OK, my turn to ask a PC PSU related question...

I have a home built PC based around an Asus A8N-SLI motherboard (manual). An XP3800 CPU, 1GB RAM, decent HD, decent video. At least for its time a few years back. Generally it's operated just fine over the years and is just fine for SWMBO to dor her Facebook, email, Itunes etc.

A number of years back it wouldn't power up unless you held the power button down. At the time, I didn't have the get up and go to fix it since I found that by tying the green PS-ON low on the power supply it would start. Do that and it would start. Press the power button, remove the PS-ON short and it would keep going until it was shut down. Since it stayed on permanently I perservered with that for a couple of years. I did try another supply temporarily and it worked correctly at this stage. That supply got stolen for a HTPC though.

Finally after a couple of years, the supply wouldn't start at all - even permanently shorting PS-ON to ground. So I took it in and they replaced the supply just inside the 3 year warranty (Antec supply from a Sonata case). I got a better supply as a replacement in fact.

OK, so I thought "cool, new supply, will all be good again". Now however it will only run with the PS-ON shorted to ground which means I cannot shut it down properly without using the power switch on the back. I want to hibernate it in the interests of being a bit greener and cutting back on the power bill (100W 24/7 adds up to ~$130/year). If I leave the PS-ON to ground short permanently fitted and use the rear power switch, I can't do things like Wake on LAN if I need remote access. It's a bit of a pain to crawl on the floor to turn it off and on too.

Whilst in "off/standby" mode, the green motherboard standby LED still lights up as expected. I also have checked the extra ATX+12V connector is fitted. I even tried manually shorting the power switch header in case the power switch was a problem.

Without understanding exactly how mother boards work in terms of holding the power on, I'm at a bit of a loss. I assume the Power switch forces the supply on enough to get the motheboard up and running and then the motherboard hardware holds it on through the PS-ON signal. Somewhere in that chain my board appears broken...

Any ideas? It all points to the motherboard at this stage. I'm loathe to spend $100 on a secondhand motherboard. Since it's an old Socket 939, I'd need new RAM and a new CPU to upgrade it in any other way. Apart from this issue, it works just fine, happily running 24/7 at the moment with no issues at all.

Could it be something like this?

In particular, look at Tony's reply to my post.

tanstaafl.

You got another motherboard you can just temporarily power with that PSU? You're supposed to have a load on the 5V/12V lines when testing a PSU so you can't just short the PS-ON line with nothing else plugged in.

The motherboard is always supplied with power via the 5V standby connection from the PSU. The hardware that monitors the power switch is powered via that. You push the button and it'll signal a microprocessor on the motherboard that you want to power up. The microprocessor will then handle the PS-ON signal etc...

tanstaafl:
I don't think so. I only have a couple of USB things plugged in (mouse, keyboard and scanner). It's pretty barebones as is but I will remove all but the CPU/RAM/Video card to be sure. Even the video card and RAM could come out. At least a power up and some beeps would verify things.

tman:
I'm only shorting PS-ON to ground with the PSU connected to a motherboard. I'm well aware of the need for a load on switchmode supply.

The standby +5V is working since the motherboard standby LED is lit as it should be.

I do have another board I can try (although only 20 pin ATX and not 24 pin ATX). I'll try swapping supplies on both machines and see what happens. Hadn't done that yet because the wiring is all neatly bundled up and routed at the moment.

Is there a small microprocessor responsible for the PS-ON signal whilst in standby? I assume it must be since a board will generally still start without the main CPU won't it? If there is a small microcontroller no doubt it's embedded in some other chip making signals hard to get to. Not that replacing/fixing that is likely to be possible. Such logic could be fairly easily done without processing so I'm not sure about an actual processor being involved. I'll have to pull the board out of the case to look closer I think and maybe try and trace the power switch signal somewhere. Don't really like my chances.

FWIW, a buddy of mine brought over his "modular" PSU to get me to replace the fan on it.

After a successful transplant, we attached it to one of the many known-good ATX boxes here for a quick smoke test, and the darned thing wouldn't power up.

Well, it would briefly, while we held the softswitch in, but not for more than a couple of seconds at best.

After much panic and reexamination of the fan wiring etc.. we tried it with a different motherboard. No problemo.

Just fussy, I guess.