Muffled sound P4P800 (heat)

Discussion in 'Asus' started by Navid, Dec 5, 2004.

  1. Navid

    Navid Guest

    After playing a game for 20 minutes, the sound changes and is not crisp
    anymore. So, I have to increase volume.

    I have Audigy2 ZS on my P4P800 deluxe and have disabled the on-board audio
    in the bios.
    Before getting my Audigy sound card, I used the on-board audio and I had the
    same problem even then!
    Because sound changes after playing a game for a while, I am assuming this
    is a heat problem.
    Am I wrong?

    My speaker system is Logitech Z-640.

    What is heating up that causes this?
    I need to know to improve its cooling.

    Motherboard Monitor shows the CPU (P4C 2.6 @3.17) and case temperatures to
    be at 30 and 34 at idle and at 41 and 37 under load.

    Thanks,

    Navid
     
    Navid, Dec 5, 2004
    #1
    1. Advertisements

  2. Navid

    Paul Guest

    Here is a hypothetical sound path:
    V+
    +5V |
    | R1
    Sound_chip C1 C2 |
    DAC -----||----jack,plug------||-----+-- rest of amplifier
    | |
    GND R2
    |
    V-


    The DC blocking capacitors, C1 and C2, prevent the DC level of
    the sound chip, from upsetting whatever DC bias network, like
    R1 and R2, are on the amplifier.

    For sound to work properly, neither side of the circuit can
    "bump its head". On the output of the DAC, a 0x0 output code
    should give 2.5V. For 1Vrms output, the DAC can go to 3.9V
    and as low as 1.1V. The difference between those two extremes
    is 2.8V and to convert from Vpp to Vrms, the conversion factor
    is 2*SQRT(2) or 2.8 .

    Say the DAC tries to make a voltage higher than 5V. It cannot,
    because it is supplied with 5V. If the DAC cannot swing to the
    output requested, it "clips", and the sine wave on the output
    has a flat spot on top. This would be heard as a distortion,
    and could be what you are hearing.

    The same is true on the amplifier side. The bias network
    centers the input somewhere between the two rails. As long
    as the sine waves don't get too close to V+ or V-, the
    input stage won't distort.

    If I were to guess as to which end is overheating, I'd say
    the speakers are doing it.

    Test the speakers by running them from an audio source, like
    a CD player, for an extended period, and see if you hear
    distortion after the speakers reach their thermal peak.
    Since these are amplified speakers, the heat dissipating
    elements are inside the speakers, and it is hard for the
    heat to get out, unless a heat sink is mounted on the back
    of the amplifier unit.

    I find it hard to believe that both of your computer sound
    solutions are failing the same way.

    Only more "mix and match" testing will help you identify
    which component is really at fault.

    It could be that the audio engine in the game sucks. That
    has been known to happen. When audio channels are mixed in
    a game, the values are added together, and the same dynamic
    range issues can happen, if the added values go too far for
    the DAC to follow them.

    Taking the audio output (Lineout) from the affected computer,
    and recording on the Line_In of a second computer, then
    looking at the results with a waveform editor, might show
    the distortion. But, unless the test conditions are simple
    waveforms, you'll have a hard time identifying a problem that
    way. Waveform recording is fine for looking at clicks and pops,
    but unless pure sine waves are used for testing, you'll
    have a hard time identifying a distortion by eye.

    HTH,
    Paul
     
    Paul, Dec 6, 2004
    #2
    1. Advertisements

  3. Navid

    Navid Guest


    Paul,

    Thanks for the help.

    I have noticed the problem with my older 2.1 speakers also.

    The games that cause it are Far Cry, Halo and Counter Strike. They are not
    games that no one has heard of.
    I am now playing Half-Life 2 and have not noticed the problem while playing
    it yet.

    The MOSFETS on the P4P800 motherboard get extremely hot.
    Could that contribute to the problem?
    I guess, based on your graph, the answer is no, since the analog portion of
    the path starts on the sound card and never is on the motherboard. Is that
    correct?

    Thanks,

    Navid
     
    Navid, Dec 7, 2004
    #3
  4. Navid

    Paul Guest

    My comment about the games angle, is I have heard of some games
    that run out of memory, or have memory management issues with the
    sound engine. Then, funny things start to happen, like the
    sound cutting out. As long as you don't see reports in Google,
    of the above mentioned games having problems with sound, then
    you can dismiss that as a source of trouble. I do remember
    sound quality issues with the A7N8X after long periods of
    time, but that could have been a sound driver issue as well.
    There were a lot of complaints about sound on Nforce2
    motherboards, and it took Nvidia a long time to fix them.
    nforcershq.com should have lots of posts about that.

    I've noticed at least two MOSFETs on my P4C800-E get hot. I
    checked them with a thermistor, and the surface temperature is
    about 45C when room temp is 25C. If you hold a finger on them,
    they get even hotter, because of the insulating effect of your
    finger. 45C is not hot enough to affect the reliability of
    them, but I cannot say I'm impressed with that design. If that
    circuit is in fact a regulator circuit, a switching design would
    cost a few dollars more, but would be cool to the touch.

    I tried probing the two MOSFETs on my board with a voltmeter,
    and I get strange readings off the pins. I'm not really sure
    what that is powering. I do know an adjacent (cool) MOSFET makes
    2.5V for the DIMMs, so the other two aren't doing that.

    When you are using the Audigy, you are correct, the sound
    signals don't go to the motherboard. The motherboard
    supplies power to the Audigy, the Audigy will likely have
    a linear regulator on the card, to clean the power, so it
    is unlikely you would have a motherboard induced problem,
    without the motherboard crashing first. The Audigy most
    likely fills its sound buffer with digital data, by using
    DMA (bus master) transfers from a block of system memory.
    It is a case of "garbage in - garbage out" - if the
    sound samples in system memory are miscalculated, then that
    could affect sound quality, but from a thermal perspective,
    the case air temp would have to be an inferno before the
    sound card is going to distort.

    If you want to try another test, run Prime95 (which will
    heat up the system the same way a game would) while you
    are playing a music CD (and using digital extraction on
    the CD, so digital data is pulled from the CD and fed to
    whatever sound system you are using). If the sound
    is not muffled or distorted after a long period of
    time, I'd be looking for a software source of this
    problem. I can say this now, because you've already eliminated
    the speaker's internal amplifier as a source of the problem.

    Muffled sound here is blamed on swapping center with sub on
    a 5.1 sound system:

    http://groups.google.com/groups?threadm=gAP5a.92798$

    This post about Half Life, mentions some console settings
    and A3D. You know more about these games than I do, so
    perhaps a little Googling with "muffled" as a keyword will
    uncover more in-game issues:

    http://groups.google.com/[email protected]

    Are you up to date on any patches for the games ?

    HTH,
    Paul
     
    Paul, Dec 7, 2004
    #4
  5. If you're referring to the MOSFETs providing the core voltage to the
    CPU, they will be a switching setup - the current draw is far too high
    (approaching 100 amps on some CPUs) for a linear regulator.
     
    Robert Hancock, Dec 8, 2004
    #5
  6. Navid

    Paul Guest

    No, these are over near the DIMMs. In fact, the MOSFETs for Vcore
    have always been cool on the Asus boards I've got here. The only
    exception was when I put a Tualatin 1.4GHz in my P2B-S, and then
    the Vcore MOSFETs started to get warm.

    Of the three MOSFETs near the DIMMS on the P4C800-E, one seems to
    be a linear for the DIMMs, making 2.5V from 3.3V. At least the
    MOSFET had 2.5V on one pin, and 3.3V on the other.

    The only function I can think of for the two hot ones, is a
    push-pull linear regulator for the DIMM terminator voltage.
    In the example below, is a circuit for SSTL2 termination, and
    maybe that is used on DDR ? The 1.25V "power supply" must on
    a per cycle basic, either source or sink current, depending
    on how many ones and zeros there are on the 128 data plus 16 ECC.
    This means the termination power supply is different than your
    average conversion problem, as both sourcing and sinking is
    required. Now, normally, you can get some tiny regulator chips
    to do this (and you'd need a few to handle that many data bits),
    but as Asus are penny pinchers, maybe these two MOSFETs are
    doing a big version of the same thing. I cannot see any
    magnetics near the two MOSFETS, and the main magnetics
    are one entry choke near the 2x2 +12V power connector, plus
    three output inductors for the three phase Vcore. Basically
    the MOSFETs would be part of a 1.25V linear power amp.

    1.25V
    +2.5V | +2.5V memory termination
    | Rterm | times about 144
    |\ | |\ instances. Vterm must
    --| \----Rseries---+----| \-- sink or source current,
    |/ |/ at least as drawn here.
    | |
    GND GND
    (e.g. Northbridge) (e.g. memory chip)

    (from http://download.micron.com/pdf/misc/sstl_2spec.pdf )

    Asus likes to use quad op amps and these MOSFETS. The 14 pinner
    near the MOSFETs is a quad op amp. If you look at page 21 of:

    http://www.asuscom.de/pub/ASUS/mb/sock478/P4C800E-DX/f1347_p4c800-e_deluxe.pdf

    the two hot MOSFETs are the ones below the right most pair of
    DIMMs.

    Paul
     
    Paul, Dec 8, 2004
    #6
    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.