Die Guard Questions

Discussion in 'Extreme and Water Cooling' started by aussie-revhead, Nov 8, 2019.

  1. aussie-revhead

    aussie-revhead Member

    Joined:
    Feb 21, 2005
    Messages:
    19,402
    Location:
    St Clair 2759
    Dudes, I am messing around with my naked 7940x to try to improve temperatures and I noticed that the cpu die is flush with the die guard, is this right? Wouldn't you need a slight protrusion to get good contact with your water block? If the die is flush then your relying on the LM to bridge the gap right? Would you advise machining or sanding the die guard to make it protrude slightly? If the die is flush and I sand my water block to make it properly flat (instead of convex) then is that going to make contact and temps worse?

    Has anyone else noticed this too? Has anyone else modified their die guard to improve the fitment?

    Any help would be greatly appreciated.

    :)
     
  2. juzz86

    juzz86 Member

    Joined:
    Nov 15, 2013
    Messages:
    1,229
    Location:
    Forth, NW TAS
    I always thought they were designed flush so that the mounting pressure you applied securing a heatsink or block wouldn't crack the die mate?

    I may be wrong here, but it seems a slight die protrusion would mean the entire pressure of the heatsink would then be on the die - not somewhere you'd want it.

    Reminds me of busting an old Socket A Athlon XP actually - sad day, but lesson learned.
     
  3. OP
    OP
    aussie-revhead

    aussie-revhead Member

    Joined:
    Feb 21, 2005
    Messages:
    19,402
    Location:
    St Clair 2759
    It's quite likely and sounds logical, I'm just concerned that any pressure on the flush die will make it sink away and the block will be pressing on the die guard mostly

    Also In addition if waterblocks are generally concave then sanding it flat might make this issue worse.

    :/
     
    juzz86 likes this.
  4. juzz86

    juzz86 Member

    Joined:
    Nov 15, 2013
    Messages:
    1,229
    Location:
    Forth, NW TAS
    I see the point there for sure, but the mounts are designed to apply uniform pressure across the IHS surface so any pressure applied to the mount should, in theory (and assuming flat mating surfaces) also apply to the die.

    Are waterblocks truly concave? I must admit I don't put the Verniers over them, but subjectively the ones I've got here (old Kryos, old 380i, old Corsair AiO) are very close to being flat. I see a lot of reviews claiming a slight concavity or convexity, but it's rare to see it measured. If that's true, you'll end up with the die guard lower than the die anyway, and a little 'bubble' above the die for TIM. Being convex would be an issue - you'd potentially apply more pressure to the die than the guard.

    The old Socket A gear had 'spacers' around the outside of the package to help prevent over-application of pressure to the die - the problem arose with uneven and over-specced mounting pressures. Here's Anand's take on this very issue. Largely overcome today, but that Socket A clip mount was an absolute dog with a big heatsink.

    You've got me curious now, I'll see if I can borrow the calipers from work and have a look! :thumbup:
     
  5. deluxe

    deluxe Member

    Joined:
    Dec 9, 2003
    Messages:
    595
    Location:
    Melbourne
    Correct. You can check the relative heights for flatness measuring the diagonals with a straight edge but I can't imagine the die will be recessed. Also check your block in the same way, they're pretty precise these days but if you find any unevenness you can lap it flat.
     
  6. OP
    OP
    aussie-revhead

    aussie-revhead Member

    Joined:
    Feb 21, 2005
    Messages:
    19,402
    Location:
    St Clair 2759
    I have pics from dannyboi who lapped his EK block a few days ago, definitely concave.

    :)
     
  7. Dannyboi

    Dannyboi Member

    Joined:
    Jul 12, 2018
    Messages:
    39
    Location:
    Mt Larcom
    Higher coming out further in the centre / middle is convex not concave and gives a spin top ability to a suppoesed flat surface like block or IHS / die etc...

    I have had probably 30 waterblocks in my life and none have been flat.

    Here is EK supremecy Evo I had to lap. I could spin that waterblock like a spin top spinning for over 5 seconds..

    [​IMG] [​IMG] [​IMG] [​IMG]
     
    juzz86 and aussie-revhead like this.
  8. deluxe

    deluxe Member

    Joined:
    Dec 9, 2003
    Messages:
    595
    Location:
    Melbourne
    How much convex height did you have to lap down on that one?
     
  9. Dannyboi

    Dannyboi Member

    Joined:
    Jul 12, 2018
    Messages:
    39
    Location:
    Mt Larcom
    I'm not exactly sure how much as my goal was to make it as flat as possible which was accomplished.

    All I can say is that before lapping, it spun for 5 seconds on a flat surface before it stopped spinning so you can assume it was a fair bit. :wired:
     
  10. juzz86

    juzz86 Member

    Joined:
    Nov 15, 2013
    Messages:
    1,229
    Location:
    Forth, NW TAS
    Great info mate, thanks for chiming in.

    In that case, point on die pressure stands! The 'concave' threw me a bit :thumbup:
     
    Dannyboi likes this.

Share This Page

Advertisement: