Lapping Thermalright heatsinks

[boatillo]

I know all of their faqs say 'blah blah flat enough' or 'blah blah you could make it worse' but I want to know how many people DO lap their shiny new 120 Extreme or Ultima 90 down to a bare, mirrored copper surface??

Copper > Nickel, so wouldn't you want to get rid of it?

Only do so if your hsf shipped without being true flat?


I just bought an Ultra 90 and have paper 400 through 2500 grit so I figure I will....unless I really shouldn't!

 

[Navid]

Originally posted by: boatillo
I know all of their faqs say 'blah blah flat enough' or 'blah blah you could make it worse' but I want to know how many people DO lap their shiny new 120 Extreme or Ultima 90 down to a bare, mirrored copper surface??

Copper > Nickel, so wouldn't you want to get rid of it?

Only do so if your hsf shipped without being true flat?


I just bought an Ultra 90 and have paper 400 through 2500 grit so I figure I will....unless I really shouldn't!

I would never lap a sink before at least trying it once for the record.
Even if you decide to lap it, install it first and try it and make some measurements.
Then, after you lap, you can be happy that you improved it.

 

[jonmcc33]

The nickel fills in the microscopic layers of the copper and the nickel layer is so thin anyway. It's nickel plated mainly because copper will oxidize and look bad.

 

[wwswimming]

Copper > Nickel

the thermal resistance of a typical 100 micro-inches
nickel plate = .0001 inches = L (below)

K ~ 99 Watts/ meter - degree C (thermal conductivity by volume)

* 1 meter/39.37 inches (conversion to Watts/ inch - degree C)
~ 2.5

L/KA (formula for thermal resistance)

.0001" / [ 2.5 {watts/in degree C} * 2 inches^2 ]

(for area, call it 1.41 square, which is close and squares nicely to 2)

= .0001 degrees C/ 5 watts

= .00002 degrees C per watt

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

use a processor with 100 watts, and the net improvement
by removing the nickel is ... .002 degrees C.

if you can measure that, you have some metrology-lab
grade equipment.

ref.
http://www.engineersedge.com/properties_of_metals.htm

 

[boatillo]

Originally posted by: wwswimming

Copper > Nickel

the thermal resistance of a typical 100 micro-inches
nickel plate = .0001 inches = L (below)

K ~ 99 Watts/ meter - degree C (thermal conductivity by volume)

* 1 meter/39.37 inches (conversion to Watts/ inch - degree C)
~ 2.5

L/KA (formula for thermal resistance)

.0001" / [ 2.5 {watts/in degree C} * 2 inches^2 ]

(for area, call it 1.41 square, which is close and squares nicely to 2)

= .0001 degrees C/ 5 watts

= .00002 degrees C per watt

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

use a processor with 100 watts, and the net improvement
by removing the nickel is ... .002 degrees C.

if you can measure that, you have some metrology-lab
grade equipment.

ref.
http://www.engineersedge.com/properties_of_metals.htm

So you proved what I said Copper > Nickel
Of course I also see your point, its negligible in this instance.

Ok, so no one who read this has lapped their expensive heatsinks - how many have lapped the cover on their cpu? I am going to test how flat mine is first, but plan to do some dry sanding if necessary.

 

[Navid]

Originally posted by: boatillo

Originally posted by: wwswimming

Ok, so no one who read this has lapped their expensive heatsinks -

:laugh:
lol
How did you come to that conclusion?