I'm building a new rig, but my dad decided to buy the parts without consulting me. He bought good parts, but he got a Zalman cooler. It would work fine for stock settings, but I want to overclock. I'm looking into possibly exchanging for a Thermalright Limited Edition Ultra-120 Extreme. I'm wondering if I should lap one or both of the heatsinks? I have double of everything for 2 new rigs and I'd probably only get one TRUE because of the money. I've heard some say that the TRUE is supposed to be convex and others disagree. So tell me whether I should lap one or both of the Zalman and TRUE. Also, should I lap the CPU itself? Finally, if you recommend lapping, where can I get my hands on some like 800+ grit sandpaper. The best I've seen is 800 grit in stores near me. I would like to go to at least 1500, 2000 if I can get it.

Here's an excerpt from a review at FrostyTech:

Source: Frostytech.com

If you follow the link you can see a picture of the base. The best thing is to also get some responses from guys here that have the TRUE to see if they saw the same thing on their heatsinks.

I lapped my TRUE and I can say it definitely wasn't flat. I didn't take pictures, but I think it's very much worth the effort and I've heard a lot of people here say the same. Thermalright has a history of not-flat bases all the way back to at least the XP-120 which a lot of people (including myself) lapped as well.

As for lapping the cpu... it's easy enough and they're usually not flat either, but it's also a lot easier to mess up and a lot more expensive if you do. So that's a decision for you alone...

To be able to find out if it needs lapping use a razor or straight edge ruler. Napa or Auto Zone would have the 1000, 1500, 2000, 2500, 3000. But the finer grits would be used more for polishing rather than lapping. When I lapped my processor and heatsink. I used 320, 600, 800, and finished with 1500 grit paper. Do not forget to use a a nice size piece of glass. Possibly a mirror or the glass out of a picture frame.

Never even bothered to check mine. My temps are at 60C full load at 4.2GHz.

ending on 800 is fine... I did my TRU to 1500 and I think I did my XP-120 to 2000, I definitely did my Thermaltake Silent Boost to 2000+ as it really was a mirror finish, but it really doesn't make much difference at all over 800 in terms of thermal transfer as it is very flat and smooth by the time you get to 800... anything higher *looks* nicer though

if you just did 400 followed by 800, it would be a lot better than stock, and probably perform within 1°C of any higher grits

apart from looking nice, the higher grits allow you to evaluate the flatness by the distortion in the reflection... no distortion = flat

So from what all of you posted, I take it that it should actually be flat and not convex like it comes. Does anyone have any test results or at least temperature differences between lapped and not lapped so I can decide if it's worth my time? Is a progression of 320, 400, 600, 800, 1200, 1500, 2000 good? I know it's not necessary to go past 800, but I like the pretty mirror finish, lol. And one last thing, is it all wet sanding? Do you sand in circles or do you do one direction then rotate the HS and go the other direction, I've heard people say both ways, is there a difference?

what Zalman did you get?

if it's the 9700 I would just keep it as it's very on par...

why spend the extra money on a Thermalright and then have to spend extra time to lap it flat... isnt this what you are paying a premium for...

I just bought a XP-120 for my HTPC and had to spend a hour and a half lapping it cause it was one of the worst heatsinks for flatness...it was crowned inwards on both axis and the center wasnt even touching the heatsink at all...it was out a whole millimeter...

I have bought dozens of Zalmans and all of them have been the flattest I have ever had....not to mention the XP-120 was 5c higher in temps with the exact same TIM application compare to the 9500 I took off it...

I will NOT settle for a high price on a heatsink and then have to resort to making it flat in the process nomatter what the analogy is...

if you have a 9500/9700 just run it and see....I have a 9700 on my Q6700 and I have had it to 3.8ghz and temps never went over 60c across all cores at 1.5v and thats on a Quad...dual core wouldnt put out as much heat...(only slightly less)

there you have it...

as for lapping...I rough cut with 320grit and finish it off with 400grit...with 400grit and very light pressure it does as good a job as 800grit...

you really dont need a mirror finish on a heatsink to do the optimal job... quite a few I have found a 400grit job does as good or better than some mirror finished job...

the important part is flatness...

I'd definately do it!

Just don't do anything like this : http://www.tomshardware.co.uk/forum/253302...800gt-110c-idle

I'm not retarted! Now that we have that cleared up, I think I'll test my Zalman, then possibly lap that. If I still want lower temps, then I'll buy the TRUE and test that, then maybe lap the TRUE. Now if only I had a motherboard and some RAM, lol.

As I said....

I havent had a Zalman yet that needed lapped...I have 8 of them here and have bought dozens...

What I do is get a machinist straight edge that measures the flatness...you'll know right away if it's out or not...

I will tell you this I upgrade to a 120 extreme and an antec 1200 case over this past weekend. I lapped the cooler and added a sythe fan at 133cfm and my temps went from 40 idle and 49 at load down to 27 idle and 36 under load so in my mind it is every bit worth the effort to lap the cooler. I am estatic with my temps so I will not be lapping my cpu. And to boot these temps are using Thermalrights chill 2 thermal compund. Have Arctic silver 5 but read some very positive reviews on the chill and decided to try it and I must say the reviews were right on.

true....

when you achieve results like that it tells you the cooler wasnt flat....if it was lapping it would have brought NO change in temps....

Contact patch of the mating surfaces are what you look at....it's okay if the heatsink is concave as long as the IHS on the CPU is convex the same amount or vise versus...doesnt matter what angle the axis is as long as both surface mate the same angle...

Wish I had a pic of my old Q6600 before I lapped that. I swear the heatsink must have been re-routed to the soup ladle machining department by mistake...

To put in quantitative terms, I lapped my Q6600 and TRUE and went from 30 degrees at idle and 50 degrees load to 24 degrees idle and 36 degrees load with no change in the fan or TC.

After Lapping: image

I had to RMA 2 of the TRUE's before I got something I could work with

http://forums.overclockersclub.com/index.php?showtopic=81238


I'm glad I lapped mine. After reading peoples stories both here and the Newegg Reviews, I can't believe that Thermaltights QC lets these go out like this... Still the same, I trust it more now that I know it's flat

I have never seen the need to lap my heatsink or processor but some of the results do seem to show improvements. I like dr-bowtie have usually used Zalman heatsinks and think they are excellent heatsinks but I don't usually overclock.

Could some of the improvements in temperatures be put down to the heatsink compound used I wonder? The improvement seen by mistabonz from 40 to 27 and 49 to 36 is quite large. Was this using the same heatsink compound or is it a case of going from a generic brand to something like Artic Silver 5. This information is essential in judging heatsink performance.

I can understand the need for a flat surface but why have it highly polished. If the surface is slightly rough, to what degree I don't know, will this not improve the surface area for the heatsink compound to do its job. If you go to the other extreme and have both the heatsink and cpu flat and very highly polished would you reach a point were heatsink compound is not needed?

Regards

Paul

To answer your question Paulktreg I went from a Zerotherm Nirvana 120mm cooler to the 120mm extreme. The next thing in the equasion, I think helped a great deal, is I went to the Antec 1200 from Nzxt Apollo mid. In doing so I think I dramatically improved airflow in my case. Lapping the true was a no-brainer because of how rough the base was and very uneven to boot. I was using Arctic Silver 5 on the Nirvana cooler and deciced to use the Thermalright chill compound because all the reviews I have read showed it was on par with arctic silver. So I think as a whole, in doing everything mentioned, had the desired affect. So to point at just one thing that dropped the temps so dramatically I can't do.

I thought the improvement was quite large thats all and it's interesting to see it was a combination of factors and not just the heatsink. It would of been interesting to see the temperature difference with just an heatsink change.

Regards

Paul

With less variation in the surfaces (ie, higher polish) the need for TC is reduced, not made unnecessary. Less variation simply means you need less volume. I literally used just enough TC that it would smear to roughly the size of the core (quarter size) under pressure when the heatsink was attached. It took a few tries to get this amount right, but the results were worth it. For reference, I would imagine there was roughly enough TC that the thickness of the layer it created was about tissue paper thin.

well....I dont know about others but I have my opinions from testing....I have lapped several Thermalright heatsinks and a few stock Opteron heatpipe heatsinks (Foxconn) and a couple Evercool VC-REs....

I can tell you that on all cases I didnt see any improvement from the finish of the base passed a light pressure rub of 400grit paper to a 2000grit shine...

I believe the surface being flat is the ultimate part of getting nice temps...from my personal understanding of the metal copper...I think that the finer grit you use to rub on it either just smears the metal over the pores or fills in the pores with the fine grit it rubs off....thus filling the pores and not letting the TIM do the job as well as it could....

The pores are still there but the finer the polish the more covered up they are...I use 400 grit as a polisher for a semi-finished surface...lighter pressure will do the same as 800grit with a tad more pressure...

I have used the lapped heatsinks with no TIM and only see a 2c difference on load temps...and I think that is just the air trapped in the pores with no TIM present...

I have tried just mounting the sink with no TIM on several occasions as a test to see how well the lapped sink is...even overclocked...you'll know if it's better...it will be no worse than when you started....

I use the Zalman grease on all builds...I dont like AS5 or much others...with the zalman grease it's thing enough I can get the TIM application dont very thin and with repeatable results on every single test...I cannot get AS5 as thin... AS5 always turns to crap after 6months to a year and in a couple cases turned brown and flaky....the Zalman gease has always been the same...

Contact patch is very easy to test also....I also tested the contact patch and wiped the grease off the heatsink and retested the contact patch and still had a sufficient contact with even less TIM left with no change in resulting temps...

it's all in application....I usually tell people to apply the TIM and attach the heatsink and lock it down...pull the sink back off and wipe all the TIM off the heatsink and re-apply it...if temps drop...you had too much TIM...if they go up...the surface(s) arent true...if it stayed the same...well you got a flat surface and less TIM...

I use a socket 939 Athlon X2-4200 I once had here for a boat load of test on different heatsinks from Zalmans to Thermalrights to stock coolers as this particular chip would idle at 80c with the stock cheap heatsink....65c with a lapped Opteron sink and a modded fan...55c on a 9500 zalman

it was the hottest chip I ever ran...I ran it 2 weeks fully loaded on Orthos at 85c and it didnt die so it became the official heatsink test chip....after all those test it got a new home...lol