too much thermal grease. Does it really matter?

Discussion in 'Overclocking & Cooling' started by dr.kevin, Sep 17, 2012.

  1. dr.kevin

    dr.kevin 2[H]4U

    Messages:
    3,055
    Joined:
    Feb 17, 2006
    Note: I am not asking how to apply grease.


    Here's the grease you typically see on laptops straight from the factory.
    They use the large pea method, then clamp the heatsink down.
    grease oozes out the sides, and surrounds the die completely.

    Many people have parroted that this 'insulates' the cpu.
    Where exactly is the insulation? around the sides where the grease oozed? Do you actually want heat to escape out the sides of the die? Doesn't that ooze prevent heat from escaping from the sides of the die, and transfer it directly to the heatsink?

    as I see it, the die has good contact with the heatsink, even with all that grease.
    The large pea method probably eliminated most air bubbles.


    [​IMG]
     
    Last edited: Sep 17, 2012
  2. -(Xyphox)-

    -(Xyphox)- [H]ardness Supreme

    Messages:
    6,527
    Joined:
    Sep 9, 2004
    Yes it matters, With thermal grease, less is better
    each processor is different on how to apply
    I usually put a very tiny drop on the chip and then speard it out to make a very thin even coat.
     
  3. Kyle_Bennett

    Kyle_Bennett HardOCP MasterChef Editor Staff Member

    Messages:
    39,958
    Joined:
    May 18, 1997
  4. ironforge

    ironforge [H]ard|Gawd

    Messages:
    1,083
    Joined:
    Feb 7, 2006
    That is waaay to much.

    Thanks Kyle for the link.
     
  5. dealmaster

    dealmaster Limp Gawd

    Messages:
    467
    Joined:
    Jul 12, 2007
    Thermal paste has horrible thermal conductivity compared to metals. Here are some values (higher is better):

    Aluminum: 200-250 W/(mK)
    Copper: 350-400 W/(mK)
    Silver: 400-450 W/(mK)

    Arctic Silver 5: 8.7 W/(mK)

    So looking at this, pure aluminum is over 20 times better at conducting heat than AS5. As bad as thermal pastes are, they're way better than air, which is an insulator. You want as thin a layer of thermal paste as possible to remove air gaps, but direct metal contact is best.
     
  6. Gorilla

    Gorilla [H]ardness Supreme

    Messages:
    4,967
    Joined:
    Sep 14, 2007
    One of these days someone should make a sticky on this subject. I swear we get this question or some variant of the "how do I apply it" question at least once a week.
     
  7. Mark111

    Mark111 n00bie

    Messages:
    7
    Joined:
    Sep 17, 2012
    Hi, new here!
    Since there's already a thread up about thermal paste I will ask here..
    I have just replaced my H100 high performance cooler with the stock cooler that come with it. But instead of cleaning the thermal paste of I just left it on.
    Will it be OK?

    H100 has to go back because it made a horrible rattling noise...

    Thanks in advance.
     
  8. DopplerEffect

    DopplerEffect [H]Lite

    Messages:
    72
    Joined:
    Aug 26, 2012
    Less is more when it comes to thermal paste.
     
  9. Gorilla

    Gorilla [H]ardness Supreme

    Messages:
    4,967
    Joined:
    Sep 14, 2007
    You want to apply fresh thermal past every time. If you don't remove the old stuff you will get air bubbles when you apply the new cooler, which means worse temperatures.
     
  10. dr.kevin

    dr.kevin 2[H]4U

    Messages:
    3,055
    Joined:
    Feb 17, 2006


    i'm not asking how to apply paste, or any variant of it.

    i'm asking about purported consequences of a sloppy paste job like the one on my laptop.

    I've heard all the "less is more" and 'too much paste insulates' sayings a million times.

    But what I want to know is where exactly is it insulating?
    When you clamp and screw down the heatsink, all the paste will be squeezed out enough to make the thinnest possible layer between the die and heatsink.
    After all is squeezed out, is it really insulating?

    so is the excess grease on the sides of the die increasing temps by a few C ?

    I have to wonder because I've read about other people using a better grease than factory grease, and only getting a 2C improvement after a clean paste job.
    A better paste alone would account for the 2C temp drop, so it looks like it doesn't matter whether you do it nice and clean, or sloppy like the factory.

    On the flip side, there are people who use so little and spread the paste out so thinly, only to later find out that the die and heatsink only made partial contact.




    There is no direct metal contact on the circled part of the cpu, where the excess paste is. If air is worse than paste, does it even matter that there is excess paste there?

    [​IMG]
     
    Last edited: Sep 17, 2012
  11. Gorilla

    Gorilla [H]ardness Supreme

    Messages:
    4,967
    Joined:
    Sep 14, 2007
    How much thermal paste (essentially what you are asking here) falls under how to apply thermal paste.
     
  12. rtangwai

    rtangwai [H]ard|Gawd

    Messages:
    1,056
    Joined:
    Jul 26, 2007

    A sloppy job has several consequences. If the paste itself is conductive, it may run when it is hot and short something out. In a laptop the fans may be below the heatsink so the paste in theory could run into the fan spindle (I have actually seen that once, but the person who was doing it was an idiot and thought you pasted CPUs the way you frosted a cake). If there is too much on the die surface then you are reducing the direct contact. The paste around the die should be harmless enough as long as it is non-conductive - there is so little surface area along the sides of the die I doubt it could radiate enough heat for the thermal paste to affect it in a a measurable way.
     
  13. ITSTHINKING

    ITSTHINKING [H]ard|Gawd

    Messages:
    1,260
    Joined:
    Mar 1, 2006
    Ithink I know the answer to this but the cheese was a joke right? The article was close to the beginning of the month of april. It gave me a great laugh! I have always been partial to AS5 i dunno just what I started with and it has never failed me (not to mention you can get it often thrown in to budles for a cheap price). OP: Less the better just spread the surface and none more is what I've always gone by, and I have had good luck. Good luck to you!
     
  14. wabbitseason

    wabbitseason [H]ard|Gawd

    Messages:
    1,516
    Joined:
    Jun 16, 2010
  15. ITSTHINKING

    ITSTHINKING [H]ard|Gawd

    Messages:
    1,260
    Joined:
    Mar 1, 2006
    Yes? That's a pretty shit response. Post count much? :rolleyes:
     
  16. wabbitseason

    wabbitseason [H]ard|Gawd

    Messages:
    1,516
    Joined:
    Jun 16, 2010
    That's all the answer you need. Whine more please.
     
  17. XacTactX

    XacTactX 2[H]4U

    Messages:
    4,062
    Joined:
    Dec 13, 2010
    I disagree with the theory that "it insulates the CPU". Air only conducts 0.04 W/mK. Your bog standard silicone grease conducts 0.6 W/mK. If you cover the sides of the CPU with thermal grease then more heat should get removed from that area (due to higher thermal conductivity).

    Now, the real question is, will it make a noticeable difference in the real world? Given that the sides of the processor are not attached to a heatsink, either way the end result should be negligible.

    Do you know what the problem is with using a ton of paste (like in the picture above)? The problem is that there is too much paste, and it doesn't allow the heatsink to completely clamp down on the processor. The smaller the gap between the heatsink and the processor, the lower the bond line thickness. The lower the bond line thickness, the lower the thermal resistance.

    If I had to choose between:

    1.) A large glob of paste that creates a thicker bond line and covers the sides of the processor.
    2.) A thin layer that has a thinner bond line and leaves the sides of the processor exposed.

    I would go with #2 every time.
     
    Last edited: Sep 20, 2012
  18. ITSTHINKING

    ITSTHINKING [H]ard|Gawd

    Messages:
    1,260
    Joined:
    Mar 1, 2006
    No thanks troll. Hope you use the same demeanor in the real world as you do here, you just might get bitch slapped! One could only hope. :D

    FYI: Yes isn't even an answer considering you only wrote the word yes and had no indication of who/what you were responding to regardless of the fact it was below my post. Thanks, for making my first/last visit in the OC subforum such a pleasant one. People like you are why some visitors never return. If that was your goal congrats! Carry on with the ways of the neckbeard. ;)
     
  19. wabbitseason

    wabbitseason [H]ard|Gawd

    Messages:
    1,516
    Joined:
    Jun 16, 2010
    I gave a response to the OP, not you 'ITSTHINKING'. You felt compelled to post an accusation that couldn't be farther from reality; I simply didn't have time for a long response and don't care one way or the other about post count. Yes is the answer to the OP's question and I'm not at all inclined, especially now, to explain to you why that is.

    It's your attitude that's the problem here, and you aren't even the OP.


    Edit: PM sent OP
     
    Last edited: Sep 21, 2012
  20. blade52x

    blade52x 2[H]4U

    Messages:
    2,834
    Joined:
    Sep 9, 2006
    Most factory TIMs are pretty crappy compared to the stuff we use. You can put a glob of something like MX-2 and be fine. What really matters are three things

    1) Contact area
    2) Thickness of TIM over the contact area (putting more doesn't necessarily mean you'll end up with a thicker layer)
    3) Quality of the TIM