The processor is easy and you do not need to take it to a mirror finish.. thats just for show

It just needs to be flat and do be careful how far you go in sanding large amounts of material off. It only needs to be flat and any very deep grain lines removed.

As for the HSF.. that one is a bugger and I have done it twice with the TR120EX.. the third one I just had a friend do at a machine shop with strong warnings about how much to remove... enough was enough  LOL

It takes slow and steady moves and is very difficult to hold down by passing the fingers through the base between the heat pipes to hold the top of the base flat to the surface. Again, its does not require a mirror finish and the goal is to simply remove the nickel plate down to the copper and flatten the surface at the center removing any deep grain lines

The only issue after is since you are removing materials, small amount as they may be, you may wish to make sure the upper mount bracket has sufficient pressure to absorb the space distance which was removed. In some cases, and depending on how far you go with sanding, it may require a small thin washer be places between the upper X bracket and the top of the base plate @ center to give it just a touch more tension as it is tightened.

Wingo wrote on Jan 15^th, 2009 at 3:48pm:



no.. thats just what the hardware sites do to look impressive

Mirror finish you are getting into theoreticals, trust me..

In reality the more 'cap' you sand off the processor and get it closer to the cores under that cap the better heat transfer and that can be dangerous because the i7 is not made like the old Q's

the goal is to bring both surfaces down to copper and get them to make as much contact as possible flat face to flat face.

I would say start with automotive 280 (or somewhere in there) and work your way up .. 400 then 600 is just fine.. you can go to 800 if you want to


The processor I normally do dry.. the HS I use a bit of water to help slide the unit easier but I dry sand the initial to get the major off of it first, then add water to the 280 and work up

Alcohol is used to clean everything up

sometimes I may add a touch of water to 600 to finish off the processor but do take care as that moisture can get up under the cap if you are not careful, use very little.


here is a real engineer showing the process on the CPU

http://www.thetechrepository.com/showthread.php?t=71

Thanks for the info NickN. Is there anything that you don't know about computers? I swear your brain must weigh twice as much as a normal one with all the information in it. I'm going to lap my HS first and leave my CPU as I want to make sure my CPU still works when I finally get my motherboard back, just in case it needs to be RMA'ed. (I never determined which was the true cause of the fault, the problems pointed to the motherboard but I can't say for sure).

I have a few more questions to ask, if you don't mind answering them. Why did you need to lap the 120 more than once? Was it to clean it before reapplying a new lot of paste?

When you say the i7s aren't designed like the Qs, what do you mean exactly? Do they need a bit of heat to run well?

Last question is to do with thermal goops. Does the actual goop you use really matter that much, or is it more in the way it is applied? I was reading this article which is a bit dated but seems to prove a point: http://www.dansdata.com/goop.htm It shows that the cheap compunds can work as well as the Arctic Silver 3 as long as they are applied as well as the AS3 and stresses that the goop isn't there to actually conduct the heat as much as fill the gaps and leave as much metal to metal contact as possible. There are better conducters of heat than the goops, but the goops are designed to last as well as fill the gaps whie other materials will conduct the heat but not fill the gaps for a prolonged period of time. So in other words metal to metal is preferred over goops, goops are a compromise between conduction and longevity and are preferred over air pockets.

when I lapped my e8400 and hr-01+ i did the processor dry in figure 8 patterns.

with the heatsink I put water and dish soap onto the sandpaper. this keeps the HS from stuttering as you pull it across but it still sands well enough. It all does take time though, i spread it across 3 days, doing little bits here and there.

also, with the processor, at even intervals you should rotate it 90 degrees to make sure you are getting everything the same.

all in all i was able to shave 2-3 degrees (celcius) off my idle temperatures and 10 degrees off load; well worth it.