1. This forum section is a read-only archive which contains old newsgroup posts. If you wish to post a query, please do so in one of our main forum sections (here). This way you will get a faster, better response from the members on Motherboard Point.

Really stumped: PS inoperative

Discussion in 'PC Hardware' started by Haines Brown, Apr 13, 2008.

  1. Haines Brown

    Haines Brown Guest

    I raised this issue before in this group and received some very helpful
    advice regarding the proper procedure for testing out a power
    supply. But my problem is more basic than that.

    I had an old Enermax power supply in a new box that seemed to work to
    the extent I could boot to the MB spash screen, but then it stopped
    working altogether (no fan spin, etc.) If I played with the computer
    front panel on/off switch (depressing a few seconds, releasing a few
    seconds and repeating the cycle), I could get the CPU can to make a few
    turns, but that's all. I checked the PS fuse, and it's OK. No visible
    signs of trouble inside.

    I decided to purchase a new power supply, but when I installed it, same
    problem. When I pushed the on switch the first time, the CPU fan rotated
    a couple of times, and then nothing. MB LED on, but no fans or anything

    I took the old PS by itself, not even installed in a machine, and
    connect one of its molex connectors to a fan. I connect power, turn the
    supply on, and the fan does not spin. The 115/220 volt switch is in the
    right position (115 visible); the power cord does deliver the
    voltage. The fan is new and I assume works.

    I might be tempted to circuit trace the power supply if I didn't have
    the same problem as the new one. Obviously I'm doing something that is
    very elementary wrong.

    If a power supply is getting juice, the switch is on, and the fuse not
    blown, shouldn't the periphery power connectors cause a fan to turn?
    Does not at least that part of the power circuit become active as soon
    as the supply is turned on?
    Haines Brown, Apr 13, 2008
    1. Advertisements

  2. Haines Brown

    Bob Eager Guest

    (assuming this is an ATX PSU with the 20 pin connector)
    The PSU needs to have a particular pin (PS_ON) grounded to run properly
    (and run the fan). It generates a standby voltage on a different pin,
    which is enough to run the power control part of the MB, which then
    grounds PS_ON when the MB considers itself to be 'on'. The MB may not be
    holding that pin at the correct level.
    You need to strap the appropriate ground pin to the PS_ON signal.
    A small part does, but it doesn't run the fan.

    See here:


    and look at chapter 4.
    Bob Eager, Apr 13, 2008
    1. Advertisements

  3. Haines Brown

    Haines Brown Guest

    Just what I needed to know to start tracking down the problem. When I
    sort pin 20 on the main power connector to the black ground next to it,
    the fan runs. So it's beginning to look like I have a problem with the
    power-on circuit outside the power supply.

    If you are willing, you might help me with a further question. In the
    power supply there's apparently a resister between the standby 5+ volts
    and ground that holds the Power-On to some level, which does not allow
    the power supply to power its rails. When I grounded pin 20 on the main
    power connection, that dragged the voltage down to 0, which enabled the
    power supply to deliver power to the rails.

    What I don't know is how the power on switch on the front of the
    computer chassis ties in to this. It is a push-button which that closes
    a circuit when depressed and is normally open circuit. Apparently this
    temporary closure causes the Power-On connection to the power supply to
    toggle its state between being grounded and being open. How is that

    I've assumed this control circuit that includes the front panel power
    switch is not polarized.
    Yes, thanks. Useful.
    Haines Brown, Apr 13, 2008
  4. Haines Brown

    Arno Wagner Guest

    That is as expected. An ATX psu gets its poer_on signal from
    the mainboard and the power-switch on the front is in turn
    connected to the mainboard. In order to turn a PSU on in
    "stand alone" mode, you have to short the /power_on
    pin to ground. Pinouts for 20pin and 24pin ATX can be
    found here:


    /power_on is supposed to be a green wire, ground is black.

    You do.

    The standby 5V must become active. It is supposed to be
    on a purple wire, see also the "5VSB" pin in the diagrams
    referenced above.

    Arno Wagner, Apr 13, 2008
  5. Haines Brown

    Arno Wagner Guest

    Some complex ciecuitry on the maonboard. Complex, because most
    mainboards also support power on by other events, e.g. network
    packets, modem ring, keyboard, timer, etc..
    The front switch is not polarized in any way.

    What you can do is depress the front switch, and observe
    what happens to the voltage across it. You can also monitor
    the PSUs power_on pin while pressing the front switch.

    At the moment it looks like your mainboard is shot, but keep
    looking a little longer.

    Arno Wagner, Apr 13, 2008
  6. Haines Brown

    Bob Eager Guest

    My understanding is that circuitry on the motherboard latches the
    grounded state. This can be unlatched by the motherboard itself, which
    is how a power down and switch-off by software is done. There is a
    standby (always-on) power line in that 20 pin plug, and that powers
    enough circuitry to work the latch. Failure to latch indicates some kind
    of motherboard problem.
    No, it's not.

    BTW, I bought a useful gadget for testing PSUs. Plug the PSU into it,
    and it grounds PWR_ON as displays the state of all of the power lines.
    It also provides test points for more accurate measurements. But then I
    do quite a lot of system building/repairing. It cost me (in the UK)
    about $20 (US).
    Bob Eager, Apr 13, 2008
  7. I had an similar failure on my machine. In the second repair attempt it
    turned out that the CPU socket was the reason, why the cooling
    construction had no *permanent* contact with the CPU. That caused the
    CPU to turn itself off, about 2 seconds after power on. Perhaps you have
    a similar temparature/cooling problem?

    Hans-Peter Diettrich, Apr 13, 2008
  8. Haines Brown

    Haines Brown Guest

    Zero voltage across the front power switch when not depressed, and zero
    voltage at the Power OK pin on the main power connector, whether or not
    the front switch is depressed.
    I hope not. I just bought the board to replace another one. At least if
    it's defective, it's under warranty.
    Haines Brown, Apr 14, 2008
  9. Haines Brown

    Haines Brown Guest

    Yes, I suspected as much. Despite what I might have just reported in my
    other message, with the PS power switch the green PS_ON wire reads
    3.7V. When the front panel switch is depressed, the voltage drops to
    zero. If I push the front button again, but hold it for a few seconds,
    the voltage climbs back up to 2.5V. So it seems the latching mechanism
    on the MB is not entirely dead.

    The grey PWR_OK wire on the main power connector reads zero after the PS
    switch is off for a while and when it is switched back on. If I now
    press the front power button, it causes a little activity at this point,
    but it quickly settles back to the 0.004V. This seems wrong. I gather
    that the PWR_OK connection should be 5 V initially, and when the power
    on button is depressed, drop down to 3.3V. I'm not getting anything like

    If my impression of the proper voltages here are correct, then the
    question is, what determines them? Is the initial voltage supposed to be
    5V, and is that drawn from standby power in the power supply? That is,
    is the initial voltage independent of the motherboard?
    Haines Brown, Apr 14, 2008
  10. Haines Brown

    Haines Brown Guest


    Sorry, but a slight correction to my prior message. The voltage across
    the front panel button when the power supply is turned on is 0.06V. When
    the switch is pushed, this voltage slowly drops to zero. When the front
    switch is released, the voltage goes back up to 0.06V.

    Sorry for voltages provided in my first message, for I also happen to
    have a faulty front panel switch, and I have to simulate the conditions
    for the measurements.

    See my note to Bob that suggests maybe the latching mechanism is
    working, but that the PWR_ON voltage does not have a sufficiently high
    voltage and does not respond to MB power latching.
    Haines Brown, Apr 14, 2008
  11. Haines Brown

    Arno Wagner Guest

    If your measurements are correct, then it is the board. I would
    call your mainboard's power-on circuit stone dead.

    Arno Wagner, Apr 15, 2008
  12. Haines Brown

    Arno Wagner Guest

    It should be around 3-5V. 0.06V is in range for a measurement error
    or a short circuit on the board.
    The power-on voltage is supplied by the PSU itself. The board just
    shorts it out to ground to turn the power on. It cannot have
    an insufficient level, unless the PSU is broken. You can test
    that by turniong on the PSU without the board being connected to it
    and then measuring the volgate on /power_on.

    I still think the board is dead.

    Arno Wagner, Apr 15, 2008
  13. Haines Brown

    Arno Wagner Guest

    Then it woirks ok.
    Again, entriely ok. Note, however, thet this is /ps_on, and not
    ps_on, i.e. it is active low. That means a low voltage tells the
    PSU to switch on.
    It is. It means the PSU could not get the voltages up.
    No. It shoul be false with the computer off and true with it being on.
    Ok, first thiese are logic-level signals. <0.8V is false, >2.4V is true.

    Power_OK needs to be false if the PSU is turned off and needs to go
    to true after switch on (via front-button) within a very short time.

    I see an additional possibility here: One or more of the power
    lines might be shortened out. This would not even need to be
    a problem with the mainboard. Quich check: Disconnect _everything_
    except the mainboard and see whether you can get it to start then.
    Also remove all add-on cards, except the video card.

    A popular way to short out some PSU power lines is also to
    mount the mainboard wrongly. To check that, remove the
    Mainboard from the case and test in on a pice of cardboard or
    other insulating material.

    Arno Wagner, Apr 15, 2008
  14. Haines Brown

    w_tom Guest

    Important numbers are each wire voltage before and when the power
    switch is pressed - as summarized in a previous post. For other's
    benefit, what is happening here: isolated parts of that 'less than two
    minute procedure' are eventually being performed.

    What is called a 'latching circuit' is a power supply controller.
    Power switch is only one controller input. Measuring that switch was
    unnecessary if numbers from that two minute procedure were provided.
    Again, numbers both just before and when power switch is pressed.

    Irrelevant are voltages on green and gray wires without first
    providing purple wire voltage (both before and when switch is
    pressed). Purple wire provides power to the power supply controller
    and can make that circuit work intermittently. Purple wire voltage
    must measure above 4.87 VDC - and should be posted here because those
    numbers may provide other information.

    Before switch is pressed, green wire voltage must be well above 2.0
    volts. When power switch is pressed, then green wire voltage drops to
    below 0.8 volts. Last posts suggest that one part of the power
    supply controller is working AND that the power switch is completely
    OK. Two minute procedure made power switch testing unnecessary.

    Moving on. Not provided were voltages on red, orange, and yellow
    wires - each measured as the switch is pressed. Without what happens
    on each wire and resulting numbers, useful information has been
    withheld making the next paragraph an incomplete reply.

    Gray wire must be something near zero before switch is pressed
    (again, what was the actual number). When switch was pressed, what
    happened to that gray wire voltage in the next 5 seconds? As I read
    it, that voltage either starts to rise or never increases. Voltage
    must be well above 2.4 volts within 2 minutes. That explains why
    replacing a power supply would accomplish nothing. But without
    specifically stating what gray wire voltage did and without voltages,
    then too many possibilities still exist. No useful reply can be
    posted without all those voltages in the two minute procedure.

    I might bet one of the orange, red, or yellow wires will indicate
    where to look next - what to suspect. But at least five other reasons
    still exist which is why nothing more can be recommended until those
    numbers are provided.

    This paragraph is for others to learn from this discussion. A new
    power supply solved nothing because the problem was not identified
    first. A two minute procedure with a 3.5 digit multimeter that can
    eliminate lost time and labor is "When your computer dies without
    warning....." starting 6 Feb 2007 in the newsgroup alt.windows-xp

    Among many things answered immediately in less than two minutes: is
    the power switch good? Is the power supply defective? Are wires
    properly connected? Nothing faster could have answered so much so
    w_tom, Apr 15, 2008
  15. Haines Brown

    Haines Brown Guest

    You know, that worried me, too.

    1. HDD LED. Here I'm not certain. The manual shows one pin as a square
    and the other as a circle, and I was concerned that this implied
    polarity, although it shouldn't. As of now I've an orange wire on the
    square and a white wire on the circle. Could I be wrong?

    2. Reset. OK

    3. Power LED. Since the front panel connector accomodates a three pin
    socket, and its wires are for pins 1 and 3, it could not be inserted the
    "power LED" section of the front panel header, which presumes a two-pin
    connector. So instead I put the Power LED connector to another header
    that the manual calls "Alternative power/sleep LED", which calls for a
    three pin connector. Was I wrong to do this?

    4. Front panel audio connector. The wires are only long enough to reach
    the side of the board nearest the front panel, while the mb header for
    the "front panel audio connector" is on the far side of the MB. Since it
    couldn't reach, I just left it off. Some day I may make or find an
    extension so that it will reach.
    Haines Brown, Apr 15, 2008
  16. Haines Brown

    Haines Brown Guest

    By bad. I simply skipped reporting on that. The front panel power button
    goes to a "Power SW" connector, which I have connected to what the
    manual calls the "On" pins of the header for front panel connections.
    Haines Brown, Apr 16, 2008
  17. Haines Brown

    Haines Brown Guest

    Sorry that other things were interfering with getting all of the
    voltages before, but I've jury-ricked the equivalent of a front power
    switch and here are the results.

    1. Purple standy-voltage. When PS is toggled on, this rises to 5.20 V.

    2. Green PS_ON. Starts at 3.7V. When front panel power switch is
    pressed, drops to 0.001 V.

    3. Grey PWR_OK. Starts 0.004 V, when front panel power switch is pressed,
    there's some flickering (no voltage to read), but quickly settles back
    to 0.004V. This should have remained above 2.4V.

    4. Although given the above, there may no point in these, I add them for

    Orange: 0.00V jumps to 0.80V, but drops right back to 0.00V
    Red: Starts 0.00, momentarily jumps to 0.36 but drops back to 0.00V.
    Yellow: Starts 0.00, flickers a bit, but remains at 0.00V.
    Haines Brown, Apr 16, 2008
  18. Haines Brown

    Arno Wagner Guest

    Ok, the mainboard sends the right signals. Looks more and more
    like a short somewhere. The orange/red/yellow measurements
    migth indicate that +12V (yellow) is shortened out. Incidentially
    the CPU voltage is generated from it and a MOS FET (as used in the
    switching regulator) can short out a PSU without problem.

    Arno Wagner, Apr 16, 2008
  19. Haines Brown

    w_tom Guest

    Demonstrated is a point made in most every post. What was
    speculated as "no point but added for completeness" was the only
    useful information.

    . Defect on 12 volts either in the load or in the power supply.
    Finally we have taken the first useful step from the original post;
    the original problem. Those with best knowledge, for the first time,
    are able to post something useful because numbers from the two minute
    procedure have been provided. Again, the point for all others who
    would learn from this discussion. What was thought not important was
    THE most important fact posted. Never filter information. Never
    short information from those with technical knowledge. Those
    'irrelevant facts' are the only infomation that made a useful reply

    The problem is now isolated to a short list of components. What
    uses the 12 volts? Again, keep following the evidence. Problem
    exists on something that draws 12 volts.

    Clearly not relevant now and not relevant earlier are: 1. HDD LED,
    2. Reset, 3. Power LED, 4. Front panel audio connector, or other
    missing connections. Those other suggestions demonstrated classic
    wild speculation; not how one 'follows the evidence'. Evidence made
    woefully obvious by following the evidence: the yellow wire voltage
    is shorted out.
    w_tom, Apr 17, 2008
  20. Haines Brown

    w_tom Guest

    What is suggested clearly is not correct - completely ignores the
    useful facts provided by the OP - numbers. Voltage on the green wire
    both before and when power switch is pressed means that wiring was
    always perfectly fine. Also obvious from numbers (once they were
    finally provided) was that a missing connection obviously is not the
    problem - which is why solutions come from 'following the evidence'.

    One reason for so many posts, so much speculation, and no useful
    replies were no numbers. All this could have been answered in the
    very next post had numbers from that two minute procedure been
    provided as requested. Once the OP provided those numbers, quickly
    the problem was isolated down to a few possible defects. The OP's
    next step is to determine which of few items actually creates the
    defect - by again 'following the evidence' and by not speculating that
    "it could be this or could be that". Less than two minutes and the
    resulting numbers would have made how many posts here completely
    unnecessary? Follow the evidence is a concept that many have
    difficulty grasping. And 'follow the evidence' is the only reason why
    the OP finally has his first useful replies.
    w_tom, Apr 17, 2008
    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.