X5DA8 Failure

Discussion in 'Supermicro' started by G Farris, Jan 28, 2005.

  1. G Farris

    G Farris Guest

    I am running a SUPERMICRO X5DA8 board with two Pentium4 XEON processors, and
    the board has developed a problem. The computer will not run any more, because
    the board faults the power supply.

    With the computer OFF, the power supply produces a loud clicking sound, as it
    goes into protect, probably due to overload. If I remove the 5VSB connection
    from the main board, the clicking stops, the power supply starts and all fans
    and drives run - but the computer won't boot - apparently the board cannot run
    without power on the 5VSB connection.

    I can measure the resistive load from Pin 9 of the ATX power connector (on the
    board) to ground. It is about 9 ohms, which is well within the loading
    capacity of the 5VSB on the power supply. The power supply itself is an NMB,
    model GM460WTXW01SSV. The 5VSB output is rated at 2 amps, but in tests I can
    load it much higher - I have put a 1 ohm load on it, and measured 5 amps of
    current for a couple of minutes, without the power supply overloading or going
    into a protect mode. As a final test, I have used one of the other 5V outputs
    of the power supply (with the board powered up) to supply the 5VSB connection,
    and this shuts off the power supply instantly. I believe it is safe to say
    that the power supply is not the cause of the problem.

    I have the jumpers for USB and JPWake both set to 5V source (not 5VSB), but I
    have tried all configurations of these jumpers, with no change. I have
    examined every millimeter of the board for damage, visible shorts or stray
    wires, dirt etc, and I have vacuumed the board on both sides to make sure
    there is nothing conductive shorting this connection.

    The condition persists even if I remove everything from the board - both
    processors, all memory modules, all fans, drives etc. I have checked the
    battery on the board, and it measures just over 3V.

    What is going on????

    Thanks,
    G Faris
     
    G Farris, Jan 28, 2005
    #1
    1. Advertisements

  2. G Farris

    Paul Guest

    Disconnect the ATX power connector from the motherboard. Get a
    multimeter, and check the resistance from each rail to ground,
    on the ATX power connector. Leave everything in place for your
    first measurements (just disconnect the PSU from mobo). Maybe
    you'll find +12V, +5V, or +3.3V is partially shorted to ground,
    or even one of the two mostly unused rails. (Phone Supermicro,
    and see if they can tell you what "normal" readings are for your
    model of board. My experience in the lab, is electronic assemblies
    have fairly good clustering in terms of those resistance numbers,
    and you can get some health information about a board, simply
    based on the resistance reading from each power rail to ground.
    Perhaps in your case, the symptoms will be quite obvious, with
    a dead short indicated in one of your readings.)

    Remove components from the motherboard one at a time and measure
    the rail that was shorted again. Continue until the motherboard is
    empty. (Place all components in antistatic bags as you remove
    them.)

    If the board still reads a short on one rail, remove the motherboard
    from the case. Place it on top of a piece of cardboard. Measure
    the shorted rail again, and see if the problem was a short underneath
    the motherboard somewhere.

    If it is still dead, you could have a switching converter failure on
    the motherboard. Perhaps a MOSFET is fried or something. With your
    empty motherboard in hand, prepare the shipping container, and
    send it back for repair.

    If you get a new board to replace your current one, don't forget to
    measure the resistance of each rail to ground when the new
    board arrives. Write the values on a piece of paper, and stick the
    paper inside your computer case. If you have trouble in the
    future, you'll now know what the "normal" values are.

    To debug a motherboard like this, you need a schematic and a layout
    for the board, plus the CAD tools to load the design and allow
    cross probing. On a complex board, that is the only practical way
    to associate schematic to components in a time efficient manner.
    If you had a paper schematic, it would have to be extremely high
    quality, with abundant cross refs and component locations on it,
    and most modern schematics are useless in this respect. If you send
    the board back to the manufacturer, they can probe with a bed of
    nails, and find the short in seconds. If copper planes are shorted
    somewhere, they'll likely throw the board out.

    Something else you should consider, is the percentages are
    that your power supply has failed, and not the motherboard.
    Before wasting good money, only to have the manufacturer tell
    you "no fault found", try another power supply. It is a
    simple test that could save you a lot of time. (Sometimes
    power supplies become weak, and they cannot take a decent
    load. The power supply may operate at no load, but protect
    itself before the rated limits are reached.) Since you have
    a multimeter, don't try the power supply swap if you are
    already reading a dead short on the mobo :)

    Paul
     
    Paul, Feb 2, 2005
    #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.