Steve said:
P-IV running Windows XP Home. Windows appears to be shutting down as
the screen goes dark, but the "turn off" has to be done manually. Any
help will be greatly appreciated & thanks ...
Was it working at one time ?
There can be hardware or software reasons, and if it was all working
at one time, that might help eliminate the software reasons.
*******
The hardware looks like this. The front switch is momentary contact,
and the motherboard logic converts the pulse from that switch, into
a solid continuous level on PS_ON#. If PS_ON# is zero volts, that
means the supply should come on. IF PS_ON# is at a 5V level, then
the supply should go off. Because the main cable of the supply has
a pullup resistor on it, that holds off the supply when no motherboard
is connected to drive that signal. That's why the PSU won't run by
itself when its sitting on the bench (jamming PS_ON# to COM starts it).
Front_Switch ------ Southbridge/SuperIO -----------------> ATX PSU
^ PS_ON# |
| +5VSB |
+----------------------------------+
In terms of hardware, it could be a (motherboard) driving end problem.
Or, it could be a problem with the thresholding of the signal by
the ATX PSU.
It's theoretically possible for a person to characterize the behavior
of both ends. You can check that the pulldown function of the motherboard
works. You can check that the pullup function on the PSU works. If
you don't know how to do that, you'd replace the motherboard and PSU
and test by substitution (which is a pretty expensive way to do things).
Since PSU failures are more common, you'd start with the PSU end.
Sticking a multimeter on the PS_ON# line, and looking at the voltage
levels, would give you a hint. If the line is swinging from 0.4V to
close to 5V when the computer is "supposed" to be off, then that
would tell you the PSU is likely having a problem reading the signal
level (bad PSU). If the line swung from 0.4V to 1.4V, then you'd have more
trouble guessing at who is at fault. That would require more
characterization of either end of the line. The motherboard end
can only pull down on the line, since the signal is open collector.
But when you see only a 1.4V level, is the pullup weak, or is the
motherboard driver still tugging on it ? That is indeterminate. You would
then need to separate the motherboard and supply, and test them separately.
These are ATX supply specs. I'll use the middle one, to discuss how
to use one of these. The third one is for 24 pin connectors.
http://web.archive.org/web/20030424...org/developer/specs/atx/ATX_ATX12V_PS_1_1.pdf
http://www.formfactors.org/developer/specs/atx/ATX12V_1_3dg.pdf
http://www.formfactors.org/developer/specs/ATX12V_PSDG_2_2_public_br2.pdf
On page 30 of ATX12V_1_3dg.pdf , you can find the main cable pinout. The
PS_ON# signal is green in color and on pin 14. The body of the main power
connector is open at the back, and you can jam the probe of your multimeter
in there, when the system is running. (That is easier to do, when the
system is removed from the computer case and sitting on your table.)
I connect the ground on my multimeter, to a screw on the back of the computer
in the I/O area. That way, I only need one hand to guide the red probe to
the main connector. That avoids shorting the meter tips together.
Setting my meter to the 20VDC range, allows me to measure 5V logic signals.
On page 20 is shown the behavior of the input. Normally, for a TTL
level circuit, the driving end might manage a swing of 0.4V to 2.4V minimum,
while the input threshold is set at 0.8V and 2.0V. Since 2.4V is higher than
2.0V, the input says "that's a logic 1". When the 0.4V level is seen, that is
less than 0.8V, so the input says "that's a logic 0". Values between
0.8V and 2.0V (like if you saw a 1.4V level), will confuse the logic
and the input will either guess it is logic zero or one. In some cases,
this failure to swing properly, causes the less than normal behavior.
I've even heard of cases, where a 1.4V level presented on PS_ON#, causes
the power supply to be "half on". The supply works, but won't meet full
load specs, and "falls over" easily. So the thresholding on PS_ON# on the
PSU end, isn't guaranteed to be "purely digital". In some instances,
the circuit on the end of the signal behaves in an analog way, so a half
way signal makes for a half-weak power supply. You're more likely to see
that on an older power supply.
Paul
, and give