# Using PWR_OK to Activate an ATX Power Supply Automatically

Okay so, I have an ATX PSU that isn't connected to a motherboard, and I'm currently using the simple paper-clip trick to ground PS_On in order to trick the PSU into providing power when I need. However, doing this cleanly means wiring in a switch to do this, but my PSU already has its own switch on the back which I'd like to just use instead, rather than two separate switches. Now, I know that simply leaving PS_On grounded all the time isn't really safe to do if I have devices connected.

But then I found out about the PWR_OK signal, which is a +5V logical high signal that is triggered when the PSU considers itself ready (shortly after mains power is provided).

What I was thinking I could do is somehow wire up the PWR_OK, PS_On and a ground pin, such that when PWR_OK goes to high, PS_On is automatically grounded to activate the PSU automatically, but safely. I may still put a toggle switch in there for good measure, but I basically don't want to be able to ground PS_On without the PWR_OK signal.

I'd very much appreciate any help in describing how to do this. Please treat me as a relative novice for any answer; although I'm comfortable soldering things, mostly all I've done is things like creating my own splitter or adapter cables for computers and even then, only simple ones. It's been quite a while since I had to design a circuit, work out what resistors I need (and where) etc.

• You might like to consider ATX Breakout Design - it doesn't address your idea directly, but maybe it will help. – RedGrittyBrick Nov 23 '13 at 12:06

The way a PWR_OK works, or at least the ones I have worked with, is by signaling the stabilization of the power output with a certain tolerance. This is used for sequencing purposes (e.g. When 5V line is ready enable the 3.3V PS).

This means you won't be able to use PWR_OK unless PS_ON is high first.

• Ah! In that case I've misunderstood the specification. I'll probably have to think about what to do then, as currently I have no sequencing; as soon as PS_On is grounded my devices start drawing power; although my particular PSU seems to have a very stable supply, I want to make sure I don't fry anything :) – Haravikk Nov 24 '13 at 12:02
• In that case (making sure you don't fry anything) the feature I have seen in some PS is the control over the initialization. e.g. increasing the capacitance on the Vref pin (if there is any) to create a longer "ramp-like" behavior on the output. This avoids major overshoots but it may delay your control loop, so be careful not to make it too large. – joseppinilla Nov 25 '13 at 16:05

As the other have posted, PWR_OK comes after PS_ON so what you propose won't work. However, there is no reason you cannot just leave PS_ON grounded all the time. For as long as a computer is turned on, the motherboard is keeping it grounded. You just want to make sure that you turn off the power supply before you connect or disconnect any load.

• Uh, no, the computer does not keep the power supply on. It uses the standby 12v or 5v to turn the (rest of the) power supply on, when you turn the computer on. And you can hotswap loads at any time. – Passerby Nov 23 '13 at 23:38
• To clarify: the motherboard is always powered by +5VSB (stand-by), at least as long as the power cord is connected and the back switch is on. To turn on the rest of the power supply, the logic on the mother board (as powered by +5VSB), grounds PS_ON and keeps it grounded as long as the computer is on. When the motherboard ungrounds PS_ON, the power supply turn back off (again excluding +5VSB). As for my no-hot-swap comment, yes, the power supply is fine with adding and removing loads hot. However, this assumes the properly rated connectors... – DoxyLover Nov 25 '13 at 17:20
• The OP is talking about using a PS in a non-computer environment so I'm assuming he is not using such connectors. Thus, my connect. – DoxyLover Nov 25 '13 at 17:21

As far as I understand how ATX power supplies work, you have to pull down PS_ON to turn the power on, and then you get PWR_OK eventually. Without pulling PS_ON, you'd wait forever for PWR_OK.