Re: checking temperature on Radeon HD 4350 graphics card

Discussion in 'Computer Information' started by Paul, May 20, 2011.

  1. Paul

    Paul Guest

    Fredd Wright wrote:
    > "Fredd Wright" <> wrote in message
    > news:...
    >> I have had an ongoing problem with my PC shutting down. I just found out
    >> this is due to my Radeon HD 4350 graphics card is overheating. The card
    >> has a fan on it (not a heatsink) which is running. I did have another card
    >> installed right "below" the graphics card (by "below", i mean next to the
    >> card but on the opposite side of the fan). I took the card out since it
    >> wasn't being used. Could that have been the source of the overheating? If
    >> not, what else could it be?
    >> Does anyone know how i can monitor the temp on my graphics card? I
    >> downloaded ATItool but it's freezing on me plus before it froze it showed
    >> a temp of -1C so i think it's not recognizing my card. Does anyone know
    >> of any other utilities i can use to monitor the temp?
    >> Fredd

    > More info. I downloaded GPU-Z and the temp is reading 95C with a load of
    > 8%. Also, the fan speed reads 58% but at -RPM (it is running). This seems
    > like way too high a temp for the card to run at. Especially with no load.
    > Any ideas?
    > Fredd

    Physically check the fan.

    It has likely stopped spinning. The fan is tiny, and sometimes even a
    wire jammed in the fins, is the cause of the problem.

    GPU-Z is exactly the right tool for the job, so you know you have a problem.
    The card should not run 95C at idle, so it isn't getting cooled properly.

    On some graphics cards, when the fan fails, the chip can get so hot,
    that plastic parts melt. I hope your card has not damaged anything,
    and you can work on replacing the fan or freeing up rotation with
    a little oil or whatever. The longer you run the card in an
    overheating condition, the more likely something will melt.

    The fan can be delivered a variable voltage by the card. In some
    cases, the power circuit feeding the fan cable has failed in some
    way, and no voltage is present. If that happens, you can look at
    operating the fan from a Molex disk drive connector.

    For example, my video card right now, the cooling fan power
    connector has been disconnected from the video card itself,
    and is being fed through a voltage adjustment circuit, directly
    from the power supply. I've been running it like that for over
    a year now. Doing it that way, prevents the software-controlled
    fan mechanism from making a nuisance of itself. By observing how
    the original setup ran the fan (fan speed under gaming conditions),
    I reproduced the conditions exactly with my voltage adjustment circuit,
    and get identical cooling. My card runs perhaps 45C (it's a different
    card than yours, but I also use GPU-Z like you, to verify the temps).

    If it's just a fan failure, then you can come up with some scheme to
    cool the card. There are companies that make after-market coolers,
    which replace the entire copper heatsink assembly, so that is another
    way to do it.

    Before buying an after-market cooler, there are *many* mechanical
    details and dimensions to check. This is *not* as simple as it
    seems. And the web site of the manufacturer, doesn't always make
    it easy to figure out. If you look at the video card itself,
    you'll see a mechanical outline printed in the white silk screen,
    and that captures the ATI or Nvidia recommended mechanical footprint
    for the cooler. As long as the aftermarket cooler screw holes
    line up, it might just work. Care should be taken, to make sure
    the product also cools the video card memory chips - some of these
    products don't get enough cooling effort to the memory chips, and
    the memory chips are "insulated" by the cooler resting just above

    So fixing your fan, is the "least risk" way, but if the fan is
    melted, or you have doubts, there are other ways to deal with it.

    Paul, May 20, 2011
    1. Advertisements

  2. Paul

    Paul Guest

    Fredd Wright wrote:
    > Well, the fan was working. The temp kept rising to 118C at which point the
    > PC shut down. I got a new graphics card (EAH5450). Not a great one but it
    > does the job and so far it's holding at about 60C with minimal load. Still
    > seems a little high to me but much better than the other one. I'm guessing
    > the other card had a short or something.
    > Fredd

    Video cards can have two switching regulator circuits on them. One to power
    the core of the GPU (around 1.0VDC or so, at a high current). The second one
    to power the RAM chips. If the GPU Vcore were to rise significantly above
    1.0V, that could make the chip hot. I'm not really familiar with what
    setting and measurement capabilities exist, but there may be some
    kind of application (Rivatuner or the like) which can access it.
    If the regulator is defective, then the setting may say 1.0V, but
    some other voltage could be coming out.

    Something else you might inspect on the video card - is how well
    the pins or screws are holding the heatsink to the surface of the
    GPU. If the heatsink is loose, or there is an air gap between
    the GPU and the cooler, that can cause the temps to skyrocket.

    Paul, May 21, 2011
    1. Advertisements

  3. Paul

    Paul Guest

    Fredd Wright wrote:
    > I took off the fan. It seemed to be on fine. It was directly on the GPU
    > attached with thermal compound. Upon further inspection, i noticed that one
    > of those little copper teeth that plug into the PCIe slot had about 1/3 of
    > it chipped off (that is the tip of it was missing). I wonder if that was
    > the problem. Perhaps one of those regulator circuits you mentioned wasn't
    > working because of that.
    > Fredd

    The front section of the PCI Express (nearest the faceplate), carries power.
    In the pictures here, you can see which pins on the component and solder
    side carry 12V and 3.3V.

    I wouldn't expect that would make any difference, except if the teeth
    were burned off and the card could no longer get power.

    If the rails got shorted together, then more things than just the
    video card would be upset. (A good power supply, would notice out of
    range current or voltage if that happened, and then the supply would
    switch off pretty well immediately.)

    The video card chips, don't use those power rails directly. There
    is a tendency instead, to use switching regulators to convert the power.
    The 3.3V rail is low enough in voltage, to power some CMOS circuits,
    but I think for power reasons, they like to use the lower voltages
    if there isn't a communications need for the higher voltage 3.3V source.
    12V on the other hand, is just too high to be used directly, so some kind
    of switcher would be in the path there.

    In any case, I doubt a copper tooth would be doing it.

    Paul, May 21, 2011
    1. Advertisements

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

It takes just 2 minutes to sign up (and it's free!). Just click the sign up button to choose a username and then you can ask your own questions on the forum.
Similar Threads
  1. Silverstrand

    PowerColor 128MB Radeon X800 GTO Graphics Card

    Silverstrand, Feb 25, 2006, in forum: Front Page News
    Feb 25, 2006
  2. zxcvar

    CD Rom in the Kodak Dx 4350

    zxcvar, May 11, 2004, in forum: Digital Photography
    May 12, 2004
  3. zxcvar

    For Mr. Ron Baird - Lens adapter for Dx 4350

    zxcvar, May 12, 2004, in forum: Digital Photography
  4. PAUL

    a fair price for an IPAQ 4350

    PAUL, Feb 14, 2005, in forum: NZ Computing
  5. Ian

Share This Page