- Apr 23, 2001
A little while ago by my own stupidity and enthusiasm I posted this in the wrong section, I thought that the post would get eventually moved but alas it was overlooked.. so today I posted it again, in the right place as I am still really interested in your views
As most of you know, I am quite new to overclocking but have a knowledge that is ever expanding.. When researching my system, I was not aware of lapping a heatsink and thus built my system without doing so. Today however I built a PC for a friend and obtained the same heatsink as mine. I recalled a article on the front of overclockers.com which suggested that excessive lapping could actually hinder the heat transfer. I believe the article was concerning machining materials to precise measurements as well. I also read the Lapping Tips section and then attempted this task.
In the Lapping tips, it mentions rubbing grinding paste between the CPU and heatsink as an alternative to lapping.. so this is what I did to one heatsink and I lapped the other (lapped down to 600 paper).
To obtain a reasonably accurate result, I checked the room temp and fitted both FOP-32's to the same PC using the same clip and fan, while this is not scientific research as there may be manufacturing differences between the heatsinks, or between the amount of Artic Silver I used to assemble them each time..perhaps it suggests something.. but heres the temps reached after running the burn in on SI-Sandra for 30 mins
FOP32- Normal CPU @ 52 c | SYSTEM @ 27 c
(both FOP made same same temps at standard)
FOP32- Lapped CPU @ 48 c | SYSTEM @ 26 c
FOP32- Ground CPU @ 46 c | SYSTEM @ 26 c
Upon request I also lapped the heatsink a bit further the following day using 1500 grade paper (couldn't get 2000 grade ) and monitored room temperatures until they reached the same levels used in the previous test.
The heatsink was lapped using Artic Silver as the lube.. to ensure that there was no high or low spots the heatsink and CPU core was also brushed with engineers blue.. This complete process was completed on a mirror for a ideal flat surface (5 sheets of A4 paper were used to lap) These are the temps reached after running the burn in on SI-Sandra for 30 mins
FOP32- Lapped (1500 grade) CPU @ 47 c | SYSTEM @ 26 c
So there you have it a drop of 1c.. I am however unsure of if this can be compared to my previous results as it was done at a later date and things change such as dust in the air.. Perhaps lapping with 2000 grade paper will bring it down another 1c to match the grinding paste temps which today still read 1c lower than the lapped heatsink.
As previously stated.. perhaps not scientific.. but interesting all the same
The origional post (from which this information was copied can be read here)