P5K WS: CPU fan speed control and CPU temperature

Discussion in 'Asus' started by ken k, Sep 22, 2008.

  1. ken k

    ken k Guest

    I have installed a new P5K WS motherboard with a Sunbeam Core-Contact
    Freezer cooler. The fan for the cooler is a bit loud (it runs at 2000
    rpm); unable to see how to modulate the fan speed, I disconnected it
    for a test. Running BurnIn for 60 minutes with the cooler with the
    fan blowing through the tower, the idle temp raised from 29 degrees to
    33-34C and held. Running without the fan, the temperature is about
    44C. The MB temp stays about 39 degrees (temperatures are from PC
    Probe II)

    First of all, is 44 degrees terrible? I am running BurnIn for the
    CPU (maths and SIMD [whatever that is]), memory, video 2D and 3D at
    100 per cent) I am thinking that temps in the 65+ degree range are
    not good.

    Second, is there a way of controlling the speed of a CPU fan on this
    model motherboard through the motherboard itself? If not, is there a
    reliable 3rd party fan controller? (the control provided by
    Sunbeamtech does not work when added between the board and the fan for
    the cooling device)

    Thanks
    Ken K
     
    ken k, Sep 22, 2008
    #1
    1. Advertisements

  2. ken k

    Paul Guest

    Generally, we're only interested in the peak temperature, under load.
    Under no load, the CPU will run at some more reasonable temp.

    So, run Prime95 with a thread per core, and see how hot it gets. As you
    say, a CPU temperature of under 65C under load, is desirable. If you go
    too much higher, the Intel processor will start to "throttle". A program
    like RMClock can be used to detect throttling.

    http://www.mersenne.org/gimps/p95v256.zip

    You can see RMClock detecting throttling, here.

    http://www.xbitlabs.com/articles/cpu/display/oc-guide_7.html

    The fan controller on the Sunbeam, is shown as a three pinner. You should
    have been able to plug that into a four pin CPU fan header (leaving the
    PWM pin unconnected), and then the adjustment knob should have worked.
    You could test, using a three pin motherboard header, and plug in one
    of your case fans to it.

    The motherboard usually wants to see pulses on the CPU fan header and its
    RPM signal. The BIOS may shut the board off, if it thinks the fan has failed,
    or there is no fan present. You could, if you wanted, plug the Sunbeam
    fan controller, into another (three pin) fan header, but you'd want to
    plug some other monitored fan into the CPU header. So whatever you
    decide to do, there should be a fan with a working RPM signal, plugged
    into the CPU header.

    They list a ridiculously low theta_R for this cooler, so I hope it
    actually achieves those numbers in practice.

    http://www.sunbeamtech.com/PRODUCTS/cooler/Core-ContactFreezer.html

    For example, say you had a 100 watt TDP processor. The theta_R is
    listed as 0.092C/W . Say the room temp is 25C, and the computer
    case air temperature is 35C (a relatively well cooled computer case).
    Then the CPU temp would be 35C + (100W * 0.092C/@) = 44.2C .
    That can only happen, if the flatness of the two surfaces
    is such, that the heatpipes are making good contact. I
    remember seeing a review for some other product, that uses
    the "raw heat pipe" contact technique, and it actually sucked
    at it. So load up Prime95, and do the math for whatever processor
    TDP you've got, and see if the thermal performance matches.
    Based on the math, you should have no trouble meeting 65C,
    and should be able to turn the fan down.

    I make my own, home made fan controllers. I get a pack of
    1N4000 series diodes, at a local electronics store. They cost
    about $0.10 each. Each diode drops about 0.7V when forward
    biased. I put six or seven diodes in a row, and connect to
    +12V, to get a reduced voltage for a fan. The tough part,
    is packaging the solution. A little piece of perfboard would
    likely do a better job, than what I put together :) I like
    the ghetto solutions. And no, it isn't adjustable (except
    by adding or removing diodes). I suppose I could put a
    very expensive wafer switch, to make them selectable,
    but then the price would be well outside the $0.70 total
    so far.

    Zalman makes the FanMate2 adjuster, but I believe it has
    a 6W limit (12V @ 0.5A). The Sunbeam current is under
    that value, so it would probably work. But I'd rather see
    more choices than that in the industry. Using a bulky
    tray mount solution (three channel) for this application,
    is overkill.

    Paul
     
    Paul, Sep 22, 2008
    #2
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.