How does this CPU lapping look?

[Zaskar]

Just lapped my Q9450, First time Ive lapped a CPU, went from 320 grit up to 2500grit and took a few hours.
It seems perfectly flat, I don't know how people get a perfect mirror without using polish, though Ive read you really shouldn't be concerned about that.

http://img183.imageshack.us/img183/2608/cpuprelapsp2.jpg
http://img260.imageshack.us/img260/1550/cpulappedjq1.jpg

Have a few questions about it, around the edges you can see some silver a bit where it bevels down, are you supposed to sand off enough where there is no more bevel and it is copper from edge to edge?

Also does the surface look ok? the temps don't seem much different with my TRUE with 2 fans, and the 3rd core (from 1-4) is still 10c higher then 1-2 and 14c higher then core 4 and thats on the same die.

If the lap looks ok, think a bowed block like a EK Supreme or Fuzion can fix the discrepancy? Or do you think with the temp difference as it is still with it being lapped, that something is just wrong with the sensor for that die? Is it even possible for 1 core on the same die to have that high a temp difference?

[biftek.]

You did lap it on a perfectly flat surface right? Like glass or marble?
And I dunno if it's just your camera but you have a lot of scratches going in all directions.

It honestly looks more like you sanded it instead of lapped it. You shouldn't have silver on the outer rim if you lapped it. (Unless your IHS was bowed)

[jarvo1]

Same as above, if it's lapped on a perfectly flat surface, there shouldn't be any silver left on it, it should be all copper.
For the mirror finish, glide the CPU lightly back and forth over the high grit paper in straight lines.

When I did my Q6600 a couple of months ago, the middle and edges were the first places where copper came through, eventually leaving me with 4 blobs in each corner and finally looking like this. It's not perfect but mirrored, I dont have a closer pic to show the faint back and forth marks though
http://i250.photobucket.com/albums/gg247/njjarvis/26032008303-1.jpg

[Zaskar]

You did lap it on a perfectly flat surface right? Like glass or marble?
And I dunno if it's just your camera but you have a lot of scratches going in all directions.

It honestly looks more like you sanded it instead of lapped it. You shouldn't have silver on the outer rim if you lapped it. (Unless your IHS was bowed)

Yea I picked up a piece of glass as a surface.

On an angle it looks mirrored and without the flash face on you can see your reflection, its cloudy though. But most the pics Ive seen have them look mirrored and I don't know how people do that, even the finer grit sand papers caused micro scratches. (Any advice there would be helpful if Ive done it wrong)

From the first pic you can see the angles of the outer edge, it tapered down, id of had to take off a decent amount more surface to have the very edge be pure copper, thats why I was asking if i should of done that. There is no patches that I missed or anything, just the fraction of a millimeter at the edges where it was originally beveled.

[ZOMGVTEK]

How the hell did it take a few hours? The first time i did a cpu (my current) it took me about 12 mins. I did 320, 600, 1000, then polish. Razor blade says its flat.

And it dosent look lapped, the sanding is uneven and surprisingly circular.
The whole idea is to get the thing perfectly flat, not shinny. If you truly went up to 2500 grit, it would be dam near a mirror shine. It is nowhere near that.

Go in slow strokes as long as you can in one direction for about 50 strokes (on moistened + a drop of soap on glass) then rotate it 90º, repeat, 90º again, and again. So you now did 200 strokes in every dirrection on the 300 grit. Look at the thing, does it LOOK flat and even? Are the strokes uniform and copper showing everywhere? The lowest grit is obviously the most important, it is what sets up the base for the next steps and actually makes it flat. The next stages just make it slightly better and shinny. If your processor looks good, step it up to 600 grit or whatever... and repeat...

Its quite logical, got the concept?

Oh, and if you want a mirror finish, your going to need to polish it. Its not going to drop your temps, if anything they would be worse. Its just for bling.

[Big_Daddy]

How much pressure did you apply? for that last run, just the weight of the processor should suffice..

[Zaskar]

How much pressure did you apply? for that last run, just the weight of the processor should suffice..

Used mostly CPU weight pressure for the whole thing, thats what I read in most guides.

Didn't know I was supposed to go in 1 direction, from what I read it was recommended to use figure eight type movements. Seems that was wrong.

Are you really supposed to use a small amount of water on the sand paper? Also read that shouldn't be done for the CPU?

[fluke420]

Go in slow strokes as long as you can in one direction for about 50 strokes

O'h yeah.

[ZOMGVTEK]

Used mostly CPU weight pressure for the whole thing, thats what I read in most guides.

Didn't know I was supposed to go in 1 direction, from what I read it was recommended to use figure eight type movements. Seems that was wrong.

Are you really supposed to use a small amount of water on the sand paper? Also read that shouldn't be done for the CPU?

It is easier to apply even pressure when you go in one direction. And to make sure its as flat as possible, rotate it until you get back to where you were. And soak the sandpaper in water and put a drop of soap on it and it would make the process much faster.

I never heard of this figgure 8 method, sure its going to work, but i think straight strokes are more logical. And use as much of the sandpaper as you can.

Just dont soak the cpu and its going to be fine. Although i lapped a few celerons and allways rinsed them underwater after to clean them. Never had a problem. Its not a big deal, just try to avoid soaking the thing.

[Zaskar]

Cool thanks.

On that first pic you see how the outer edge is beveled and on the second pic how there is a bit of that left, should I be lapping it down to completely remove all remains of that? Would require a decent amount more of the surface to be removed before it got that low.

[ZOMGVTEK]

Cool thanks.

On that first pic you see how the outer edge is beveled and on the second pic how there is a bit of that left, should I be lapping it down to completely remove all remains of that? Would require a decent amount more of the surface to be removed before it got that low.

You could, but its not a big deal. The most important part is that the entire surface is flat and even, the edges being lower is no biggie, it would barely change temps, if it all.

[Zaskar]

You could, but its not a big deal. The most important part is that the entire surface is flat and even, the edges being lower is no biggie, it would barely change temps, if it all.

I was mainly afraid of taking off too much, I have no idea of how thick the IHS is on the inside, didn't wanna chance going down that low and then it being too thin or something like that. Figured I could always go back and do that after asking here, but cant undo it if it was wrong.

If thats no problem then before I put the waterblock on ill redo it and go down a bit lower. And buy better sandpaper, I had one of those dumb kits that came with like 4-5" squares. On the finer grits it always seemed like some of the grit would periodically break loose, get under the CPU and create a large scratch across the surface.

Anyone have any opinions on the temp discrepancy part though, it may not be mirrored but it Is flat.

[Th3MadScientist]

that does not look like 2500 grit

[Zaskar]

This is a shot just after finishing the coarser grain sandpaper to show the beveled corner I was talking about.
Ill be picking up some packs of higher grain sandpaper and do it again right with the tips gotten here, should I also go back to 320 grit and take down more of the surface so thats gone and it goes flat copper all the way to the edge? (the other side isn't nearly as bad as that, the bevel goes down farther on this side slightly)

http://img59.imageshack.us/img59/6344/cpupartiallapyt5.jpg

And lastly, just to make sure, is it really recommended to wet the sand paper when lapping the CPU?

[adpr_02]

U probably used 320 and scratched the hell out of it, and then didn't spend enough time with the other finer grades of sandpaper.... but that is not how 2500 gritt looks like.

I actually had this problem on mine, where the sandpaper would randomly scratch the surface, I thought it was just dust or w/e underneath the sandpaper, or maybe larger granules of the material you're sanding broke off and went between the ihs and sand paper scratching it.

Get different sandpaper, and a different surface, and try again, but from what you show in the pick, start off with 800 gritt.

Good luck.

Edit. When I sand, I rotate the cpu in my hand, because even though you're sanding in a circular motion, you're putting more pressure on the thumb-edge, so if I use my thumb to sand each edge, I think it makes it more even, but I dont' know if it's just me that does that.

[Zaskar]

U probably used 320 and scratched the hell out of it, and then didn't spend enough time with the other finer grades of sandpaper.... but that is not how 2500 gritt looks like.

I actually had this problem on mine, where the sandpaper would randomly scratch the surface, I thought it was just dust or w/e underneath the sandpaper, or maybe larger granules of the material you're sanding broke off and went between the ihs and sand paper scratching it.

Get different sandpaper, and a different surface, and try again, but from what you show in the pick, start off with 800 gritt.

Good luck.

Edit. When I sand, I rotate the cpu in my hand, because even though you're sanding in a circular motion, you're putting more pressure on the thumb-edge, so if I use my thumb to sand each edge, I think it makes it more even, but I dont' know if it's just me that does that.
I bet your right on that, I had my own full sheets of 320, 400, and 600 grain, and the higher ones were very small sheets from a kit. Ill be picking up full size individual packs for redoing this this time.

When I did it though in the figure eight pattern, I did 10 figure eights then rotated 90 degrees and did another 10 so I didn't unevenly wear down anything. Thats why Im fairly sure I atleast have it perfectly flat.

[bball4ever39]

Well that's what really matters.

The mirror shine isn't necessary, as long as its flat.

[adpr_02]

^ True that. Probably if you polish it to a mirror finish, u won't even notice a difference from what it is like now. If you're positive that it's flat, leave it - unless you're a perfectionist.

[Zaskar]

^ True that. Probably if you polish it to a mirror finish, u won't even notice a difference from what it is like now. If you're positive that it's flat, leave it - unless you're a perfectionist.

Cool, what Im mainly focused on then is if I should atleast take it down a bit further to eliminate whats left of the original beveled edge I showed 2-3 posts up.

[Martinm210]

Cool, what Im mainly focused on then is if I should atleast take it down a bit further to eliminate whats left of the original beveled edge I showed 2-3 posts up.

No, don't worry about the edges, they don't do anything anyway. I't probably best to just sand down the bare minimum, that way you maintain as much thickness as possible of the IHS to do it's job in spreading the heat out.

I just finished my Q6600, only took 15minutes on 1000 grit, pretty easy. It does help quite a bit to use some sort of cutting fluid. I'm old fashioned and just use spit!...lol!

[Zaskar]

Thanks man, do you put pressure on it for the coarser grits? I was using hardly more then CPU weight when doing it, thats why it was taking so long, I didn't wanna mess it up since I haven't done this before.

[ZOMGVTEK]

Thanks man, do you put pressure on it for the coarser grits? I was using hardly more then CPU weight when doing it, thats why it was taking so long, I didn't wanna mess it up since I haven't done this before.

I used a fair amount of pressure. I dont see how its a big deal, youre sanding copper. The first time i did it it took me a little over 10 mins, and i have a mirror finish. It not very difficult at all.

Press down a good bit, its not like its going to explode or anything. I dont see how you can screw this up.

I saw a bunch of people say 'use the weight of the cpu' but after one stroke i realized that is useless. It would take many, many hours to get the finish i got without pressing. How does the pressure applied really even matter? As long as the thing isnt smoking hot, i dont see anything to be concerned about. Its going to be a whole lot better than stock.

[STEvil]

You can press on it, just be careful not to slip is all.

Bit of plating on the beveled edges doesnt matter, they wont be touching the heatsink anyways.