finished lapping my Q6600 (with pics and results)

[graysky]

Originally posted by: BoboKattsomeone had said... if you get water on the CPU... you're sunk. In other words when you are rubbing it against the sheet of sand paper, which will be mildly wet on the surface, is there any special way to hold the CPU or shield it or prevent any moisture at all from getting on the other surface. I am just worried that when working no matter how careful you are you might get some sort of debris or filings or anything on the exposed critical part that would be facing up.

Any suggestions? Or it really is not an issue as long as you are careful?

That statement is simply incorrect. There is a slot on one face of your quad that is rumored to go into the cores. You should take care to keep moisture out of it. The way I do it is to moisten the sandpaper using a spray bottle, blot of excess water (no drops should be present) and lap. Have a moistened (not damp or wet) paper towel ready to clean off the CPU when you are finished with the lap. You'll see some copper paste (fine particles of metal suspended in the water) come off. Wipe down the surface gently and repeat the lapping.

The key is to take your time and to pay attention to what you're doing; that's the best advice to avoid getting water in that slot. I suppose you can tape it off too, but that's a bit extreme IMO.

 

[graysky]

Originally posted by: 21stHermit

Originally posted by: graysky
Temp Decrease Data Graph

From this graph, my take is that the two cores inside the package are bonded to the IHS differently. One having a better thermal connection than the other.

I realize this is a 6-mo old thread and may have already been discussed.

Thanks for the post, best explanation of lapping I've seen yet.

Hermit

Actually, I lapped the chip one additional time to remove the slightly raised edges (4 sheets of 1000 grit paper). That dropped the temp differences to about 1 °C when loaded on prime95. See the final pic I linked in the thread

Also know that one efficient apps will equally load all 4 cores.

 

[MadScientist]

Originally posted by: panfist
I might be wrong but from my study of optics in university physics, surfaces that are reflective actually have zillions of tiny scratches in them. Hence they are not "smooth," and that is how you are able to achieve reflectiveness with something coarse like sandpaper.

By lapping your IHS you flatten it but in terms of the final grit you use, I'm sure there is an optimal grit that is a function of the type of thermal paste you use. Essentially you are forming little grooves in order for the thermal paste to flow into and form a proper layer. It would be interesting to see if anyone did an experiment where they took a few identical heat sinks and processors and tested different grit finishes.

The question where to stop when lapping a heatsink or CPU has been discussed many times here and on other forums. On one side you have the "It's got to have a mirror finish, so lap it up to 2000 grit sandpaper" and on the other the "Get it flat and stop at 600 - 800 grit". I came across this article awhile back at Overclockers. Link The article is old but the theory, in my opinion, is still valid for today's heatsinks and CPU.
It makes sense to me. Of course opinions are like you know what, everyone has one. Personally I'm in the "lap til it's flat" group and stop at 800 grit.

Good job Graysky. My lapping technique is very similar to yours. Glad to see you didn't go for super shiny and stopped at 1000 grit. Nice excel data graph. One thing you didn't mention and I've found is the most important variable that effects the CPU temp is the ambient (room) temp. This must be kept constant to achieve accurate CPU temp comparison results.

 

[rodrigu3]

I concur, lapped both IHS and HS just till flat (I only did a couple passes on 1000grit, maybe half a sheet, to get rid of some large scratches) ---> running an e2180 @ 3Ghz and temps never go above 48C under prime / orthos

Edit: lapping my ultra 120 extreme was a pain in the ass because it was so heavy and skipped a lot if I went too quickly

 

[21stHermit]

Originally posted by: graysky
Actually, I lapped the chip one additional time to remove the slightly raised edges (4 sheets of 1000 grit paper). That dropped the temp differences to about 1 °C when loaded on prime95. See the final pic I linked in the thread

In your Coretemp screenshot(s) I see a Tjunction temp of 100C, always exactly 100C. Just what is this?
After lapping, are the surfaces so true that Artic Silver is not needed?

Thanks
Hermit

 

[MadScientist]

Originally posted by: rodrigu3
Edit: lapping my ultra 120 extreme was a pain in the ass because it was so heavy and skipped a lot if I went too quickly

You can prevent the skipping by first putting a bevel on the edges of the heatsink before starting to lap. See step 2 of EasyPC Kits Instructions

 

[formulav8]

Originally posted by: 21stHermit

Originally posted by: graysky
Actually, I lapped the chip one additional time to remove the slightly raised edges (4 sheets of 1000 grit paper). That dropped the temp differences to about 1 °C when loaded on prime95. See the final pic I linked in the thread

In your Coretemp screenshot(s) I see a Tjunction temp of 100C, always exactly 100C. Just what is this?
After lapping, are the surfaces so true that Artic Silver is not needed?

Thanks
Hermit

I'm thinking it is similar to AMDs MaxT Case rating. IIRC it is the rated max temp between the heatspreader and core or inside the core?


Edit: Here is a link that has alot of temp info and such


Jason

 

[graysky]

Originally posted by: 21stHermit

Originally posted by: graysky
Actually, I lapped the chip one additional time to remove the slightly raised edges (4 sheets of 1000 grit paper). That dropped the temp differences to about 1 °C when loaded on prime95. See the final pic I linked in the thread

In your Coretemp screenshot(s) I see a Tjunction temp of 100C, always exactly 100C. Just what is this?
After lapping, are the surfaces so true that Artic Silver is not needed?

Thanks
Hermit

It's a constant. The way the DTS work is to subtract their value from this constant:

100 - value = true temp

C2Qs have a tj of 100 whereas most C2Ds have a tj of 85. It's actually nothing you need to use and never changes. Don't confuse it with the "max temp" or anything, again, it's just for the math.

 

[graysky]

Originally posted by: MadScientistOne thing you didn't mention and I've found is the most important variable that effects the CPU temp is the ambient (room) temp. This must be kept constant to achieve accurate CPU temp comparison results.

Very true... I wrote a post centering around the effect room temp has on load temps.

 

[21stHermit]

Originally posted by: graysky

Originally posted by: 21stHermit

Originally posted by: graysky
Actually, I lapped the chip one additional time to remove the slightly raised edges (4 sheets of 1000 grit paper). That dropped the temp differences to about 1 °C when loaded on prime95. See the final pic I linked in the thread

In your Coretemp screenshot(s) I see a Tjunction temp of 100C, always exactly 100C. Just what is this?
After lapping, are the surfaces so true that Artic Silver is not needed?
Thanks
Hermit

It's a constant. The way the DTS work is to subtract their value from this constant:

100 - value = true temp

C2Qs have a tj of 100 whereas most C2Ds have a tj of 85. It's actually nothing you need to use and never changes. Don't confuse it with the "max temp" or anything, again, it's just for the math.

The position of the Tj number had me wondering, now I know.

Do you still need Artic Silver?

Thanks

 

[myocardia]

Originally posted by: 21stHermit
Do you still need Artic Silver?

Yes, but you'll need considerably less of it, once both the IHS and the heatsink are lapped.

 

[graysky]

this is true... a VERY thin line will do it.

 

[aigomorla]

Originally posted by: Kromis
I don't even know what lapping is so...

I'm not even gonna ask!

Its trying to make the top of the cpu cover, ihs, flat as possible. This allows more even contact on the entire chip.

Great results.

Heres one of my lap jobs which usually impresses people

http://i125.photobucket.com/al...aigomorla/IMG_0703.jpg


And yes even with my kind of lap job, you should still apply a bit of AS5, or some form of TIM.

Microbubbles is what you should be getting rid of. The more flat your ihs is, the less tim you need to use however.

 

[jonmcc33]

LOL @ all the lapping. Why don't you guys just rip the IHS off instead?

 

[aigomorla]

Originally posted by: jonmcc33
LOL @ all the lapping. Why don't you guys just rip the IHS off instead?

the ihs is soddered onto the die on all C2D and C2Q chips.

This makes removal very trickey at best, and if you do it, you risk ripping the die off along with the IHS.

Not really worth the gamble IMO.


So the safest way to get a bit of a performance increase would be to LAP it.

 

[justinburton]

Great job on the lapping, the only bad part is you will have a very hard time reselling this CPU. Try and explain to the potential buyer why the CPU ID is shaved off the IHS.

 

[aigomorla]

Originally posted by: justinburton
Great job on the lapping, the only bad part is you will have a very hard time reselling this CPU. Try and explain to the potential buyer why the CPU ID is shaved off the IHS.

very true, however i dont have problems selling my stuff because everyone who buys them usually know there hand picked by me.

But usually all one needs to do is pop a picture of the bios loadup, and then gaurentee the chip to be the model number displayed on the bootup.

Also people like IHS shaved cpu's because it saves them the job of lapping.

 

[ArchAngel777]

Originally posted by: aigomorla

Originally posted by: justinburton
Great job on the lapping, the only bad part is you will have a very hard time reselling this CPU. Try and explain to the potential buyer why the CPU ID is shaved off the IHS.

very true, however i dont have problems selling my stuff because everyone who buys them usually know there hand picked by me.

But usually all one needs to do is pop a picture of the bios loadup, and then gaurentee the chip to be the model number displayed on the bootup.

Also people like IHS shaved cpu's because it saves them the job of lapping.

Not a big deal for me either... My old stuff generally finds its way into my dad's computer system and if not, then it is normally sold to any of my friends who love a nice discount on my old parts, which are still twice as fast as their current parts.

Plus, when I sell my stuff, I am pretty generious overall... I don't try and get the most I can, I dunno, I guess I have a hard time doing that. I plan on selling my dad my Q6600 Lapped for $100 or so, hell, i'd will give it to him, but he will refuse and ask me what it is worth, then I give him some low BS number.

Anyway, I also had my co-worker tell me that people do prefer lapped chips, even on Ebay because it saves them the work and potential danger of doing it on their own... That lines up with what Aigo just said to.

 

[johnnyjohnson]

Originally posted by: justinburton
Great job on the lapping, the only bad part is you will have a very hard time reselling this CPU. Try and explain to the potential buyer why the CPU ID is shaved off the IHS.

I didn't hesitate to lap my e4500. It's an exceptional overclocker and I wanted to maximize my cooling. At present I have it running at 3.41Ghz with speedstepping on. Doesn't even break 1.35V under load. I don't think I'll have trouble reselling it because chances are the person who would buy a rare chip like this is the type who would lap it anyway.

 

[SickBeast]

Just as a suggestion...I'll bet a bit of Brasso on the HSF would be great...that stuff can buff scratches out of an iPod and make it look NEW. :light:

I'm just not sure I'd use it on the core...liquid+core=death...

 

[jonmcc33]

Originally posted by: aigomorla

Originally posted by: jonmcc33
LOL @ all the lapping. Why don't you guys just rip the IHS off instead?

the ihs is soddered onto the die on all C2D and C2Q chips.

This makes removal very trickey at best, and if you do it, you risk ripping the die off along with the IHS.

Not really worth the gamble IMO.

So the safest way to get a bit of a performance increase would be to LAP it.

Soddered to the die? I recall once seeing someone cut the IHS off actually.

So if you lap it does it void the warranty?

 

[SickBeast]

Originally posted by: jonmcc33

Originally posted by: aigomorla

Originally posted by: jonmcc33
LOL @ all the lapping. Why don't you guys just rip the IHS off instead?

the ihs is soddered onto the die on all C2D and C2Q chips.

This makes removal very trickey at best, and if you do it, you risk ripping the die off along with the IHS.

Not really worth the gamble IMO.

So the safest way to get a bit of a performance increase would be to LAP it.

Soddered to the die? I recall once seeing someone cut the IHS off actually.

So if you lap it does it void the warranty?

Some intel IHS are soldered on, and some aren't. It depends on the chip AFAIK.

Besides, it's too risky. If your blade goes in too far, bye-bye CPU. It's not like the core is titanium-plated (although it should be).

 

[graysky]

yeah.. not worth it... if you're in need for extreme cooling don't try that, just lap the chip and use a liquid system.

 

[jonmcc33]

Originally posted by: SickBeast
Some intel IHS are soldered on, and some aren't. It depends on the chip AFAIK.

Besides, it's too risky. If your blade goes in too far, bye-bye CPU. It's not like the core is titanium-plated (although it should be).

http://www.legitreviews.com/article/402/2/

Hmmmm, seems you are right. I stand corrected. Seems like they killed it too.

 

[Tweakin]

Originally posted by: graysky
True... I don't want to jinx myself by saying this, but have you or do you know anyone who needed to invoke a CPU warranty? In the past, I bought "white box" CPUs because they were cheaper and didn't come with the factory HSF that I usually got rid of anyway.

Then I'm your first. As it turns out my X2 4200+ has a bad IMC, and since I lapped it I can not send it in for replacement.

Guess you could say I took one for the team