Lapping

[Flava0ne]

I'm thinking about lapping my CPU and CPU Cooler, but I want to know how much it will drop my temps. Like what's the average drop if I were to lap my CPU and CPU Cooler and is it even worth doing it?

 

[Laurijan]

I heard from sneekypeet that lapping the cpu alone made temps drop 8C..

 

[Flava0ne]

Wow! That's a pretty significant drop.

 

[erocker]

It all depends on how uneven your IHS is on the cpu and the cooler itself. If you don't mind voiding your warranty go for it.

 

[freaksavior]

I'm thinking about lapping my CPU and CPU Cooler, but I want to know how much it will drop my temps. Like what's the average drop if I were to lap my CPU and CPU Cooler and is it even worth doing it?

Take A razor blade (like you would find in a box cutter) and put the sharp side against the base of your cooler. if it isn't flat do it.

I did it on my true because i relized the very middle (were it made the most contact to the proc) had a bit of an arch to it. I lapped it

the proc.. personally i would never do it, but if all you are concerned about is lower temps and a higher (maybe) clock then go far it.

There are a lot of tuts on here, i would start with 800 > 1000> 1200 > 1500 > 2000

Do each side about 8ish times back and forth (so total of 32 times with each piece)


final results of mine

You may have to do it more the first time to get most of the copper to show through..

 

[kenkickr]

Something I've done when I have a concaved heatsink or IHS is use a sharpie and make a X on the HSF base or IHS before sanding. This helps tell me were my high spots are during the sanding process.

 

[FR@NK]

There are a lot of tuts on here, i would start with 800 > 1000> 1200 > 1500 > 2000

you dont wanna use anything more then 800 grit...see this article:

http://www.techpowerup.com/articles/cooling/air/39

 

[Laurijan]

you dont wanna use anything more then 800 grit...see this article:

http://www.techpowerup.com/articles/cooling/air/39

There are different opinions about if using 800grid paper should be final.. i prefer 2000grid methode (shiny)

 

[freaksavior]

you dont wanna use anything more then 800 grit...see this article:

http://www.techpowerup.com/articles/cooling/air/39

yeah.. you CAN use that method, i started with either 400 or 800 and finished at 2000

 

[Flava0ne]

Yeah, I found this guide on the nVidia SLI Forums.

 

[FR@NK]

There are different opinions about if using 800grid paper should be final.. i prefer 2000grid methode (shiny)

Yea there are alot of opinions. I would recommend W1zzard's opinion:

[SUBHEADING]Do I want a mirror finish?[/SUBHEADING]
If you start lapping your surface beyond the 600 grit range your gaps in the heatsink will be very small. Thus you MUST apply a very very small amount of thermal paste for proper thermal transfer. In most cases where you have a mirror finish your final result will be like this:

There's no contact between the two metals because of excess thermal paste. The heat transfer of this is worse than with a surface that is only 600 grit because of the boundary layer of TIM between heatsink and CPU.

Either way, a 2000grit mirror finish will be better then not lapping at all :/

Here is the difference its made from lapping one of my HD4870s:

 

[nanohead]

There are different opinions about if using 800grid paper should be final.. i prefer 2000grid methode (shiny)

Yeah, you know what they say about opinions. In reality, lapping has been around for over a century, (but obviously not applied to processors ). Its been a standard process for machined metal surfaces, especially when they have to be sealed.

Efficient thermal transfer can occur without perfectly flat surfaces, but its much more important that the surfaces are aligned well. I hardly believe the stories of dudes lapping all their stuff and having 10-20C temp reductions. Its just not realistic, unless there was something wrong before the surfaces were cleaned up. Some of these folks report temps so low, its got to be below room temp, which leads me to believe they live in a meat freezer.

I personally, would NEVER EVER "lap" a processor cap, unless the warranty meant nothing to me. Cleaning up the surface of a heatsink has some merit, but most are pretty good to start with (especially most of the ones most of us would buy).

As far as abrasives, if both surfaces were nearly perfectly flat and shiny (higher number grit), and they mated perfectly (nearly a seal-like joint), then thermal transfer should be optimal. You probably would not see much difference with any compound applied.

I think its mostly hocus pocus when people debate abrasive grit levels. The simple facts are, that it requires a certain coarseness simply to "cut" the metal at all! You can do only 2000, but it would take a week to remove the same amount of material that you could remove in minutes with 400 or 600.

People confuse the objective of this exercise, which is to flatten and align the surfaces. Flattening means removing material, not making it shiny. A shiny surface can be a warped as an orange peel, but still be shiny.