2D E6400 @ 3.2Ghz at 1.34v
Zalman CNPS9500LED

Before lapping I had a concave IHS and my idle temps at these speeds was almost 55c. Load was up past 65c all the time, and it hit 70c at one point. Not good. After lapping I hot load at 60c and idle around 47-48c. This is a very good result with no other changes.

Typical IHS from a P4 for refrence

My fully lapped IHS from my E6400

Notice the nice reflective surface

Screenshot of CPU-Z etc

thanks for all the info... this will definately help1!!

 

[neuralnut]

I only lapped it until it was flat, tested with the edge of a razor blade and a bright light. This was pretty much as shown in the photo, only down to the copper at the edges.

When lapping it is very easy to take more off the edges than the center so you end up with a convex IHS rather than a concave one. Checking for flatness and stopping at that point is important.

I was seeing temps of about 48 - 50 degrees C under load (ambient temps are quite low here) and this dropped to about 40 degrees. However my absolute temps are not that relevant as you don't know my ambient, overclock or voltages. Temperature deltas are much more useful.

I also have the advantage that I have an identical machine to use as a reference. The two sit side by side and one always ran significantly hotter than the other. Now that the CPU has been lapped this machine now runs 2 degrees cooler under load than my other machine.

 

[cmdrdredd]

Originally posted by: Baked

Originally posted by: cmdrdredd
snip

The right side looks concaved still. Damn, you must have really Fed up IHS. Nice temp improvement.

No that's my shadow because I had to turn off the flash on my camera to avoid getting an overbright image. That darkened area is not there when you look at it normally.

 

[cmdrdredd]

just a small update. After some debating and thinking I decided to drop down to 400x7 (2.8Ghz) and a lower voltage. Running 1.29 according to PC Probe and 1.288 by CPU-Z.

Less heat buildup and not all that much of a drop in performance on the whole. It's still one hell of a good CPU at 2.8Ghz. Maybe if I had something better than a CNPS9500 I could do higher than 3.2

 

[dopee123]

So, has anyone sent back their CPU as a result of their crappy IHS, or did everyone resort to lapping. I don't want to void my warrantee so I'm sending it back. I would like if anyone had experience replacing stuff directly by Intel to please share . I need some reassurance before I send it in.

 

[phile]

I seriously doubt there's a case for a warranty exchange unless the CPU can be shown to be running at excessively high temps, at stock. Unfortunately, we are not technically entitled to any specific performance when overclocking.

-phil

 

[TheRyuu]

Since it's an IHS there's really no harm it doing it. It's not like you can damage the processor or die.
And it seems really hard to screw up.

Seems a lot easier then taking OFF the IHS on my Opteron.

 

[Conky]

Originally posted by: phile
I seriously doubt there's a case for a warranty exchange unless the CPU can be shown to be running at excessively high temps, at stock. Unfortunately, we are not technically entitled to any specific performance when overclocking.

-phil

No kidding, if it only run's at it's stock speed then how can you complain? :laugh:

 

[cmdrdredd]

Originally posted by: Beachboy

Originally posted by: phile
I seriously doubt there's a case for a warranty exchange unless the CPU can be shown to be running at excessively high temps, at stock. Unfortunately, we are not technically entitled to any specific performance when overclocking.

-phil

No kidding, if it only run's at it's stock speed then how can you complain? :laugh:

exactly...at stock my E6400 was fine...no high temps and everything was good. When I overclocked I noticed the issues. So I investigated.

 

[Aluvus]

Originally posted by: sjm
Why does that linked guide mention elbow grease if none should be used?

http://en.wikipedia.org/wiki/Elbow_grease

 

[dasmokedog]

Thanks for the great info. I have a 6300 running at 3.5GHz and my idle temp is 51c (Core Temp) with load hitting 71 - 74. I am not too worried since the TCaseMax is 85c, BUT I will now lap my IHS and see what change I can get.

Thanks again!!!

P.S. Yes, I am Prime95 (on each core) stable for 24 hours now!!!

 

[lopri]

Originally posted by: lopri

Originally posted by: neuralnut
Had a lot of temperature problems with my E6400, I'd reseated the HSF a couple of times and got large variations in temps. Finally pulled the CPU out and had a close look at it, basically a straightedge touched the IHS at the edges but not in the centre.

I lapped it for a short time, and as you can see from the enclosed photo, the lapping hardly touched the center of the IHS - all the meat was taken off near the edges.

Put it back in and it ran 8 degrees C cooler under load. So the lesson is, if your conroe runs hot, check the IHS, some are not nearly flat enough.

That'd the weirdest IHS I've ever seen. Do you have a picture of the CPU before IHS lapped?

Fixed. Sorry about typo. But seriously, can I see the picture of the IHS before lapping? If there is one.

 

[neuralnut]

Sorry, didn't think to take a photo until I got such a dramatic result in the first minute or so of lapping. Prior to lapping a straightedge was clearly touching the IHS only at the edges, but I couldn't estimate exactly how much the gap in the middle was. I can say that a reasonable amount of light shone through when held up to a lamp.

The thing that made me suspect the problem was when I first took the heatsink of (Zalman 9500) I could clearly see a rectanglular imprint on the arctic silver on the bottom of the HSF, but it clearly had not touched anywhere else. I had applied the AS using the thin spread method, and this clearly only works if your IHS is flat or convex. For a concave IHS it's a disaster.

When I reseated the HSF I used the single spot of AS method and the temps dropped a fair bit but were still high. I expected them to improve as the AS bedded down and the HSF settled down, but after a week of no improvement I pulled the HSF off again. This time there was the same rectangular imprint on the HSF base, but the single spot had only spread out to about a 10mm circle and was clearly still quite a thick layer. This was a clear indication that the IHS was never going to be good enough without some lapping.

I only used 600 grit to lap my CPU, but by the end the paper was pretty worn so was effectively a higher grade. I didn't go overboard on the finish, at the end of the day the most benefit comes from flatness, getting a high polish is not going to improve temperatures much more. With a lot of lapping you run the risk of making your IHS convex due to subtle presure on the edges of the CPU as you push it back and forth. This is going to reduce thermal efficiency more than the gains from the additional surface finish you are aiming for, so it is better to stop early than make things worse.