Power supplies have on/off switches now?

[Mxsmanic]

your kidding right? for YEARS ATX's came without a power switch.

I suppose that for years I had no reason to look. On the last few PCs
I had before I began to construct my own, the button on the front was
an AC switch (in some cases, working through a bizarre lever
arrangement with the PSU in back). It may be that the PSU had no
built-in switch but there was a switch external to it. I don't really
remember now.

if the user knows anything about it, or remembers to use it. fail safe?
nope

Fine. You can build your machines without a real power switch, if you
want.

 

[Pelysma]

Hitting the power button serves no purpose if the main switch has
completely denergized the power supply.

Apparently it doesn't. There are capacitors still holding a charge until it
is released or bleeds off. More importantly, there is a voltage
differential in the memory slots that can potentially damage memory being
removed from the slots.

Try this: shut down your computer, unplug it or switch it off, then open up
the side and watch in there as you hit the power button.

Several LEDs light for about half a second and fans move slightly. Then
you're all balanced out and it's safe to work inside the case (although you
still need to take precautions for static charges on your clothing and
skin).

 

[Mxsmanic]

Apparently it doesn't. There are capacitors still holding a charge until it
is released or bleeds off.

They aren't going to be discharged by hitting a completely unrelated
momentary contact switch.

Try this: shut down your computer, unplug it or switch it off, then open up
the side and watch in there as you hit the power button.

Several LEDs light for about half a second and fans move slightly. Then
you're all balanced out and it's safe to work inside the case (although you
still need to take precautions for static charges on your clothing and
skin).

That doesn't happen on my PC. Nothing lights and nothing moves unless
the main power switch on the PSU is turned on.

One wonders how anyone worked on computers before that little
momentary-contact pushbutton existed, eh?

 

[Jim]

I mentioned alternatives to using the on/off switch on the power supply.
These are even "more elegant", and more convenient as well.

.............
Jonny

 

[Pelysma]

They aren't going to be discharged by hitting a completely unrelated
momentary contact switch.


That doesn't happen on my PC. Nothing lights and nothing moves unless
the main power switch on the PSU is turned on.

It does happen on both of mine, a storebought Dell 2350 and homebuilt
P4-based system. I honestly can't say about the two older ATX machines.

For me this is just habit now: when I open it up, I first power it off with
the start button, then unplug or switch off, then pull the speaker cord,
then finally hit the start button again. After that I either wear a wrist
strap or keep one hand on the frame or PSU box.

Quick poll: Both P4's and one of the PII/400's have switches on the power
supply. The other one I have to unplug (well, actually, I turn off the
power strip). The remaining two computers in my household are a 486 and an
odd proprietary AT/ATX hybrid, and both of those have AC switches in the
front.

And, if I were buying a PSU, I would definitely want one with an AC switch.

 

[Andy]

There seems to be a lot of disinformation around this subject.

An ATX power supply may have a switch on the rear panel and, if
present, this is an absolute power on/power off switch. In other words,
if you set the rear switch off then there is no power anywhere.

Most confusion comes from what happens with the on/off switch on an ATX
PSU/Motherboard combination...

In the olden days, the PSU (Power Supply Unit) switch was a 240v /
kill-you-dead switch. Now what happens is that when the PSU has any
power at all, the switch is always being fed a signal. For the techies
amongst you this means that, while the PSU appears to be down it is
still applying power to a handful of rails (Notably VSB on pin 9, of
which more later).
This then allows us (case and computer manufacturers) to add a
low-power on/off switch which is effectively a momentary contact switch
between two connectors on the ATX PSU connector.
The ATX connector has pin 14 designated as PS-ON (Power Supply On) and
when you hit this it connects pin 14 to ground (any of the ground pins
on the ATX PSU connector) and this instructs the power supply to supply
voltage to the main power rails.
In return for this good grace, the PSU expects a *good* signal to be
applied to the PWR_OK line (pin 8) within 0.25 seconds.

PWR_OK is an output from the power supply. As long as it remains low,
the CPU is held in a reset state. Once the power supply voltages have
stabilized, the power supply raises PWR_OK, allowing the CPU to start
running.

The tricky bit in all this is that grounding pin 14 (PS-ON) after the
system has powered up is interpreted as a shutdown signal to the
operating system and it is up to the O/S to begin an orderly shutdown
when it receives this signal.
If the circuitry sees a PS-ON to ground (i.e power off) signal for more
than 4 seconds then it will perform a PSU shutdown.

There's nothing tricky about PS_ON#. As long as the motherboard holds
it low, the power supply remains on. When the operating system is shut
down, it causes the motherboard to release PS_ON#, turning off the
power supply. The other way to turn off the power supply is to press
the power button for up to four seconds, which causes the motherboard
to release PS_ON# either immediately or after four seconds, depending
on the BIOS setup setting.

 

[skotl]

Thanks for the correction regarding PWR_OK, Andy, but PS_ON should not
be held low by the motherboard - it's a momentary contact to ground to
power on and again to signal the OS to shut down or, if held to ground
for 4 secs, instructs the PSU to pwer down.

Regards, Scott

 

[Pelysma]

I'm noticing that all power supplies nowadays seem to have an on/off
switch on the back. My 5-year old Antec doesn't have one.

Working on two Gateway pc's tonight, one a year old, the other practically
brand-new, and neither one has an AC switch in the back.

 

[Andy]

Thanks for the correction regarding PWR_OK, Andy, but PS_ON should not
be held low by the motherboard - it's a momentary contact to ground to
power on and again to signal the OS to shut down or, if held to ground
for 4 secs, instructs the PSU to pwer down.

Regards, Scott

If you have a volt meter, measure PS_ON# while the computer is
running. What you're describing is the behavior of the power button on
the front of your PC. It is possible to configure in the BIOS setup to
make the computer go into standby by hitting the power button. This
behavior of the power button has no direct relationship to PS_ON#.

 

[Aldric]

Pelysmawrote
Working on two Gateway pc's tonight, one a year old, the othe
practically

brand-new, and neither one has an AC switch in the back

THIS is because gateway is a bad company. :

 

[Mxsmanic]

Working on two Gateway pc's tonight, one a year old, the other practically
brand-new, and neither one has an AC switch in the back.

No two gateway machines are alike, though, so one must be careful
about drawing conclusions from them.

 

[larry moe 'n curly]

but PS_ON should not be held low by the motherboard - it's a
momentary contact to ground to power on and again to signal
the OS to shut down or, if held to ground for 4 secs, instructs
the PSU to pwer down.

I thought that the mobo converts the pulse from the front panel power
button into a constant low signal to PS_ON. The PSU will run only as
long as its PS_ON signal is held low.

 

[skotl]

Well, this is where my confusion increases.

With apologies to more learned posters who may be able to correct me,
the front panel switch is a momentary connect (i.e. pull to ground) and
the ATX spec recommends/enforces that some debouce circuitry is
employed that treats a number of mechanically generated
bounce-to-ground as a single signal.

This is where I part company with other posters (with regard to the
original spec) that the switch should be a momentary ground, and should
then be left to float. This leaves the potential later on to ground the
switch for an OS shutdown, or a forced power off (4 seconds of PS_ON to
ground).
As it happens, you can latch PS_ON to ground and leave it there, and
the PSU *should* power up and stay up. This, however, is a happy
coincidence and is not what the original ATX specs intended.

Ta,
Skot.

 

[larry moe 'n curly]

(e-mail address removed) writes:

I don't recall ever seeing a power supply without one, although it
would not surprise me if some low-end models lacked a switch. It
seems a bit unsafe, though.

Every cheapo PSU I've seen had a rear switch, but some high-end ones
didn't, including a PC Power & Cooling, a Delta, and a Zippy Even the
300W Channel Well given to me lacked one, but otherwise it's identical
to an Antec SmartPower.