PCbanter - View Single Post

**Paul[_32_]** · July 21st 18, 07:21 AM posted to microsoft.public.windowsxp.general

wrote:
On Sat, 21 Jul 2018 00:43:39 -0400, Paul
wrote:

Andy wrote:
I have my computer plugged in to a power strip.

Is there a way to turn on my computer just by powering on the power strip instead of having to hit power button on computer?

Thanks,
Andy

"It's always too early to quit."
There can be a BIOS power policy setting on desktop
motherboards, which allows booting as soon as the power
is restored. It goes something like this:

"When the power is restored" [Keep computer off] == typical default
[Start computer] == what you want
[Last state] == put computer OFF or ON
depending on whether it
was previously OFF or ON
at the instant the power
dropped (mostly useless
setting).

Some motherboards and their chipset, have actual support.
There won't be any "computer twitching" or "fake starts"
to figure out what power state to use. The computer
instantly responds according to that BIOS setting.

On motherboards where the chipset doesn't really support
the feature, the feature is emulated. On power recovery,
the computer *always* comes on for at least one second.
In that one second, the computer checks the BIOS setting.
It can then decide what to do after that.

On some computers, you get "power on, power off, power on"
as the sequence at startup, as stuff like the above
happens. Some people assume their computer is
busted when it behaves like that, but it's actually
caused by the need to emulate that feature.

Paul

Back in the early days of ATX supply before the BIOS got smart enough
to start on power restore but it still needed a start signal my hack
was a 5v or 12v relay with the N/C contact connected to the power on
switch (to ground)
When power comes up it sees "power on" and as soon as the supply comes
up the power on shot goes away.
Simply grounding pin 14 might not work.

Well, grounding pin 14 will work, because it's wired-OR
logic. The motherboard uses open collector drive.
You can't damage the motherboard by shorting pin 14 to
ground. (Not unless a motherboard designer violates
the rules.)

The trick is, to decide when to stop driving pin 14 like
that. I don't know if an OS like Windows 10, puts up
a "it is safe to turn off your computer" like in Win98
days, to indicate it's OK to drop pin 14. You don't want
to drop pin 14, until all caches are flushed.

And you don't need a relay for pin 14 - if you want to
do your own circuit, you can do open collector drive
on your circuit too.

In the old days, for reasons I don't understand, they
used to use powerful open collector drivers. Like a
74F series (74F38?) with 64mA drive. Yet the pullup
resistor on pin 14 isn't supposed to source more than
a couple milliamps. And the other observation is, we
seem to have motherboards where the OC driver fails,
more than it should. So either the notion it's a couple of
milliamps isn't always right, or the designers of
systems 20 years ago were nuts :-) PS_ON# has always
been a bit weird that way. Certainly with your idea of
using a relay, you should always "win". I think modern
designs use less potential drive on PS_ON# than the design
from the old days.

There was some kind of OC driver that I've forgotten the
number now, but it had sufficient drive to drive a
doubly terminated 50 ohm transmission lines. That's how
I buffered clocks on a board with a lot of
discrete clocks. Four of those, one per quadrant.
So in terms of drivers for the PS_ON#, we don't lack
for "gorilla" chips to do the job. The idea of using
8mA drive is so... yesterday. There is a chip
which will drive a 2 amp load, but it's not OC,
so would be inappropriate for the job. I never
got one of those for eval though - would have
been fun to play with. That was a CMOS chip
where the circuit wasn't even considered a logic
gate, more like an amplifier with a gain of 0.9
or so (the output amplitude is slightly smaller
than the input signal).

Paul