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.

How does PC power-on work ?

Discussion in 'PC Hardware' started by no.top.post, Dec 15, 2010.

  1. no.top.post

    no.top.post Guest

    Apparently PCs often have problems if left off for weeks?
    Understandably the battery/cell becomes discharged.

    But I think 'dry joints' also develop in some of the many connections.
    Once my PC failed to run after it had been out of use for 6 weeks, and
    eventually I determined that the 'Start switch' had become a dry-joint.

    Now I've got the opposite problem on another PC.
    After the previous owner had not used it for 20 months it failed to power up.
    IIRC the CPU fan didn't start.
    After I left it on-power for 24 hrs, while I was getting some 'sevisol'
    [aerosol for dry-joints] it recovered.

    I've been using it for 4 months: 24 hrs a day for 4 days a week.
    But when I came back to it after 3 days off, the CPU-fam started immediately the
    PC got power, and before the 'Start switch' was activated.
    Apparently the CPU isn't running, but the IDE connector voltages are OK
    [5 and 11.8] and the CDrom/music runs/sounds independently of the CPU.

    What is a likely fault cause?

    In principle I'd like to know how the 'start' key/circuit/system works.


    == TIA.
     
    no.top.post, Dec 15, 2010
    #1
    1. Advertisements

  2. no.top.post

    JTF Guest

    My tries....lol
    Reseat memory, cpu and any add on cards...corroded connections are
    usually fixed by reseating....
    Check other connections and reseat (power etc) to remove any corrosion
    caused connection issues.

    Corrosion isn't so much rust as it is tarnishing of the card edge and
    plug connectors.....reseating will "scratch" off the corrosion
    restoring connection.

    My system has pretty much been on for over 4 years now and I haven't
    had issues.
     
    JTF, Dec 15, 2010
    #2
    1. Advertisements

  3. no.top.post

    mike Guest

    I had a laptop that wouldn't boot.
    I checked the CMOS battery and it was low.
    I charged the CMOS battery.
    Still wouldn't boot.
    The clock chip seemed to be counting.
    I reset the cmos chip...no help.
    I removed the cmos battery for a few minutes
    then put it back in. The computer booted just fine
    and has worked ever since.

    The battery was soldered in on the bottom of the mother
    board necessitating complete disassembly of the system
    and hooking the parts back up so it would run disassembled.

    My hypothesis is that the battery discharged far enough
    for the internal state of the cmos chip (or peripherally connected
    devices powered by the cmos battery) to be corrupted
    in a manner that wouldn't be fixed by resetting the chip.
    I don't have any idea why this could happen, but removing
    battery definitely fixed it.

    I've fixed similar problems in a Magellan GPS by shorting
    the backup battery momentarily, but I DO NOT RECOMMEND
    the process.
     
    mike, Dec 15, 2010
    #3
  4. no.top.post

    Bill Marcum Guest

    ["Followup-To:" header set to comp.os.linux.hardware.]
    In desktop computers, a BIOS setting determines whether the computer
    starts when power is applied, or you have to press the power button.
    The BIOS may also have been set to wake on LAN.
     
    Bill Marcum, Dec 16, 2010
    #4
  5. no.top.post

    david Guest

    My machine did this same thing yesterday. Reseating the power connector
    header on the motherboard fixed the problem.
     
    david, Dec 16, 2010
    #5
  6. no.top.post

    Paul Guest

    The power control is implemented partially in the SuperI/O and
    partially in the Southbridge. The reason for that path and those
    chips, is to include "wake events", so certain waking events
    can turn on the soft power (PS_ON#) control.

    As Bill says, one way of getting immediate power on, is a setting
    in the BIOS that is there for that purpose ("Restore On
    AC Power Loss"). But in your case, you may not have changed
    the BIOS setting. And if the BIOS settings were randoonly
    corrupted, there is a checksum that the BIOS can use, to
    detect a percentage of those corrupt bit patterns.

    The other way of causing immediate power up, is stressing
    the Southbridge. The Southbridge is a multi-rail chip. For
    example, the RTC and CMOS RAM, sit in the "CMOS Well", a
    chunk of circuitry powered by the CMOS battery (or by
    3VSB when the ATX supply makes +5VSB available). The well
    is isolated with transmission gates, to prevent backfeed
    current flows. That is there, to prevent the CMOS battery,
    from being used to power the entire computer (wearing
    out the battery in short order).

    The CMOS Well may also have some of the logic gates used for
    turning on the power.

    Now, I've had the following happen to me, on a 440BX. One
    day, I was inside the computer, and half unseated the IDE
    cable. Half the pins were still making contact, and the
    other half were not. Upon switching on the power at the
    back, the computer started immediately. After firmly seating
    the cable again, with the power off, it operated normally
    again.

    It could be, that something that plugs into the Southbridge,
    is shorting or putting an abnormal load on some logic
    signals of the Southbridge.

    The fact the processor can't start, could be related. If
    the Southbridge is "under attack", then it isn't likely
    that a probe to the BIOS interface is going to work.
    Then, the processor may be ready to run, but "bus faults"
    as soon as it tries to read the BIOS chip. And then it
    can't start.

    Double check your cabling. Reseat cables. Do it with the
    power off. Then, try again.

    Paul
     
    Paul, Dec 16, 2010
    #6
  7. no.top.post

    no.top.post Guest

    Thanks for the in-depth info, which I've filed.
    I didn't want to expose all the complications initially:-
    I'm operating in an african/3rd-world environment.
    I move the PSU & 2 IDEs between 2 PC [locations] every week;
    I forgot to remove the mains, before I unplugged the PSU from PC2,
    and when PC1 'got' the PSU it behaved strangely - as described.
    PC1's CMOS battery is 'down', but apart from having to set the
    RTC and boot-sequence every week, I don't know the meaning of
    most of the BIOS settings.

    When I came back to PC2's location [where I have inet access]
    I was able to confirm that the IDEs [which are most important]
    were still OK, but PC2 now has strange, apparently unrelated faults:
    hda,b are undetected; so I had to 'set' the BIOS and /etc/fstab
    to continue via hdc.
    I'm taking a scrap but tested MOBO and PSU to location PC1,
    so that I can process my inet-fetched material on the 'movable'
    IDEs, and some servisol [engineer in a spray-can] to check PC1.

    BTW, I had already tried removing the CMOS battery/cell.
    I guess the BIOS just goes to workable defaults when there no
    CMOS-power, since this worked OK for months.

    == TIA.
     
    no.top.post, Dec 17, 2010
    #7
  8. no.top.post

    Paul Guest

    It could be, that something got damaged, when it was unplugged with
    the power still applied.

    As for the CMOS battery, yes, in some cases, the defaults reloaded
    by the BIOS, while running off +5VSB instead, are useful for booting.
    But I have several computers here, that when the battery is flat, and
    I power them for the first time, it means a few minutes work turning
    stuff back on in the BIOS screen. Generally, the "dumber" the PC
    (like a Dell), the better the odds the BIOS default values will work.

    Paul
     
    Paul, Dec 17, 2010
    #8
  9. no.top.post

    Arno Guest

     
    Arno, Dec 21, 2010
    #9
  10. no.top.post

    mm Guest

    YOu left on for 24 hours while the CPU fan wasn't spinning, while you
    got a lubricant? I think you should have turned it off until you got
    the lubricant. Isn't that obvious? You're lucky you didn't burn it
    out.

     
    mm, Dec 23, 2010
    #10
    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.