Tip for CPU Lapping

[BonzaiDuck]

I expect this to fall to the bottom of the post queue, and doubt there will be many replies.

Everyone is all antsy about lapping their CPU IHS heatspreader-cap, at least everyone who has posted here recently with their "concerns."

The retail processors from Intel come with a plastic protector cap covering the gold-dot conductor-array on the underside of the processor.

Just leave the plastic protector on the processor until you're finished lapping! [Or save a couple of those things, and use them for processors you choose to lap retroactive to initial installation.]

For people worried about grinding away too much of the IHS -- the nickel-plate is just a thin coating. Grind the suckers down to the point where most of the nickel is gone. You'll get a better temperature-drop that way.

 

[Phunk0ne]

The End Result

 

[SanDiegoPC]

Phunkyman I'm impressed!

 

[BonzaiDuck]

Yup. Did that, too. I should've started with 220-grit to save time, but used an old piece of 400-grit and found myself looking for the unused corner of the sandpaper -- getting impatient. Surprised at how shiny it was afterward. Also, I left a patch of nickel on it about a quarter-inch wide -- near one corner. That won't matter much, either.

 

[PCTC2]

I lost my processor cover.... oh well. I guess I'll have to buy a new QX9650 so I could get a new cover so I can lap my Q6600... Oh Well...

EDIT: Thanks for the tip BD.

 

[mrdavidpham]

wow thats a nice lap job!

 

[Phunk0ne]

Thank you

I used an old mirror, took it out of the plastic casing, for a nice smooth surface to put the grit paper on. I bought 2 of each grit level 200-400-600-800-1000-1500 and cut out 4 pieces of 4x7 inches (which was the size of the mirror) of each grit level. I used the wet lapping method, just a few sprinkles of water onto the grit paper. I did make sure that the paper was firmly attached to the mirror, I used ducktape on all 4 corners, and that the little opening on the side of the CPU was completely sealed off before starting the job. 10 strokes each time before turning 90 degrees and made sure I would use all of the grit paper I can. In the end, it took me about 2.5-3hrs to complete lapping the CPU. I used Brasso to get rid of all the remaining impurities left behind from lapping and finished it off by cleaning it with arctic Clean #2.

OT: BTW Brasso is perfect to clean your dirty yellow crusted car headlights

 

[BonzaiDuck]

PhunkONe -- I admire your attention to methodology.

I just don't think it's necessary. I can hold down a piece of sandpaper on a half-inch-thick glass coffee-table surface. I stopped worrying about the water and impurities getting under the IHS and into the works. I blasted the CPU with canned-air after I was done.

Like I said -- there was a quarter-inch spot of nickel left on the IHS close to one corner. But -- for just using "already-been-chewed" 400- and 600-grit sandpaper, I don't think I spent 2.5 hours. It was more like 30 minutes to just over an hour.

Not as shiny as yours, but it seemed remarkably absent of significant scratches, 95% bare copper. I'll be interested in the temperature results in comparison to another similar processor I've installed without lapping -- same voltage setting, roughly the same FSB, etc.

But I'm confident there will be a marked difference between the two.

 

[Phunk0ne]

BD: the mirror shine IHS is completely unnecessary, As long as you have a flat smooth surface showing the copper you are good to go
But being me, I just could not stop there
I was afraid that the duirt during lapping would get inside the IHS through the little opening, so I taped it off to make sure (I am a natural walking disaster on two legs), but with patience and a some time, I made sure I would do this as "perfect" as I could. And as you see, the pictures show the end result.
I spent 89 cents on each grit paper and another ~3 bucks on Brasso. Over 21 bucks, and unnecessary, but to me personally worth it
I will most likely use the exact same method when I am going to sand down the Q9450.

 

[tshannon92]

I did the same as Phunky using the small picture frame glass and duct tape. I stopped at 1000 and then did the HS also and the load temps were truly amazingly better. By almost 10c but my AFP7 was terribly milled so I expected great results.

I bought another Freezer pro just to see how the milling was (and for extra clips just in case they ever break) and it was terrible too. I guess I should've just bought a Tuniq Tower... I may still do that.

 

[BonzaiDuck]

Well, we're all coming out of the same detail-obsessed mindset. You'll see here (I think) someplace I mentioned using surgical gloves. [and that was a pretty silly idea, except for my piece of mind.] You're talking about a "hole" where whater could get in, and you must mean the little rectangular break at the center of one side of the IHS. The rest of it looks pretty flush to the PCB component, and I don't think water would get into the works except possibly in that one place -- not without total immersion.

But like I said -- with the messy coffee-table, little room on it to place the sandpaper, I just cleaned off a flat area and held the sandpaper flush. Now I'm more confident about doing this more extensively with some other processors.

I think we got into this business totally conditioned to handle the CPU with focused care and concern. I'll still do that in the future, but I don't think I'll develop anxiety and heightened pulse-rate before lapping an IHS again.

 

[Idontcare]

Originally posted by: BonzaiDuck
Well, we're all coming out of the same detail-obsessed mindset. You'll see here (I think) someplace I mentioned using surgical gloves. [and that was a pretty silly idea, except for my piece of mind.] You're talking about a "hole" where whater could get in, and you must mean the little rectangular break at the center of one side of the IHS. The rest of it looks pretty flush to the PCB component, and I don't think water would get into the works except possibly in that one place -- not without total immersion.

But like I said -- with the messy coffee-table, little room on it to place the sandpaper, I just cleaned off a flat area and held the sandpaper flush. Now I'm more confident about doing this more extensively with some other processors.

I think we got into this business totally conditioned to handle the CPU with focused care and concern. I'll still do that in the future, but I don't think I'll develop anxiety and heightened pulse-rate before lapping an IHS again.

I have "immersed" my QX6700 many times in water. There is no harm unless you plug it into a mobo and power it on before the water has dried out from under the IHS.

Water alone will not harm your CPU, it takes voltage. At any rate all you have to do to ensure the water is fully removed from under the IHS if you are nervous at all is put a few drops of isopropyly alcohol on the hole while holding it at an angle, shake it around a bit, and then wick it back out by placing a coffee filter or a paper towel onto the hole.

All I can tell you is from experience I have had my processors pretty darn wet and they suffered no ill fate because of it. However there is absolutely no harm in being overly cautious. Taping over the hole or wearing surgical gloves, while unnecessary they cause no harm either.

 

[jedisoulfly]

If you do a lot of lapping and want to ensure a perfectly flat surface check out this site. The 6" x 8" one for $12.95 ($20.85 after shipping great shipping considering it weighs 12lbs) is more than large enough IMO for the lapping of HS and IHS. Has bilateral accuracy of .0001"

I have seen advertised for some HS (can't think of what ones off hand) that they hand lap to 1200 or sometimes 1600. Most likely they are lapping on surface plates just like this. The average table, counter or desk top is going to have a much higher tolerance than .0001"

http://www.grizzly.com/product...=granite+surface+plate