Mirror vs. Not

[JigPu]

OK... I just heard about this on the "Better than lapping" post. Somebody mentioned that there was a debate over having a mirror finish on a heat sink vs. having a rough finish. What are the reasons behind each type of finish as the better? I would believe that they would be about the same... The rough finish would have greater surface area, but with the mirror finish, you could be completely assured that there wern't even microscopic air bubbles where the compound didn't get to.

Please explain.....
JigPu

 

[inertia]

I believe the arguement from the 1200 grit camp goes something like:

If you've got it too smooth, whatever grease you use will suspend the HSF above the CPU thereby limiting metal to metal contact and disrupting heat transfer. If it's roughish, the grease will ooz into the cracks and allow the HSF to settle.

I believe the arguement from the infinity grit camp goes something like:

More is better.

I stop with 600 grit because by then it's definitely flat, I don't see what can further be accomplished, more isn't always better and I'm lazy. I also don't mess with the CPU 'cause I'm not interested in buying a new one.

 

[bodezafa]

Now tell me if I'm wrong, but the ONLY reason to use thermal grease is to fill the microscopic air bubbles from uneven contact with the proc and hs.
The flatter the surface, the better the contact, and the less need for thermal goop.

 

[*spazzed*]

bodezafa (Jul 11, 2001 04:31 p.m.):
Now tell me if I'm wrong, but the ONLY reason to use thermal grease is to fill the microscopic air bubbles from uneven contact with the proc and hs.
The flatter the surface, the better the contact, and the less need for thermal goop.

That's what I was thinking.......now if you had the HS completely flat........and the die completely flat, the thermal compound wouldn't be neccessary to use, since you can get the most heat transfer out of the chip having both surfaces flat ??? D'oh I just lost my self

 

[dozier768]

i seriously doubt it would be at all possible to get it flat enough with your hand and some marble or granite you assume to be flat, not to mention who knows if the core is perfectly flat or not. if everything were perfect though that would surely be advantageous to the transfer efficiency

 

[[Oc]acaridans]

there is know way to get anything perfectly flat with out the use of very percise equipment, so going to a mirro finsih is really needed....when i lapped my HS, I did just enough to smooth out the ridges that were left from manufacturing and that was it. I used 1500, the surface that was left was smooth to the touch and looked scratchy, kinda like wire brushed alumunum.

 

[Kingslayer]

I have a BP-6 motherboard and twin 500's Celeron's. It's actually the wifes machine. It has to off the big heatsinks that Global Win made before the FOP series. One is lapped with a little texture, the other is mirror finish.

They both run at the same temp.

 

[outhouse]

Kingslayer has a good point and it has been discussed before, one guy said he did not lapp his HS and he just used compound and made it shine like a mirror and had a 1C drop in temps then the same HS lapped to 600 and he also continued to lapp and test until 1500 but the temps were identical with 600 as with 1500 but remember all HS have different bottoms and i'm a firm believer in lapping just to make sure its flat ive seen a few now that were way off from being flat so to me lapping to make it flat is more important then lapping to take the high spots off so that you get better contact, but i will always do both, i'm afraid of polishing it because many polishes have wax and if any is left in the microscopic pores and valleys then it will not let the thermal compound do its job correctly, the same goes for finger prints this too is a big no no.

 

[JigPu]

All very good points.. Thanks a lot, I really wanted to know what a difference would do, and I was about to ask if anyone had taken a temp with mirror finish, and then after scratching it up a little...

Thanks for the help
JigPu