ocmooz Posted November 20, 2008 Posted November 20, 2008 Stopping at 1200 or 1000 (like I did) is ok it just needs to be more flat and smooth. As for what I used to monitor NB temps I got my readings from PC Wizard 2008, made by the same people that brought us CPUID, but I only run it while stability testing because of its inherent glitchiness (iTunes, Firefox, and a handful of other programs have conflicts with it). Share this post Link to post Share on other sites More sharing options...
Verran Posted November 21, 2008 Posted November 21, 2008 Well, if I do decide to lap the IHS would a good order of sandpaper be 400 > 800 > 1200 > 1500 > 2000? Yeah, I'd consider the 2000 step optional though. It's not that important and it can be hard to find paper that fine anyways. Share this post Link to post Share on other sites More sharing options...
Andrewr05 Posted November 21, 2008 Posted November 21, 2008 I remember someone on OCC worked at a professional polishing place (or something) and they said that being too flat or too polished actually had negative effects... Of course if I remember correctly such flatness/polishness (is that a word? er... shininess?) wsn't even capable of being done with sandpaper anyways... Share this post Link to post Share on other sites More sharing options...
DLS2008 Posted November 21, 2008 Posted November 21, 2008 Andrew: I have heard that sanding and polishing too finely will fill in the natural pores in the metal and reduces its heat transfer ability. I don't know if there's any truth to this though. Can anyone support or disprove this? Share this post Link to post Share on other sites More sharing options...
DLS2008 Posted November 21, 2008 Posted November 21, 2008 I'm guessing that a 7C to 10C difference between my cores is bad? 2 cores are at 58C and 2 are at 66C under 100% load from Prime 95. My Q9450 is 450 x 8 = 3.60 GHz at 1.280V in CPU-z and 1.3325 V in BIOS. All temps are and load are from Core Temp. I would love a water cooling setup, but don't have the money myself. Hmm... Christmas is coming up soon. Can I get some estimates on a price for a good water setup? Share this post Link to post Share on other sites More sharing options...
Crazy_Nate Posted November 21, 2008 Posted November 21, 2008 I remember someone on OCC worked at a professional polishing place (or something) and they said that being too flat or too polished actually had negative effects... Of course if I remember correctly such flatness/polishness (is that a word? er... shininess?) wsn't even capable of being done with sandpaper anyways... I don't think your general overclocker has the capacity of doing extreme polishing without specialty tools. The two materials labs I used (one at NIST, and one at RPI) both had Leco machines (just like this one: http://www.leco.com/products/metallography...p_consoles.htm). At NIST, I used a specialty multiple sample grinding / polishing machine as well as doing them by hand. Typical progressions were like this: Cut samples to the right thickness with a diamond saw. Grind using 120, 240, 480, 800, 1600 grit...all wet (there is a very large difference between wet and dry - wet is coarser but removes more material because the paper doesn't get "clogged") After that, you switch from sandpaper to using a cloth and various diamond pastes, which are rated by size of the diamond particles in the solution. Typical progression: 15 micron, 5 micron, 1 micron. Last, but not least, I would finish with a ceramic polish, which usually consisted of an alumina powder that was mixed with de-ionized water on the lapping cloth. Particle sizes were on the order of 1/20 micron. Not bad, you say? Well, consider this...in between steps, you have to run your parts through an ultrasonic cleaner (any inclusions - pits or pores, etc - will hold on to grit from the sandpaper or the diamond / alumina paste, and can contaminate your next stage). Considering the prices of some of the pastes (a small spray bottle of diamond could easily cost as much as your processor ), you don't waste the stuff... Also...you optically verify that you are indeed finished with the stage that you are on...using a microscope and looking at the size / orientation of the scratches... For 'lapping' a CPU, you probably aren't getting near the typical 'lapping' stage in metallography...you're simply grinding down the surface, and getting it as flat as possible. Having your IHS and your heatsink as flat as possible allows for excellent contact at all points, and in turn, allows for less thermal paste to be used (lowers the thermal resistance in doing so). I will stress that you are achieving flatness, not just a "really shiny surface" To Andrew's Q... The lapping stages of metallography involve a lapping cloth, which is not as rigid as silicon carbide sandpaper...it can tend to round edges or prevent you from keeping the same level of flatness. The pastes can also get stuck in the microscopic inclusions (read: material defects / pits), and the average person does not have the means to remove them easily. I cannot speak as to how that would affect the thermal performance, other than the particles are very, very small, and hopefully inclusions are few and far between. I had a task at RPI of measuring diameters of these micro-tube test sections (that I was fabricating) for a heat transfer experiment. I wouldn't bother with the finest polishes because they would round the ends of the tubes and it would make taking focused pictures (microscope + CCD) a pain... Share this post Link to post Share on other sites More sharing options...
DLS2008 Posted November 21, 2008 Posted November 21, 2008 Gotta love Crazy_Nate and his amazingly detailed responses. Share this post Link to post Share on other sites More sharing options...
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now