Since the late 90s, I have come across quite a few PCs that do not power on.
I have checked out the power supply, the power switch, removed everything
except one stick of RAM, and they still lie dead.
I would expect there was a simple on-off connection to fire up the ATX power
supply, but there must be some logic circuit on the mainboard that checks
if a CPU is plugged in or whatever, and that is frizzed out.
Those were various brands of mainboard, the last one I had was Gigabyte.
You don't need a CPU, to turn on a PC power supply.
If you cable up a bare motherboard to a PSU, connect the
power button to the correct two pins on the front panel
header, you can turn it on. For example, on one of my
older PCs, the VID pin coding table for the CPU has an
"all 1's" table entry where the output of VCore is set
to zero. When the CPU is not present in the socket, the
VCore is turned off so the socket is not powered. The rest
of the motherboard remains powered and operational. The
clock generator delivers nominal clocks, from reset (so all
of the logic can initialize without a CPU present). The BIOS
can't run without a CPU, but the dumb power logic works
just fine without CPU intervention.
The motherboard converts an "ON pulse" from the front panel
button, into a steady ON signal on the PS_ON# wire of the
main cable. That's how the second half of the power supply
is turned on. The first half of the power supply is always
running, and it makes +5VSB. That +5VSB powers the motherboard logic
that sits there looking for a pulse from the front panel button.
The front panel button is a momentary SPST switch, not a rocker
or toggle or bat handle switch which would continuously indicate
the desired state. One pulse means ON, a second pulse means OFF.
That's called "soft power". Power control done with logic gates,
capable of turning on power conversion with a relatively weak or
low power signal.
You have to work your way through that logic, to figure out
why a PC won't start. Checking the level carefully with an
analog meter, on the PS_ON# wire, is part of that. (That's because
the ATX PSU does analog measurement of the PS_ON# wire and
doesn't use a purely digital method - a power supply can be
"half-ON" as a result.)
The motherboard logic is rather complicated, in that the logic
tree also includes lots of "Wake" signals. There is also back
feed protection features (analog power cutoff) in the design,
which complicate the time and voltage analysis of what happens
at start up. For example, there are some motherboards that
pretend "I'm still awake" in terms of their internal state, and
then they don't generate a RESET pulse to kick off the boot process.
And this is a failure in the sequencing logic on the motherboard.
On one computer, this was caused by current leakage coming from
the LCD monitor. If you pulled out the LCD plug, pushing the
button on the front of the PC then worked properly. Naturally,
the user discovering this, was pissed.
So the power control uses logic gates, and enough of the
motherboard works, that you should still be able to
turn an ATX supply on and off, using motherboard, PSU,
and front panel POWER button.
This diagram may not be completely accurate, but it's intended
to show how the momentary pulse from the front panel switch
(active low), makes the PS_ON# go active low in response.
The PSU turns on, when PS_ON# is around 0.7V or less.
And the PS_ON# signal type is open collector, so you
can connect a bypass jumper to ground to the signal,
and cause the supply to come on (not really all that
good an idea, but I have to mention it). Notice that
PS_ON# on the main cable, stays at logic 0 for the
entire computing session.
____ ________________________ _______
Front button |___| |___|
________ _______
PS_ON# |____________________________|
If you defeat PS_ON#, by grounding pin 14 on the main 20 pin
cable, it does turn on the PC. But that also defeats the
automatic turn-off functions, such as overheat protection.
If you use a grounding jumper on 14, the PC could burn up,
even through THERMTRIP is saying "turn off please". The
PC would be defenseless against faults with the jumper
in place.
Paul
