On Saturday, I wanted to experiment with sanding down a heatsink to a mirror finish, so I decided to try it on the heatsink that was on my old 9200se video card.
!http://farm1.static.flickr.com/157/378826091_02f67b8c73_m.jpg!:http://farm1.static.flickr.com/157/378826091_02f67b8c73.jpg
After taking it off and wet sanding it for a bit, it immediately became apparent that the heatsink wasn’t real copper (I really should’ve guessed) and was just coated in some kind of copper-colored metal. It really doesn’t matter what kind of metal the heatsink is made of, though, so I just continued with the sanding.
!http://farm1.static.flickr.com/184/378826181_55a92dc982_m.jpg!:http://farm1.static.flickr.com/184/378826181_55a92dc982.jpg
Here is a picture taken after I was finished sanding with 400-grade sanding paper — as you can see, it’s just barely starting to get reflective.
!http://farm1.static.flickr.com/159/378826358_a4db801f47_m.jpg!:http://farm1.static.flickr.com/159/378826358_a4db801f47.jpg
Here it was after I was done sanding with 800-grade sanding paper — at this point, it wasn’t really any more reflective, and I was starting to wonder if it was working… at least I was starting to sand off more of that copper-colored metal coating.
It’s a very tiring process, lapping a heatsink. You have to constantly “cross-hatch” your sanding — basically, you have to make sure that you’re always, always sanding _against_ the grain in order to get an overall smooth appearance. What this means is that you have to remember to turn the heatsink a 1/4 turn every minute or two so that you’ll always be sanding against the direction that you just were. Got it? Good. 
Well, I really didn’t know what to expect at this point, until…
!http://farm1.static.flickr.com/184/378826943_bc097df255_m.jpg!:http://farm1.static.flickr.com/184/378826943_bc097df255.jpg
Wow. There’s quite a difference now, isn’t there? Turns out that nothing really changes all that much until you get to 1000-grade sandpaper and get to sanding with _it_.
The jump in reflective quality at this point was amazing.
!http://farm1.static.flickr.com/159/378826778_547e5bc070_m.jpg!:http://farm1.static.flickr.com/159/378826778_547e5bc070.jpg
Here’s another view, taken from a slightly different angle.
!http://farm1.static.flickr.com/168/378827554_0050784b38_m.jpg!:http://farm1.static.flickr.com/168/378827554_0050784b38.jpg
…and here’s my trusty old ATI 9200se. While it was never a powerhouse, it gets the job done. (And it’s far better than any laptop graphics. ) I’ve put a couple of big-ol’-honking heatsinks on the RAM chips, because I’m planning on overclocking the hell out of it — don’t worry, it can take it. I’m almost certain the card was severely _underclocked_ when I got it (166mhz core and 166 memory).
!http://farm1.static.flickr.com/152/378827863_79dc418c54_m.jpg!:http://farm1.static.flickr.com/152/378827863_79dc418c54.jpg
Here it is with the heatsink attached. What’s that bread-tie wrapping around it for, you ask? Well…
!http://farm1.static.flickr.com/135/378827974_8f4ff4303e_m.jpg!:http://farm1.static.flickr.com/135/378827974_8f4ff4303e.jpg
Okay, okay, okay — I know, it looks like shit. However, this was the only solution for now. (le sigh…) The heatsinks were too close to the motherboard’s RAM, and I had to “bend” one of the GPU’s heatsinks down a little bit to get it to fix. Of course, by this point, the thermal adhesive that it came with had already lost its adhesiveness, and I didn’t have any of my own, so… hey, it works, okay? It’s better than not having one on that chip at all.
All in all, it took about an hour and a half, but that’s just because I took my time with it. I’m thinking about doing it to the heatsink on the Titan’s CPU, just for the hell of it.
Why did I do this? I must admit, it makes absolutely no noticeable difference in anything performance-wise, and I can’t even tell if it’s lowering the temperature any (since the 9200se didn’t have an internal temperature monitor).
Doesn’t matter — it’s the _thought_ that counts, people!
