Soft-off state detection

[Vyruz]

Does anyone know how the method a motherboard uses to detect that the user is requesting it to go into soft-off state?

the following problems i've been experiencing on some pc's:
at random, once every few weeks, some PC's will abruptly shutdown, exactly like they would be enterring the soft-off state.

In our pc's we have custom build power buttons, which are soft of hidden in the rear panel, to prevent users from switching off pc's just like that.

so i thought that there might be a problem in these buttons/cablings.

You probably are aware that the power button short-circuits the 5V lead with the ground, which tells the MB that the button is pressed.
So i thought there might be some sort of short-circuit going on in the cable.

After an intensive test session on a scope, we've seen that sometimes, there will be short circuits for a few milliseconds.

You're probably also aware that the power button should be pressed for +4secs in order to request a soft-off state to the MB, so those few milliseconds of short-circuit, could never cause this to happen.

EXCEPT in the following case:
the MB will use sampling to request the line status, this would mean that MB would check the line every couple of milliseconds.
Then, in very rare situations, imagine that the spikes in the cabling, match with the sampling checks on the MB.

this seems offcourse far-fetched, but remember that this only happens once in a couple of weeks/months on only a few PC's out of a couple of thousands.


So my final question is actually simple:
how does the MB check the power button line, continuous, or via a sampling method?

any help would be greatly appreciated

 

[bobsmith1492]

Check for overheating; the computer I built for my parents started shutting down randomly because I set the shutdown temperature at like 60 deg. C and it was hitting that and shutting down. (1 possibility anyway)

By the way, the power button doesn't short the +5V to ground, as that would simply short the +5V line, causing a nice spark and either sending the MB wires into fuse-test mode or shutting down the power supply - the hard way, where it goes into self-protection.

The most obvious way to check the soft-off power function would be to have a depressed power button charge a capacitor and use an RC time constant to create the delay. Then again, I usually think analog before digital, so they might not do it that way. I don't know how it would be done digitally, but it does seem pretty far-fetched (the idea you're proposing).

One thing you could try would be to disconnect the power buttons from a group of computers and then watch them to see if they shut down ever. If one of them ever shuts down like you described, it would prove the button/cabling itself wasn't the problem.

I would suspect a software issue first and foremost. Hardware is usually pretty clear-cut: it works, it doesn't work, or sometimes it's flaky... ok, so maybe not so clear-cut. Anyway, yeah, there are a lot of steps to initiating a shutdown, and while the power button is one way to start the process, it is by no means the only way, or even most common way; usually shutdowns are initiated in software. I'm afraid I can't help you there.

Good luck!

By the way, you may be better off posting this in the "Technical Support" forum; I know you posed a technical question, but I don't think that's the actual problem. Plus, more people would see it there.

Edited for spelling and grammar.

 

[krotchy]

Originally posted by: Vyruz
You're probably also aware that the power button should be pressed for +4secs in order to request a soft-off state to the MB, so those few milliseconds of short-circuit, could never cause this to happen.

Im a little confused, the way all of my computers work, is that if I hold the power button for 4+ seconds I get a hard shutdown. If I push my power button for less than a second, I get a soft shutdown.

Unless im missunderstanding what you mean by soft shutdown. I consider a hard shutdown, everything just turning off, as if I cut the power. I consider a soft shutdown about a 15-20 second affair where windows closes out every application and then shuts off.

 

[Vyruz]

Originally posted by: krotchy

Originally posted by: Vyruz
You're probably also aware that the power button should be pressed for +4secs in order to request a soft-off state to the MB, so those few milliseconds of short-circuit, could never cause this to happen.

Im a little confused, the way all of my computers work, is that if I hold the power button for 4+ seconds I get a hard shutdown. If I push my power button for less than a second, I get a soft shutdown.

Unless im missunderstanding what you mean by soft shutdown. I consider a hard shutdown, everything just turning off, as if I cut the power. I consider a soft shutdown about a 15-20 second affair where windows closes out every application and then shuts off.

I was confused by the same thing earlier.
but in most manuals this operation is reffered to as the soft-off mode.
e.g. (from a PIIXE motherboard):
When the PWRBTN# signal has been continuously asserted for greater than 4 seconds, the PIIX4
automatically transitions the system into the soft off state and clears the PWRBTN_STS bit.

 

[Peter]

Power buttons do NOT short 5V and Ground, that would be completely insane.

What they do is assert a dedicated signal input pin on the chipset southbridge. This generates an ACPI message to the operating system, which will then perform whatever action is linked to that button press. (In Windows XP, see "Power Management" properties.) This action typically is "Hibernate" or "Shutdown", both of which will eventually end up sending the south bridge into S5 (Off) state. The south bridge being the system's power control unit, this pulls everything else into Off state as well.

If you hold the button in for a few seconds, the south bridge will go into S5 without asking the OS first. This is useful to force the system to Off when the OS has crashed, or in OSes that don't do ACPI.