i7 Lapping Mini Guide

Discussion in 'Modding Worklogs' started by Trucrymz, Nov 14, 2009.

  1. Trucrymz

    Trucrymz (Banned or Deleted)

    Joined:
    Jul 16, 2008
    Messages:
    362
    I've gotten lots of help and great advice from members of this forum, so I thought I'd try to input something useful. As part of my current project I lapped my 920 and made a mini guide along the way.

    Lapping is really very easy and although it can be time consuming, the results can be amazing considering it costs next to nothing! All you need is some wet & dry sandpaper, a flat piece of glass or any glass surface, some sticky tape and some time.

    The theory is very simple, by making the 2 mating thermal surfaces; your CPU and your heatsink/waterblock, as flat as possible, you will maximise contact and increase thermal transfer efficiency. You will get the best results by lapping both surfaces, but in this guide Im going to start with the CPU, the waterblock lapping will follow at a later date.

    Just be warned, lapping your CPU will most definitely void your warranty and can (albeit very rarely) damage your CPU. Lap at our own risk.

    Im going to be using 400, 600, 800, 1200, 1500 and 2000 grit paper. You can find most grits up to 1200 at any hardware store, and is essientially all you need. The finer grits can give you a smoother, even mirror-like finish, but are more difficult to find. I found mine at Autobarn. Any auto paint specialist should have it

    So I started by taping up the CPU with masking tape.
    [​IMG]


    I marked a corner for reference
    [​IMG]


    Taped down sandpaper to the glass surface. Make sure the glass is very clean and flat
    [​IMG]


    I start with 400 grit sandpaper and place the CPU heatspreader face down on the sand paper and lightly push forward & backward in a straight line. Do this 30 times, then turn the CPU 90' degrees and repeat, working all the way around.
    **Important: do not apply downward pressure when pushing the CPU, let the weight of the CPU do the work**

    After 1 sheet of 400 grit paper. As you can see, this particular CPU's heat spreader is very uneven, its concave and bowed in the centre.
    [​IMG]

    Gradually work your way up to the finer grits. You can use a little water to lubricate the paper if you wish. I personally do not use water, which means I go through the paper more quickly, But I don't risk the water getting under the heatspreader.
    After 600 grit. Usually by now, the nickel plating has been completely removed, but on this particular CPU there is still some nickel plating left. It means that this CPU's heatspreader was very concave and slightly slanted as well. It would have had rather poor contact originally.
    [​IMG]


    After 800 grit. The surface is basically flat, but not very smooth.
    [​IMG]


    After 1200 grit. Starting to be pretty smooth, with some dull reflections.
    [​IMG]


    After 1500 grit. Its now smooth and quite reflective. At this stage, I could have called it a day, as any further lapping is really just for aesthetics.
    [​IMG]


    After 2 sheets of 2000 grit. The surface is now very smooth and shiny. I used 2 sheets because at these finer grits you have make sure the paper is very clean any loose bits could scratch the surface too deeply. A few drops of water would help, just remember to continually wipe the grime off the lapped surface.
    [​IMG]
    [​IMG]
    [​IMG]

    Once your finished, remember to give the whole CPU a good clean with some alcohol to get any metal dust off. You can use isoprol. I use medical alcohol swabs for cleaning all my electronics and it works a treat. You can get boxes of a 100 for a few dollars from any chemist.

    As you can see, its very easy to lap a CPU, and combined with a lapped waterblock or heatsink, the results are very worthwhile. We often spend hundreds of dollars to get better temps, expensive heatsinks, watercooling, etc, but few options are as easy or as cheap. For about $15 worth of sandpaper and a few hours of your time, you can gain on average 5-10c drop on your load temps.

    I do hope you've enjoyed this little CPU lapping mini guide! it can be used with any CPU that has an integrated heat spreader (IHS).

    If you do use this guide to lap your CPU, please post some pictures of your results!

    For anyone interested, you can find the work log for my WaterCooled PC Mini tower here;
    http://forums.overclockers.com.au/showthread.php?t=821013
     
    Last edited: Nov 16, 2009
  2. Shadow_n

    Shadow_n Member

    Joined:
    Jan 15, 2009
    Messages:
    991
    Location:
    Brisbane
    oooh, shiney.
     
  3. Eagertman

    Eagertman Member

    Joined:
    Jul 17, 2008
    Messages:
    863
    Location:
    Brisbane
    Anything over 1200 grit and you're getting next to nothing in return for your effort.

    Nice lapp.
     
  4. dreamaxx

    dreamaxx (Banned or Deleted)

    Joined:
    Jan 28, 2008
    Messages:
    2,134
    Location:
    Gold Coast
    Does it still work?
     
  5. OP
    OP
    Trucrymz

    Trucrymz (Banned or Deleted)

    Joined:
    Jul 16, 2008
    Messages:
    362
    thats true, the goal is to get a flat surface, not necessarily a shiny one. But hey, it looks nice!


    LOL hopefully....
     
    Last edited: Nov 15, 2009
  6. Alba

    Alba Member

    Joined:
    Jan 2, 2002
    Messages:
    4,754
    Location:
    Shellharbour
    That's actually convex not concave which is a good thing. When you fit the HS onto it and apply pressure, you will get the best contact in the center where the cpu's cores are. Ever seen water block mods to make it convex? If you get a cpu with a good surface like the one you have(had) you should leave it alone.
    Some cpu's don't have a nice convex center like that and can be so bad that they are concave which means the HS will sit on the corners of the cpu and have no contact with the center at all. I had an i7 920 that was kind of like this, the top and bottom edges were around 1-2mm higher than the center. The easy way to check your cpu is with a sharp flat blade like a stanley knife blade, if the center is the highest point then don't touch it, all you will do is make it pretty which wont matter once it's in the PC, you'll void the warranty, it may even be worse than it was heat wise and there's no going back once you've done it.
     
  7. OP
    OP
    Trucrymz

    Trucrymz (Banned or Deleted)

    Joined:
    Jul 16, 2008
    Messages:
    362
    As I said, its concave, and bowed in the centre. Its is not convex as the sides are (was) higher than the centre. I did check it with a razor blade, hence I lapped it. Even if it is convex, the notion that it will be more efficient than 2 perfectly flat surfaces, is arguable. Im no thermal engineer, so I cant say for certain, but IMO two perfectly flats surfaces would be better.

    the mods done on the Swiftech GTZ waterblocks to bow the centre would arguably only benefit CPU's with IHS that is not perfectly flat. It is a quick method to avoid having to lap both surfaces and ultimately not as efficient.

    From experience, I've lapped 4 other CPU's and have never seen a degradation of temps.
     
    Last edited: Nov 15, 2009
  8. Alba

    Alba Member

    Joined:
    Jan 2, 2002
    Messages:
    4,754
    Location:
    Shellharbour
    Going by that pic, the center is all copper meaning it was sitting higher hence it sanded away first which means it was convex, am I seeing it wrong?
     
  9. OP
    OP
    Trucrymz

    Trucrymz (Banned or Deleted)

    Joined:
    Jul 16, 2008
    Messages:
    362
    If you look at the picture the sides are all copper as well, and whilst I was sanding it was definitely the sides that wore down first. Also if you look at the picture after the 600 grit, you will see that the heatspreader on a whole was slightly slanted, causing one side to wear down, whilst the other side still had nickel plating in the valleys.
     
  10. AussieJester

    AussieJester Member

    Joined:
    Jul 12, 2004
    Messages:
    2,552
    Location:
    Perth Western Australia
    Nothing over 800 grit actually been tested years ago theres already many guides here on lapping need another like a hole in the head IMHO...Polishing also does absolutely nothing for performance.

    KiM
     
  11. OP
    OP
    Trucrymz

    Trucrymz (Banned or Deleted)

    Joined:
    Jul 16, 2008
    Messages:
    362
    Its not an exact science, so I dont think you should be stopping at a set grit, rather as I pointed out, the goal, is too get a perfectly flat surface, not a shiny one. Use whatever grit is needed to achieve that.

    More guides just means there's more resources for anyone who's looking for a guide. It makes absolutely no difference to those who are'nt, I dont see how this could be negative...
     
  12. Glock

    Glock Member

    Joined:
    Oct 4, 2006
    Messages:
    1,252
    Because it's just muddying the waters?

    It's not concave - the sides wear way first because they're the first to contact the paper on, and push it down flat. Take a perfectly square block of aluminium and rub it around for a few minutes, you'll get rounded edges.
    They nickel plate those things for a reason - have fun with your corroded cpu in a few years.
    Oh, and I wouldn't be so confident the spreader was slanted - just as likely you ended up grinding down that corner less.
     
  13. dreamaxx

    dreamaxx (Banned or Deleted)

    Joined:
    Jan 28, 2008
    Messages:
    2,134
    Location:
    Gold Coast
    You couldn't pay me to lap my i7, playing with fire and pointless.
     
  14. qnutnut

    qnutnut Member

    Joined:
    Feb 25, 2009
    Messages:
    173
    Location:
    Sydney
    not to rain on your parade or anything, but I believe this is true. Since there is no real way to make sure that you havent ever so slightly applied more pressure to one side than the other...Well not with a human hand anyway.
     
  15. OP
    OP
    Trucrymz

    Trucrymz (Banned or Deleted)

    Joined:
    Jul 16, 2008
    Messages:
    362
    Again, yes it was concave. If a block is flat, it doesnt matter how you rub it, its not going to wear in the same way, that heatspreader did. And yes Im very confident it was slanted, even in the first pic, if you look carefully, the "centre" is actually off centre. I use a proven method to lap by turning the CPU 90degrees after every 30 repetitions, so it is unlikely that one side gets rubbed more than the other enough to cause that.

    I've got a Q6600 I lapped nearly 3 years ago, and it had no signs of corrosion whatsoever (I took it out when I upgraded the motherboard). Have you ever owned a lapped CPU for a few years? I somehow doubt it....

    Im not trying to convince you or anyone to lap their CPU, I provided a guide to what I have done to mine and of my personal experiences. Yes, there are pros and cons to lapping, and it is up to the individual to decide whether the potential gains in cooling are worthwhile or not. Having said that, a potential 10 degree drop is nothing to sneeze at, many people are willing to spend hundreds of dollars on heatsinks, watercooling etc to obtain that sort temperature drop

    Seriously, I couldnt care one bit whether you guys like to lap your CPU's or not,but thread crapping with negative comments is not constructive and generally unappreciated
     
    Last edited: Nov 15, 2009
  16. Agg

    Agg Lord of the Pings

    Joined:
    Jun 16, 2001
    Messages:
    31,422
    Location:
    A Reported Post Near You
    Thanks for the guide. :) Good to see people still doing this occasionally.
     
  17. OP
    OP
    Trucrymz

    Trucrymz (Banned or Deleted)

    Joined:
    Jul 16, 2008
    Messages:
    362
    You are most welcome. Thanks for the positive comments!
     
  18. OP
    OP
    Trucrymz

    Trucrymz (Banned or Deleted)

    Joined:
    Jul 16, 2008
    Messages:
    362
    Firstly, the guide is there as a resource, it does not muddy anything, because the reader is free to decide whether or not he uses it. If anything, it provides an option that many readers may not yet know about.
    The paper is taped down to a perfectly flat surface, so the sides do not push the paper flat as you have stated.

    There is, and its described in the guide. By turning the CPU 90 degrees after an equal number of repetitions, and not apply any downwards pressure, you can minimise any differences to the amount of wear to each side

    To "ever so slightly applied more pressure" would not result in the above results. There would have to be a significant difference in pressure to create such a slanted result as seen in the pictures of after 400 and after 600 grit
     
    Last edited: Nov 15, 2009
  19. LuciferDarklord

    LuciferDarklord Member

    Joined:
    Mar 14, 2002
    Messages:
    86
    Location:
    WA
    i think you'll find a figure '8' motion ends up with more of an even lap. Best to turn 90degrees every so often to even out any pressure differences in your technique. Would be nice to see some before and after tests to prove if it achieved anything. You can use light machine oil on the paper - its not conductive. You can also use fine lapping paste straight on the glass (oil based)
     
  20. memnoch

    memnoch Member

    Joined:
    Jan 4, 2002
    Messages:
    517
    Location:
    Sydnet

Share This Page

Advertisement: