Post by (PeteCresswell)Family members HP Compaq/Older Athlon X2 just went belly-up.
Only sign of life is a single green led glowing on the mobo: no power
supply fan, no CPU fan, no response to DVD drive's eject button...
First thought was power supply, but when I swapped in a known good power
supply, no change.
Then I figured "Maybe the on/off switch"... so I pulled out the switch
and used a knife to jumper the two leads coming into it - figuring that
would duplicate the action of the switch... but no luck there either.
Unless somebody has a better idea, next thing I am going to do is to
revisit the power supply and verify correct voltage on each output lead.
The PS_ON# signal has lost its ability to pull to a logic low.
Just a guess.
Process starts with +5VSB. There must be +5VSB for the
system to even think about starting. The power supply
has two parts. A section with just +5VSB on it. A section
with the major power rails (3.3V/5V/12V etc).
We know the +5VSB is working, because you have a green motherboard
LED. The LED also tells me the motherboard is an Asus OEM. As
Asus is one of the few manufacturers who provides a monitoring
LED (green in color) for +5VSB. The +5VSB feeds the LED directly.
If the LED glitches, doesn't glow at full brilliance, these
are cheap indicators of trouble with the "supervisor" half
of the ATX power supply.
So we suspect your machine is getting +5VSB. You also did the
right thing, by checking for power switch. But there is an
additional step. If the Power button is "jammed on", the machine
can't come out of reset. What you really want, is to disconnect
the twisted pair for the power button, and do a momentary touch
with your "shorting wand" on the two terminals. If you own
a separate push button (I have one here for this test), you
can slide your "known good" switch and cable in place of the
one provided by the computer case.
Once you do the momentary contact thing, PS_ON# is an open
collector signal on the main cable. It pulls down a
pullup on the PS_ON# signal on the PSU. If you had a
multimeter, you'd check the voltage on the line.
A typical failure, is (for some reason) the transistor
on the motherboard, can no longer pull the signal down.
Maybe the signal drops from +5V (PS_ON# not asserted) to
around +2V. This is not sufficiently low to be
recognized as a logic 0. You might want a voltage in the
0.4V to 0.7V region as a good solid trigger.
The user can short PS_ON# to ground, and force the PSU on.
And that test then is a workaround for a defective motherboard
PS_ON# driver. The computer will likely start.
OK, so what's the downside of operating this way:
1) You've done an override on THERMTRIP. The computer
cannot now shut itself off in the event the heatsink
on the CPU falls off.
2) You've disabled the ability of the computer to do
"Soft Off". When you select "Shutdown" in the OS,
the Soft Off shuts down the main power rails. With the
shorting jumper in place on PS_ON#, now you need to
remove the shorting jumper so the computer can go off.
But for the purposes of a test, jumpering PS_ON# to GND
will help work around a defective PS_ON# response.
In fault analysis, you can either isolate to the nearest
subassembly, or the nearest three components. The three
components in this case might be motherboard, PSU, cabling.
You know that something is wrong with PS_ON#. Motherboard
end might be "too weak". PSU end might be "too strong"
(that happens sometimes). Since you've swapped in power
supply, that helps eliminate the "too strong" theory.
Leaving motherboard or some cabling flaw. So you're getting
pretty close to blaming the motherboard. Not a lot
wrong with it, but still a nuisance. And this failure
is *entirely* too common. This circuit should not
fail, yet there are way more failures than should be
happening. And I don't have any convenient whizzo
theories as to why.
When I was a new grad, an engineer at work called me
over to review something he was doing. He had premature
failures in the very kind of circuit you're looking at.
It was wired-OR logic and open collector driving chips.
I took one look at it, and I could see immediately,
he was forcing the driving chip to sink 3x the rated
current. The chip would last pretty well three days
doing this, before it would "take a crap". He would
put in another spare, and it would run for another three
days or so. That's an example of "abuse" and since this
was a prototype, nobody got hurt. But it was one of
my first experiences with what it takes to kill
a circuit like that. And I appreciated the opportunity
to see theory turned into practice (just what could
an abused chip take). I was surprised it was that
sensitive. This should not be happening on a modern
motherboard, and the chip is nowhere near the
abuse level.
At one time, boards around the 2000 era, were using
stuff like 74F07 or the like. Huge drive capability.
Now, the driver is gutless, but should still be
well suited to the task (in other words, no
spec violations should be present on either
end of the implementation). Which is what makes
it hard to understand why the drive is defective
on that signal.
Sticking your multimeter on PS_ON#, should tell
you whether it's making it all the way down to
0.4V to 0.7V or so. The voltage is the saturation
voltage of a bipolar junction transistor. That's what
that voltage represents. And chances are, the transistor
involves is "kinda" working, and you'll see 2.0V on it
or so. Which isn't sufficiently low to kick on the PSU.
If the power connector is 24 pin, you can get the
pinout info here. Or here.
http://www.formfactors.org/developer/specs/ATX12V_PSDG_2_2_public_br2.pdf
http://www.playtool.com/pages/psuconnectors/connectors.html
HTH,
Paul