Close

Register Now!

To Get More Info and Daily Reward.

Please login or register.
Pages: [1]   Go Down

Author Topic: Temperature question ???  (Read 3537 times)

0 Members and 1 Guest are viewing this topic.

Seranko

  • Guest
Temperature question ???
« on: 04-December-03, 22:26:36 »

Hey guyz,

I have a Thermalake Xaser III VM100A (very cool, by the way) with a water cooling system (I used my Thermaltake Aquarius II Liquid Cooling and changed it a little bit).  

The Xaser has a Temp. LCD Monitor with alarm and fan control.  I've connected the temperature sensor on one side of the CPU.  Also I've installed another Temp. LCD monitor (the same that has the case) to check the water temperature.  

Right now I have 31.5 C in my CPU and 30.6 C in my water, usually the water is 1 C above, but Core Center shows 47 C [CPU temp] and 41 C [sys temp].  Can somebody explain me why the difference ????  By the way, the temp. in the water is right, I cheched with a different digital termometer.

Thanks for your help,

Seranko
Logged

scottg26

  • Guest
(No subject)
« Reply #1 on: 04-December-03, 23:22:35 »

i have a feeling my temperature is also being reported incorrectly. i have thermaltake silent boost K8 CPU cooler, and corecentre says 49/50 degreesC idle, 56 degrees full load!!! yet the heat sink does not really feel that hot. its not too warm at all really.

Maybe BIOS 1.2 will fix this problem?
Logged

Seranko

  • Guest
(No subject)
« Reply #2 on: 04-December-03, 23:31:42 »

Scott26, where did you get BIOS 1.2 ????
Logged

scottg26

  • Guest
(No subject)
« Reply #3 on: 05-December-03, 00:07:54 »

what i meant was, maybe when BIOS 1.2 comes out it will fix this problem  :-))
Logged

Bibihest

  • SECOND LIEUTENANT
  • **
  • Offline Offline
  • Posts: 103
    • http://vivaldi.dk
(No subject)
« Reply #4 on: 05-December-03, 03:01:08 »
  • Best Answer

  • :-D)You get a higher reading from the CPU because it's inside the die and you probe is on the outside. That makes a big difference. 47 is maybe still a little high, but it WILL be higher in corecenter because of that.

    I was sanding my CPY last night :-D) and found out that the top of the CPU isn't 100% flat. Maybe some CPU's are worse than others. I have sanded both the CPU and stock cooler and i'm running without any cooling paste now and lost 5-6'c because of all those things.

    Maybe try to get another probe and compare the Xaser probe and the new one.
    Logged

    scottg26

    • Guest
    (No subject)
    « Reply #5 on: 05-December-03, 03:21:57 »

    wouldn't the water be a similar temp to the CPU?
    Logged

    TheChosn

    • Guest
    (No subject)
    « Reply #6 on: 07-December-03, 13:28:32 »

    Hey Bibihest,

    sanding sounds great :-P),
    Any step-by-step instructions for dummies?

    Greetz TheChosn
    Logged

    Bibihest

    • SECOND LIEUTENANT
    • **
    • Offline Offline
    • Posts: 103
      • http://vivaldi.dk
    (No subject)
    « Reply #7 on: 08-December-03, 02:09:24 »
  • Best Answer

  • Hmm not really, just be carefull :)

    I just jused some very fine sandingpaper for det CPU and was holdig it with a lot of toiletpaper so I wouldn't bend any legs and used a mirror as table. I think the heatplate is made of copper, because some of it got through when I was sanding the CPU. I only know the CPU wasn't very flat.
    I think my temps are better now. Lost 5-6'c on full load. But remember that the waranty is gone if you do it  :biggthumbsup:
    Logged

    riprjak

    • Guest
    (No subject)
    « Reply #8 on: 15-December-03, 22:10:04 »

    Quote
    Originally posted by Bibihest
    :-D)You get a higher reading from the CPU because it's inside the die and you probe is on the outside. That makes a big difference. 47 is maybe still a little high, but it WILL be higher in corecenter because of that.

    I was sanding my CPY last night :-D) and found out that the top of the CPU isn't 100% flat. Maybe some CPU's are worse than others. I have sanded both the CPU and stock cooler and i'm running without any cooling paste now and lost 5-6'c because of all those things.

    Maybe try to get another probe and compare the Xaser probe and the new one.

    Thermodynamics 101;
    NEVER attempt to carry out conductive cooling over a junction without a whetting agent!!!  ie: ALWAYS use a thin layer of small particle size, high thermal conductivity slurry between two solids where you plan to conduct heat with high efficiency.  NEVER use a thick layer of paste, a thin, uniform layer of a conductive slurry serves to fill the microscopic (and not so...) gaps and improve heat transfer across the junction; this is thanks to the fluid exclusion principle; the aim is not to have a layer of paste between the two sides of the junction, but have just enough to be displaced into the gaps between the surfaces.  No matter *HOW* well you sand, the use of a heat transfer slurry will result in dramatic improvements in heat transfer rate and stability.  In order, the preferred material are *SMALL* particles of carbon, silver, copper, aluminium or silica, nano-scale particles are best; but hardest to implement; micro-scale particles are most common...  ensure that the slurry is well mixed and free from agglomeration prior to application or you will create too large a gap.

    Sanding is a good idea if you dont have access to good quality thermal agent ("Arctic Silver" paste is an adequate product which is commonly available); I recommend making a slurry from carbon black and using that; the nano-scale particle size is an excellent whetting agent that will allow for improved heat transfer.  

    I use a sealed case (GMC NEO case) with only small air inlets cut iin the front sides; inlet of air driven by laminar flow fans; two laminar flow exhaust fans are used... and a Antec TruPower480W power supply; using unsanded stock athlon64 3200 and the included HSF; coupled with a carbon black/aluminium oxide slurry (this is actually an optical polising material and it is electrically conductive, so be BLOODY CAREFUL to mix it elsewhere and transfer it in a controlled environment).  Im in Australia where it gets a little warm ;) and my CPU under load peaks out at about 48ºC with a 31ºC ambient (in case) temperature.  Idle temperature is about 35ºC (as registered by the in-die thermistor) in the same conditions; which is not bad for forced-convection air cooling.

    err!
    jak.
    Logged

    Bibihest

    • SECOND LIEUTENANT
    • **
    • Offline Offline
    • Posts: 103
      • http://vivaldi.dk
    (No subject)
    « Reply #9 on: 16-December-03, 02:45:38 »
  • Best Answer

  • :-D)#ripjack

    *I'm not sure if your right about that. If you sand both the cooler and the proc and they fit 99,9% you will have more contakt than if you have paste between them. If I have to use paste in the small gaps, then I cant use any pasta anyways.

    I think my temps are great without paste, but I can try with paste and see if it gets any better  :-D)
    Logged

    scottg26

    • Guest
    (No subject)
    « Reply #10 on: 16-December-03, 06:43:29 »

    meanwhile my idle is still reported at 50 degrees C, lol  :-D)

    /methinks i need to go get te arctic silver and re seat the heat sink
    Logged

    cwigster

    • Guest
    (No subject)
    « Reply #11 on: 16-December-03, 07:42:23 »

    Quote
    Originally posted by scottg26
    meanwhile my idle is still reported at 50 degrees C, lol  :-D)

    /methinks i need to go get te arctic silver and re seat the heat sink

    I'm not so sure about that. When I first ran my system the CPU reported temps in the low to mid 30C's. When I flashed to the 1.10 bios I noticed it was reporting the CPU at 60C!!. Every so often it still does this, but generally it seems to report mid to high 30C's now. So I think there could be something flakey with the monitoring code in the bios or the censor on/in the CPU is faulty, or a combination of both.

    By the way, my heatsink is using the bog standard paste that came pre-applied to the heatsink.

    Craig
    Logged

    riprjak

    • Guest
    (No subject)
    « Reply #12 on: 16-December-03, 17:59:20 »

    Quote
    Originally posted by Bibihest
    :-D)#ripjack

    *I'm not sure if your right about that. If you sand both the cooler and the proc and they fit 99,9% you will have more contakt than if you have paste between them. If I have to use paste in the small gaps, then I cant use any pasta anyways.

    I think my temps are great without paste, but I can try with paste and see if it gets any better  :-D)

    True, *IF* they fit 99.9%, paste would be not needed.  However, even with DAYS of precision mechanical sanding followed by polishing, if you acheived better than 85% surface contact I suggest writing a paper on your technique as your method exceeds the best optical polishing techniques currently available.

    However, if you are just sanding,  I severely doubt that your contact would be better that 70~75%.  This is assuming a 2400grit paper.

    Remember, you are dealing with a relatively rough surface placed on another relatively rough surface, surface variations on the scale of micrometers will create air gaps.  Thermal paste is designed to fill these; the layer of paste MUST be very thin or strongly shear thinning (to ensure it doesnt CREATE a gap, merely fills those which actually exist).

    Properly applied thermal grease, tape or paste does *NOT* create a gap, it only fills gaps which exist anyway.  Sanding will not remove these.  Sanding and Polishing will not remove these.

    I strongly doubt that manual sanding actually improves things anyway; I would tend to suggest that it makes things worse as you are likely using a compliant surface with no means to ensure you are planing the surface level.

    ;) Apologies for the lecture; but thermo and fluid dynamics is what I do for a living.

    Of course, the monstorously low efficiency of heat sink mounted fans (great big dead spot in the middle) doesn't help the situation.

    In fact, the *BEST* heat transfer across a junction reported in literature to date was acheived with *FLAT* (as in machine ground to extreme flatness flat) but uniformly rough surfaces with a high contact force and universal treatment with a slurry of carbon fines.

    err!
    jak.
    Logged

    riprjak

    • Guest
    (No subject)
    « Reply #13 on: 16-December-03, 18:11:36 »

    Quote
    Originally posted by cwigster
    Quote
    Originally posted by scottg26
    meanwhile my idle is still reported at 50 degrees C, lol  :-D)

    /methinks i need to go get te arctic silver and re seat the heat sink

    I'm not so sure about that. When I first ran my system the CPU reported temps in the low to mid 30C's. When I flashed to the 1.10 bios I noticed it was reporting the CPU at 60C!!. Every so often it still does this, but generally it seems to report mid to high 30C's now. So I think there could be something flakey with the monitoring code in the bios or the censor on/in the CPU is faulty, or a combination of both.

    By the way, my heatsink is using the bog standard paste that came pre-applied to the heatsink.

    Craig

    Oh yeah, if you dont know what you are doing (and I am talking know as in the ability to accurately mathematically model the outcome before attempting it); Craig's approach of using the thermal tape pre-applied is ideal.

    This is a bloody expensive toy to risk voiding your warranty because you think you know better than the engineers who designed AMD's solution; particularly since they have designed a basically robust and reliable solution.  I only messed with it because the ambient temperature in my computer room at home climbs as high as 40ºC in the height summer; so I could spend thousands on extra air conditioning in a rental property, or tens of dollars ensuring that my PC could run at full load, with an ambient temperature of 42ºC and keep the CPU under 60ºC and whack in a pedistal fan to blow cold air from the lounge room onto me :)

    If your computer lives in an ambient environment that never gets over say 30ºC, then *DONT* mess with the thermal tape already applied, its just not nescessary.

    With stable hardware (ie RAM and Videocard) configuration, a processor at its factory clock speed should remain stable up to ~60ºC; although this is probably less true with a brand new product.  Most instability I have encountered comes, in order, from "generic brand" ram, poor quality power supplies, insufficient airflow through the case (ie: the air inside the case just keeps getting hotter, you cant make something cooler than the surrounding atmosphere without a phase change people) and systems which were assembled without proper anti-static precautions.

    err!
    Jak.
    Logged

    scottg26

    • Guest
    (No subject)
    « Reply #14 on: 17-December-03, 05:56:48 »

    Im using the Thermaltake Silent Boost K8 CPU cooler. i knew when i bought it it wasn't the best for cooling, but i thought it would be better than this. but then bios, speedfan and corecentre all report different temperatures so im a bit confused  :-D)
    Logged
    Pages: [1]   Go Up