Currently lapping a CPU: doing alright? (pics)

B

Bop

2[H]4U
2FA
Joined
Oct 1, 2003
Messages
3,426
It looks a little weird the way it is coming out, the nickel parts seem nowhere near to being sanded off yet:

DSC00148.jpg


How does it look?
 
looks your proc will benefit from the lap job considering how unflat it is. Looks good otherwise.
 
make sure you use your lap :D

Looks like you're on the right track. Takes a long time and mucho effort.
 
All the nickel has worn off... when reflected against the light some parts of the copper looks darker, is that alright? A marker revealed that one corner stayed slightly visible after a few laps, but if all the nickel is worn out shouldn't it be flat by now?

EDIT: Went ahead and booted it:

Load temps on the cores are now within 3-6C of each other. BIG improvement considering it was 10-12(!!!) before.

I haven't even lapped my TRUE yet.

I'm getting around 70-74C load at 3600mhz 1.416v in P95 after 30 mins :( A definite improvement and closer temps between the cores but still hot!

2ND EDIT: The gap between the cores is as high as 10C again. :( I hope lapping the TRUE will give me better results.
 
Bop said:
2ND EDIT: The gap between the cores is as high as 10C again. :( I hope lapping the TRUE will give me better results.
Click to expand...

I wouldn't worry too much about that until you lapped the TRUE as well. That HSF is notorious for having a horribly misformed contact surface and alot of times lapping the CPU alone and not the TRUE will have very little effect. As a matter of fact, sometimes the misshapen CPU heatspreader and the stock TRUE contact surface have a better connection than a lapped CPU heat spreader and a stock TRUE.
I am sure that once you lap your TRUE you will see the results you want. Just know that it will be more tedious to lap the TRUE correctly than it was to lap the CPU.
 
Im about to get my first run with the stock HSF from my e2180. Too bad this thing doesn't even have a copper core all nickle :(. Unless I do the ex trick I don't think i would even notice it. Come next weekend i'm getting my AC7 Pro :D
 
if u have a glass table, u that to lap the CPU... get some 1500-2000 grit sand paper and tape the paper down to the table, this guarantees no concavity almost. do the same wiht teh TRUE, good luck
 
I don't think it is as flat as it should be. I'm going to gather some more material and go back to 800 grit and work on that for about 15 minutes and go back up to 1000-1200.

Then I'll do the TRUE.

I miss having bare processor cores... AMD Bartons were so easy to work with to get insane OCs :( 2.7ghz on a 2600+m!
 
@ you need to lap the base of the HS as well since the thermal transfer occurs between BOTH the surfaces... doing the HS will take you MUCH longer than the IHS on the chip though...
 
Yeah... I've gotten down to pretty much a copper base on the TRUE but one damn corner is still shiny nickel! Quite annoying... probably another 30-40 mins of lapping just to get it flat.

I've taken a razer blade to the CPU but some tiny amount of light still seeps through :( Looks like it isn't flat enough but I'm worried about wearing away more copper by bringing it down to 200-400 grit and having a go at it again. The thermal paste appeared to make excellent contact with the unlapped TRUE however.

Update: Looks like another 5-10C drop :D. At 3200mhz 1.288v I'm seeing load temps of 45-52C after 6 mins.(EDIT: 20 mins later... 55-56C seems to be the max right now)

2nd Update: After about an hour at 3200mhz the max load temp was 57C. Not bad! I'm priming 3600mhz @ 1.425v right now and it hits a max of 68C. Hot, but if it passes ~20 hours I'll keep that speed as I'll probably get a new CPU by next year so longevity is not a big issue. I know it is making proper contact so I can rest easy... although I do wish my load temps were closer between the cores. I didn't do a perfect job on the lapping but well enough.

I really hope Intel gets some better QC for their heat spreaders.

Overall improvement: 8-10C for max temps. Cores stay within 3-5C of each other during load compared to up to 12C.
 
Nice job!

I'm definitely lapping (or trying to lap) the HSF, a Tuniq 120.

But as for the CPU, I'm still unsure. Was this your first time doing it or do you have some experience? How easy is it to damage the CPU?

I'm guessing that the sandpaper remains stationary and secured to a flat surface while you drag the cpu or HSF across the paper?

It'd I think it'd be insanely helpful if somebody posted a video of themselves doing it, or mock-doing it. Sounds trivial, but I'd like to know if I'm killing it, or not doing anything effective at all.

Still trying to find and obtain an E8400 as early as possible, so it'd be a shame if I went through all this trouble to get it and then just ruined it.
 
You are not being patient enough with your lapping. In your first pic you clearly have some nickel left on your chip. I spent about 4 hrs lapping my Q6600 and TRUE. You keep putting it all together only to rip it apart and work some more. If you would take your time do it right the first time IMO it would be easier on yourself and you would get better results. Start with 800 grit and work your way up to 2500 grit, use 2 sheets of sanpaper on each (proc. and HS) until they are worn then move up to the next step. I finished mine off with Mothers metal polish and cleaned with 91 percent Isopropyl Alcohol.

Here is what they looked like when I was done:

IMG_2687.jpg


IMG_2670.jpg


With my 3.6 OC I run prime95 small FFT test and max out around 61 C on the hottest core.


With some patience and elbow grease I am sure you can do the same.

*edit* Use something very flat to wrap the sandpaper on. I ordered the lapping kit from SVC com and they give you a small piece of plexiglass which is sized slightly smaller than the sandpaper they give you. I then wrapped the plexi, taped the back with some duct tape and would go back and forth spinning the processor or heatsink 90 degrees about every ten back-and-forth strokes. Worked like a charm.
 
wow excellent polished mirror finish!... but it's not really necessary and does not lower temps. Once you get a flat enough surface, the mirror finish is just comestic, temps won't lower.
 
use a piece of glass from a picture frame (4x6 or 5x7) and tape the sand paper to it. I used to go to 2500 grit. Actually going to about 800 is better since the Thermal compound has somewhere to go. I had been suggested by others that 800 or 1000 should really be the max you go.
 
I may lap it further this weekend to see if I can narrow that 2-5C gap between the cores. I know that my TRUE isn't as flat as it should be yet but I was in a hurry for results and due to time constraints. (damn you college!)
 
I recently lapped my Q6600 and Asus Silent Knight Cooler. Saved me 8c and allowed me to sit at 3.4 and under 65c after 12 hours of Prime95. Idle 42-48c. Took a few hours but was well worth it in the end.

HS
heatsinkafter.jpg


CPU
cpuafter.jpg
 
Is there such a thing as too much lapping? Like, wearing it down too much? Or do I not have to be afraid of that?
 
well, you don't want to go too far other wise youll expose all the die which is obviously not good. the other thing about going "too far" would be that the HSF has trouble contacting the IHS because the IHS is not as tall as it used to be.
 
Well after a few more hours I didn't even get a 1C benefit. Light still shone through with a razor, either the glass I got just isn't flat or my technique was bad. Either way I don't feel it's worth to pursue any further, it is better than it was stock.
 
I read somehwere the other day that the TRUE is actually a smidge indented by design to match the slightly convex shape of the IHS on the processor. I was going to lap my HS and processor but now I'm going to try it out stock when my TRUE arrives and see where it takes me.

If the article was right about the TRUE not being totally flat then the OP will have to lap the HS to compensate.

Anyone else have any opinion on the TRUE being curved to better match the shape of the CPU?
 
Did you guys use wet sandpaper?
I don't know how, but either my IHS or Heatsink got concave in the process of sanding it...
I can see the spots where the cpu touches the heatsink, and it's always on the two side edges.
I don't know how this is possible, unless the piece of glass I received from scv was convex.. or the fact that the water/dish-soap made the center of the ihs more vulnerable to sanding.
Either way, temps dropped a couple degrees.. but it bugs me, knowing that I could've done better.

Also, how do you guys get a smooth finish to it? - Applying no pressure at all, the processor glides across the 2500grit sandpaper effortlessly.. .. sometimes glides when I don't even touch it... but the sandpaper, instead of smoothing it.. leaves it's own mark on the IHS.
 
You must log in or register to reply here.
Back
Top