Lapped Heatsink temp results

pegasus1 · 19 Jan 2006 at 16:32

Having practiced on a couple of old Skt A H/S i decided to give it a go on my Gigabyte G-Power Pro which is far from smooth, it took about an hour but i ended up with a nice smooth finish with a slight sheen to it. After replacing the H/S and running prime for an hour and giving COD2 a 30 min bashing it dropped 2-3 degrees C after an initial drop of 4 degrees C (the central heating is on now so thats probably why).
Might try it next on my 7800 GTX cooler.

Zefan · 19 Jan 2006 at 17:20

I've never had the need to lap my hs's. Always nice and smooth and not shiney

.

:edit: nice temp results btw

pegasus1 · 19 Jan 2006 at 18:50

Zefan said:
I've never had the need to lap my hs's. Always nice and smooth and not shiney .

:edit: nice temp results btw

Ive just checked 6 varius CPU H/S's and 3 GPU H/S, all good makes and none could be described as being smooth. When i say smooth i mean looking across the face with a light above and seeing no ripples, scratches or dimples.
Well its now at 35C whilst im browsing the web (in a warm house)which is 1 degree C cooler than it drops too during the day when the house is cold.
Im happy with that

lordedmond · 19 Jan 2006 at 19:41

can i pose a question that has puzzled me
people try to "lap a heatsink" by hand this produces a shiny surface , but is it flat?
is the mating suface of the cpu flat?

if the surface of both have microsopic imperfections ,the the total surface area for heat conduction would be greater than a flat plane
if these ripples ( micro scopic ) are filled with thermal compound then the surface for heat transfer would be greater

thinking about laping a surface unless you use an optical comparitor ( which can read imperfection less than the wavelenght of light ) you cannot be sure its flat , but do you want it that flat

for example two pieces of steel made flat they when slid together will stick ( wrung together )

so after all that do we need these shiney surfaces or will a dull surface conduct heat better

HeadlessChicken · 19 Jan 2006 at 20:21

lordedmond said:
so after all that do we need these shiney surfaces or will a dull surface conduct heat better

I think you will find that he answered tha with his before and after temps. :rolleyes:

pegasus1 · 19 Jan 2006 at 23:41

lordedmond said:
can i pose a question that has puzzled me
people try to "lap a heatsink" by hand this produces a shiny surface , but is it flat?
is the mating suface of the cpu flat?

if the surface of both have microsopic imperfections ,the the total surface area for heat conduction would be greater than a flat plane
if these ripples ( micro scopic ) are filled with thermal compound then the surface for heat transfer would be greater

thinking about laping a surface unless you use an optical comparitor ( which can read imperfection less than the wavelenght of light ) you cannot be sure its flat , but do you want it that flat

for example two pieces of steel made flat they when slid together will stick ( wrung together )

so after all that do we need these shiney surfaces or will a dull surface conduct heat better

Well professor, the ripples and imperfections i saw didnt need scientific instraments to bring them to my attention. In an ideal world the CPU/heatspreader and HS would be perfectly flat thus needing no thermal compound but as with most things in life this is not the case, i believe that the less thermal compound used, the greater the heat transfer between CPU and HS. I have carried out a scientific experiment and found that my temps are as described above, whether or not you like it, they are the results.

Mekrel · 20 Jan 2006 at 00:02

Nice one mate,

I have an AMD Opteron which takes 1.55v to run at 3.0ghz and the Big Typhoon I purchased for it is far from smooth.

What did you use? Im thinking of buying a lapping kit I saw online here http://www.3dvelocity.com/reviews/pcviper/ultrakit.htm

Col_M · 20 Jan 2006 at 00:47

I know exactly where you're coming from lordedmond, if a surface was "rough" in the engineering sense then you could justifiably say that there is a larger area for contact and heat transfer. I think in reality unless the peaks of one surface line up with the troughs of the other and perfectly mate at a atomic/molecular/crystalline (whatever scale the roughness level is at) there is likely to be peak to peak contact between the surfaces. That would cause voids that need to be filled with TIM which is a less ideal than direct surface to surface contact. Fair point though

I can agree with the flatness, i doubt hand lapping that people do at home is that good compared to controlled machine lapping but to be honest the flatness is usually good enough and like here shows good results. Those of you at uni could try going to the engineering department and asking if you can use the polishing machines used for preparing metal samples for use in a SEM (scanning electron microscope), whether you'd get better temps i don't know as it could follow the law of diminishing returns and you may get bugger all for all your extra effort.

pegasus1 · 20 Jan 2006 at 07:13

Mekrel said:
Nice one mate,

I have an AMD Opteron which takes 1.55v to run at 3.0ghz and the Big Typhoon I purchased for it is far from smooth.

What did you use? Im thinking of buying a lapping kit I saw online here http://www.3dvelocity.com/reviews/pcviper/ultrakit.htm

I read a few guides on the net and then went to halfords and bought the Wet and Dry paper, its not hard but took around an hour.

lordedmond · 20 Jan 2006 at 07:24

hmm interesting comments

Mrchu_ · 21 Jan 2006 at 01:59

did you use a a sheet of glass underneath?

pegasus1 · 21 Jan 2006 at 08:33

used my kitch table.

Carter · 21 Jan 2006 at 11:31

Mekrel said:
The Big Typhoon I purchased is far from smooth.

Same here, I've been thinking about doing mine.

Mrchu_ · 21 Jan 2006 at 21:05

use some glass,when you do yours if you can^^^^ its really flat/smooth.+ easy to clean.
Mirror works.

Well done on the lapping & temp drops BTW.

pegasus1 · 21 Jan 2006 at 21:18

after about 20 hours its working 2-3 degrees cooler at idle (34C reported by speedfan/asus probe) but i do have very good case cooling,it was previously idling at 36-37 degrees.

Cyber-Mav · 21 Jan 2006 at 22:05

lordedmond said:
can i pose a question that has puzzled me
people try to "lap a heatsink" by hand this produces a shiny surface , but is it flat?
is the mating suface of the cpu flat?

if the surface of both have microsopic imperfections ,the the total surface area for heat conduction would be greater than a flat plane
if these ripples ( micro scopic ) are filled with thermal compound then the surface for heat transfer would be greater

thinking about laping a surface unless you use an optical comparitor ( which can read imperfection less than the wavelenght of light ) you cannot be sure its flat , but do you want it that flat

for example two pieces of steel made flat they when slid together will stick ( wrung together )

so after all that do we need these shiney surfaces or will a dull surface conduct heat better

your theory would work IF the thermal compound had a 100% heat conduction and transfer rate as the heatsinks surface. since the thermal compound is never going to be as good as a direct heatsink to core contact with no "middleman" thats why its best to have as little pits as possible on the heatsink and the core...

MikeTimbers · 22 Jan 2006 at 13:01

Use graph paper to check the smoothness. Its reflection will quickly show up any imperfections or convex/concave distortions. Before you start draw an x across the bottom and when you first start see if the x disappears uniformly or only in the middle/at the edges.

Remember: shiny != flat.

Which is flatter?