to change or not to change thermal paste?

Discussion in 'MacBook Pro' started by vonb3ta, Aug 5, 2010.

  1. vonb3ta macrumors newbie

    Joined:
    Aug 5, 2010
    #1
    Firstly, I would like to say a thank you to all the help this forum has given me (as a reader, not member) over my year of owning a macbook pro. (Mid 2009, 2.8Ghz, 9600GT 512mb, 7200RPM HDD)

    Basically I am a computer savvy Electronic Engineering student, but am wondering if there is a point changing the thermal paste in my mbp.

    Everything ran great until this one time a got an EMERGENCY temperature shutdown while encoding video, (running windows 7 at the time in bootcamp). I investigated, and then couldn't understand why I was idling at 70 and got under loads around 105 celsius.

    Noticing that my right-fan was reading 0 rpm, and warranty had ran out. I opened it up and found some debris in the right fan, blocking it all together. After having removed this, the fan worked again.

    some tests gave me results of:
    45-55 idling with Chrome open in 10.6.3 with fans at 2900-3000RPM
    90 with a 103 peak in Starcraft 2, all settings high 1280x800 in Windows 7 with fans at 6000RPM
    and yes > /dev/null peak at 77 degrees C, with fans at 5500RPM

    Now I have looked into solutions to minimize heat and have ordered the BlueLounge Cool Feet, and have cleaned out my fans completely.

    Being a bit of a perfectionist, as well as wanting great temperatures. I was wondering whether anyone though it would be worth changing the thermal paste. I have a ton of Arctic Silver laying around and all the tools... but is it worth the risk? I have disassembled and assembled PCs and Laptops before, but I am slightly more worried about a MBP than a cheap netbook.

    Some say there is not much change, others see a shift of 10 degrees.

    So my question to you guys is:
    Is it worth it? Does the risk of doing this outweigh the benefit? and does my Macbook Pro have a temperature problem?

    Again, sorry for the long text and question. Thank you to anyone who can offer me advice. I am currently torn between doing it or not :)
     
  2. ayeying macrumors 601

    ayeying

    Joined:
    Dec 5, 2007
    Location:
    Yay Area, CA
    #2
    Those temperatures are normal. I wouldn't change the thermal paste since it won't make a huge difference at least with your temperatures currently.
     
  3. Meever macrumors 6502a

    Joined:
    Jun 30, 2009
    #3
    A good paste only costs like 15 bucks and it would help by a few degrees. And if you know how to and not worried about warranties then go for it.
     
  4. Schtumple macrumors 601

    Schtumple

    Joined:
    Jun 13, 2007
    Location:
    benkadams.com
    #4
    If you're 100% sure you know what you're doing, go for it, Apple normally does a pretty crappy job with thermal paste, my MBP was absolutely covered in the stuff all around the CPU.
     
  5. Pax macrumors 6502a

    Joined:
    Dec 12, 2003
    #5
    A few points in a very long post....

    Re paste quality, you have got to look at the temperature difference between the CPU and its heatsink under full load. Run 2 x yes > /dev/null if you have a dual core, 4x if you have a quad core. You should see CPU temps the same as you saw with StarCraft, say 95 C. I expect your heatsink temp will be at 70-75 C, typical of most "stock paste" Macs I've seen. If you have a C2D you've got 25 W TDP, therefore ~1 C/W of thermal resistance from the stock paste (if you have an i CPU you've got 35 W). At idle the CPU draws something like 5 W, so your temperature difference will drop to about 5 C.

    If your thermal resistance with "stock paste" is much bigger than 1 C/W then changing paste might be a good idea. From what I've seen you can reduce it a bit, maybe to 0.6 C/W with really good paste perfectly applied. That will reduce your idle temperature by 2-3 C.

    BUT, assuming the tests above to show your paste to be "pretty good" eg 1 C/W.....

    I am pretty sure Apple's fan control algorithm is based upon CPU temperature. You can do anything you like to the paste, the Mac will control the fan speed to settle the CPU at 95 C under full load. So changing the paste will simply slow the fans. Ironically, I think this will heat up the rest of the Mac's components - the temperature difference between heatsink and CPU will be smaller, but the CPU will still be at 95 C, so the heatsink will get hotter (Say 85 C instead of 75 C) with fans running slower. This will dump more heat into the Mac's interior and make the case hotter.

    So changing paste - it depends what you want to achieve. "Great temperatures" doesn't really mean that much given the Mac fan control algorithm.

    New paste could reduce idle temperature by 2-3 C. But IMHO who cares? The CPU idles at a very low temperature as it is. I think new paste won't change your full load temperatures, it will simply reduce the fan speed under full load, and heat up the Mac's interior/casework more than before.

    If you want lower full load temperatures you need to attack the fan control algorithm, eg through SMCfancontrol or similar, to make the fans run faster and the CPU max out at less than 95 C.

    Having said all that, part of me would really like you to change your post with detailed before-and-after information, say screenshots of iStat Pro at idle & full load before and after :).

    NOTES:

    1. I suspect there are minor differences in thermal resistance depending on die size which varies from CPU model to model, and also the TDP is IIRC an upper limit and will vary from chip to chip. But around 1 C/W seems typical for various Macs I've seen.

    2. your temp of 77 C under yes > /dev/null seems too low. Perhaps you didn't run one instance per core? I would expect to see about 95 C, the same as with StarCraft.

    3. Your idle fan speed seems odd - if your Mac was truly idle (Activity Monitor showing just a couple of percent CPU load) the fans should be rock solid at 2000 rpm unless some other component is hot. Normally the fans sit at 2000 rpm until the CPU temp is well into the 90s. Maybe something else in your Mac was hot at the time (enclosure? HDD?) causing the fans to spin up. Again posting screenshots from iStat can be helpful to see what the fan speeds were and why.
     
  6. vonb3ta thread starter macrumors newbie

    Joined:
    Aug 5, 2010
    #6
    Firstly I would like to thank everyone what has given me advice about this matter. It appears from what everyone is saying that I have a pretty standard MBP.

    I am however sorry that I didn't add that I use SMCfan control under OSX and Lubbo's Fan control tool under Windows 7. therefore automatically rasing fan speeds as temperature goes up.

    Here is SMCfan control off - idle (I realised I still have Fan Control Installed, will re-post when that is done)
    [​IMG]

    Will post the rest load results soon :)
    Depending on what I get, I will probably change thermal paste. But in that case I will also record before and after results.
     
  7. Pax macrumors 6502a

    Joined:
    Dec 12, 2003
    #7
    Here's my falsifiable prediction..... (with no SMCfancontrol & assume GPU idle)

    before paste change
    - at idle, delta between CPU and its heatsink, 5 C (we've already seen that, so I admit my Nostradamus skills are not great....)
    - at full load (2x yes > /dev/null), CPU settles at 95 C, CPU heatsink at 70 C, fans at 4000 rpm, enclosure at high 30s

    after paste change
    - at idle, delta CPU-heatsink, 3 C
    - at full load, CPU settles at 95 C, CPU heatsink at 75 C, fans at 3600 rpm, enclosure 2-3 C warmer that before

    so after change at idle CPU will be 1-2 C cooler and at full load heatsink 5 C warmer, fans 500 rpm slower, enclosure 2-3 C warmer.

    :)
     
  8. vant macrumors 65816

    Joined:
    Jul 1, 2009
    #8
    Ignore that prediction, it has absolutely no useful data.

    The end result depends on your ability to reduce bubbles during application, area cleanliness, and the type of compound you use. You sound computer inclined, so I would predict that you would notice a difference.

    I strongly suggest you use Ceramique over AS5. AS5 has dried on me twice already, but my Ceramique applications have been going strong. My applications tend to be on the 'very hot overclocked' side, which speeds up thermal compound failure.
     
  9. vonb3ta thread starter macrumors newbie

    Joined:
    Aug 5, 2010
    #9
    given the choice between Arctic Silver and Ceramique, (quick I both have in quite large quantities), as you have pointed I have noticed that the Arctic Silver dries up on the processors I have used it on. (watercooled/overclocked AMD Phenom processors)

    I hesitate however to use ceramique due to the fact that it is less thermally conductive and hence may not be worth the effort of changing thermal paste.

    I have worked on may PCs, built many setups and replaced many laptop motherboards in my time. What I fear however are the "ribbons" found in the MBPs

    (as seen in step 18 here: http://www.ifixit.com/Guide/Repair/Installing-MacBook-Pro-15-Inch-Unibody-Mid-2009-Heat-Sink/1733/6)

    The only times I have delt with these are with keyboards and iPod reparations/replacement but these seem common in the MBP. Are these a big hasle?
     
  10. Pax macrumors 6502a

    Joined:
    Dec 12, 2003
    #10
    If you are at all worried about it I would suggest not doing it unless you are sure you have a paste problem.

    This guy replaced the paste on his C2D with Ceramique and got a and got a thermal resistance of about 0.8 C/W. Which is hardly better than my C2D with its "stock" paste of 0.9 C/W
    http://forums.macrumors.com/showthread.php?t=932922
    Which I think shows that the thermal performance of Apple's stock paste is nearly as good as that of Ceramique.

    Of course YMMV and if your stock paste has a very much higher thermal resistance it could well be worthwhile. But from what I have seen across a number of recent Macs the thermal resistance is very,very similar from machine to machine. So I doubt yours is much different to mine.
     
  11. alphaod macrumors Core

    alphaod

    Joined:
    Feb 9, 2008
    Location:
    NYC
    #11
    I replaced my thermal paste with Shin Etsu thermal material; it's a bit runny compared to Arctic Silver, but I think it works pretty well.

    This is what my motherboard looked like when I got it:
    [​IMG]
     
  12. vonb3ta thread starter macrumors newbie

    Joined:
    Aug 5, 2010
    #12
    Ok so I finally "man'ed" up to it. Went to iFixit.com (great website! :D) got the manual and was on my way. Everything went smoothly with no hiccups and none of the wires/ribbons where any trouble to remove.

    Using a screw/pill sorting rack, I meticulously sorted every screw according to every step on the guide. (I greatly recommend doing this!) Thus I had no confusion when it came to re-assembling.

    My assumption to why it was so easy to disassemble was because someone in China has to assemble it in the first place. The more space for error there is the greater the change of messing up hardware. Which is of great cost to the manufacturer.

    One of the only places the people actually have freedom to muck stuff up, is in how much thermal paste they wish to apply. I assume they see no stock options or do not have any benefit in being careful as most of them will never have personal benefit from doing so.

    Once I got the heatspreader off, I found the thermal paste looking like this:
    [​IMG]
    [​IMG]

    Not necessarily too much, nor too little. But still on the + side. Moreover I suppose that the constant low and high temperature shifts on a laptop causes the Thermal Paste to "age" faster than it would in say a desktop.

    Using ArcticCleaner compound, I was able to clean up the old stuff and "purify" the surface.
    [​IMG]

    Now desciding between Arctic Silver 5 and Ceramique, I went with Ceramique. Partly because of my past experiences with Arctic Silver acting as Glue, as well as reports that Cermamique lasts longer.

    [​IMG]

    Now it is all up and running, but I will post my Results later on tonight. So far so good, seeing average 5-7 less degrees :D
     
  13. Pax macrumors 6502a

    Joined:
    Dec 12, 2003
    #13
    Didja do a full load test with the old paste?, it will be interesting to see results
     
  14. belvdr macrumors 603

    Joined:
    Aug 15, 2005
    #14
    Are you serious? That's way too much. It shouldn't be pouring out the sides like that and smothering the rest of the chip. Also, it depends on how often the desktop is shutdown or slept too. ;)

    I use even less than that. The entire purpose of the paste is to fill in gaps in the surfaces so heat is conducted faster and easier across the bridge, so there shouldn't be more than very thing coating to start with.
     
  15. vonb3ta thread starter macrumors newbie

    Joined:
    Aug 5, 2010
    #15
    For the thermal paste, I applied as thin of a layer as I deemed reasonable. I made sure to avoid too much so as to not pour over the chip. I did have a little "run" but only as I was scraping excess of using a plastic card. I tried clearing that up as best I could.

    I agree however that the original thermal paste application was excessive. But not uncommon for commercial laptops and desktops.

    As for Results :p

    Before:

    IDLE
    [​IMG]
    LOAD:
    [​IMG]

    After (with reproduced load fan speed)

    IDLE
    [​IMG]
    LOAD
    [​IMG]
     
  16. belvdr macrumors 603

    Joined:
    Aug 15, 2005
    #16
    I don't think I'll be tearing mine apart for a 5 degree differential (at least not until AC runs out). :) Good job!
     
  17. Pax macrumors 6502a

    Joined:
    Dec 12, 2003
    #17
    So no change to the thermal resistance, temperature difference between heatsink and CPU is largely unchanged, maybe a couple of C better:-
    idle, before 6 C
    idle, after 6 C

    full load, before 26 C
    full load, after 24 C

    Thermal resistance of CPU to heatsink interface....
    before, stock paste, 26/25 = 1.04 C/W
    after, Ceramique 24/25 = 0.96 C/W

    So Ceramique has 8% lower thermal resistance than stock paste

    Your absolute CPU temperature is slightly lower in the "After" case, but some of that is because the heatsink is cooler, maybe because the fans are running 100 rpm faster? If the ambient temperature is the same, the fans are at the same speed and the CPU is generating the same amount of heat, then the heatsink should be at the same temperature before and after the change.

    Interesting stuff - glad it all worked out for you. Thanks for the info
     
  18. belvdr macrumors 603

    Joined:
    Aug 15, 2005
    #18
    If the paste conducts the heat more effectively (due to various reasons), the heatsink may actually be warmer.
     
  19. vonb3ta thread starter macrumors newbie

    Joined:
    Aug 5, 2010
    #19
    glad I finally got it over ;)
    It was easier than I expected, but I would like to thank everyone who contributed. .... So thanks :)

    If you have any questions please don't hesitate to ask
     
  20. AlphaDogg macrumors 68040

    AlphaDogg

    Joined:
    May 20, 2010
    Location:
    Boulder, CO
    #20
    I am happy to see my thread being referenced on the forums. Thank you.
     
  21. paintballswimgu macrumors 6502

    Joined:
    Feb 13, 2010
    #21
    [​IMG]

    you need about 1/5 th that amount, on a full size desktop you want about a half grain of rice, your just trying to fill in the minor cracks and crevaces, not put a layer between anything. You should have used artic silver as well, it cools much better. I was able to drop my load temps on my i7 930 by 6 C just by switching.
     
  22. vant macrumors 65816

    Joined:
    Jul 1, 2009
    #22
    As noted earlier, AS5 has drying problems. Tests have shown that AS5 is not that much better than Ceramique. In fact, in some cases Ceramique is better and likewise. Either way its a very small difference.
     
  23. sporadicMotion macrumors 65816

    sporadicMotion

    Joined:
    Oct 18, 2008
    Location:
    Your girlfriends place
    #23
    I just saw this thread and here's my 0.02

    I was hitting 92c under yes > /dev/null

    Here is is now after using Arctic Silver MX-3. This is after running at 100% for 35 minutes. It was fluctuating between 69-71.

    (2.66 C2D 9600m GT 256)
     

    Attached Files:

  24. ouimetnick macrumors 68020

    ouimetnick

    Joined:
    Aug 28, 2008
    Location:
    Beverly, Massachusetts
    #24
    I have had Arctic Silver 5 on a laptop and when I heard that AS5 dries, I checked it for my self. 4 years later, its still thick n creamy. :p After heavy use for 4 years, it hasn't dried out.
     

Share This Page