PSU having to be forced to turn on...

[jamena]

Got an issue with my PSU (300W in the Silverstone LC11m) where it no longer responds to the power button...

It worked fine for about 2 years, only (minor) change was to replace the old 9800pro with a passive nvidia 6200 (so...less power hungry?). This is when the psu started it's current problems.

Plugging in the mains cable at the back of the psu causes a sort of electrical whine to start up that continues for about 10s after the mains cable is unplugged again (there's no main on/off switch hence the plugging/unplugging). No response is had from the power button, or from subsequently shorting the green+black ATX pins.

Now: Starting with the thing totally disconnected from the mains this is the process I have to follow to make it boot:
1)[starting totally disconnected from the mains] short the green+black pins on the psu
2)plug in the mains cable [fans spin up, lights come on]
2a)if fans fail to light up, disconnect mains, wait for electrical whine to dissapate (~8-10s) and repeat
3)press the reset button [pc subsequently boots and runs with no further problems]

When the problem first started I was sometimes able to "un-short" the green+black ATX wires at the moment of hitting the reset switch and it'd carry on working and shut down normally. Now I have to wait a couple of minutes of the shutting down message and the flick the green+black switch, turning it all off. Not elegant but it works.

I have actually wired in a switch so I can short those pins from outside the case since it became a permanent problem. I've plugged in a new 600W ATX psu and my pc booted up as normal via the power button, so I think it's definitely something to do with the PSU. I also tried removing all the PCI cards to further reduce the load but no change, still weird problem.

Any suggestions wrt to repair before I think about replacement?
Thinking about replacement - any suggestions of other psu's that might fit?

Many thanks, in advance!

[peterb]

My first thought would be the PSU. A 300 Watt decent make PSU can be had for under £30.

[jamena]

It's a custom-size psu for the silverstone case. Rated at 300W, and can be replaced for a bit under £50 I believe. But, thing is it still runs the pc absolutely fine, rock solid, but only once forced to turn on as described above. If anyone knows more about the circuitry a psu might use to respond to the power-switch from a standard motherboard that might let me fix it. If it's just a relay or something it seems crazy to replace it...

I've plugged in a new 600W ATX psu and my pc booted up as normal via the power button, so I think it's definitely something to do with the PSU.

My first thought would be the PSU. A 300 Watt decent make PSU can be had for under £30.

[vinnyT]

I suppose you have tried the pc with the old 9800 Pro gfx card in, just to see if this affects the boot behaviour? (I really doubt it will, but then you will have tried all you can.)

[jimbouk]

Well the power switch goes through the motherboard so have you thought about that?

[peterb]

Traditionally a "soft start" type of system used a small, permanently on auxilairy PSU which was used to switch a relay, which was then held on by the main PSU.

However modern switched mode PSU's are a little more complex, although the pribciples remain the same.

Most of the functions of a SMPSU are handled by a dedicated controller - shorting the two connections on the loom applies a signal to the PSU controller which switches on the main PSU. There is also an interactiuon with the mobo co control shutdown (pressing the on/off switch for more than 4 seconds) or standby - and that is a bios/software function, so the interaction is complex.

However the orinciple remains that there is a small always on auxilary PSU (which may be rerlated to the WOL auxilairty supply - not something I have researched)

My best guess (and it is only an educated guess) is that the auxiliary supply (located within the main PSU) is faulty. It might only be something as simple as a failing capacitor - or it could be a problem obn the circuitry associated with the PSU controller.

The fact that the system works with a different PSU tends to point in the PSU direction.

It may be fixable, but diagnosing the problem down to component level would be time consuming, and given that we are ultimately probing around high energy circuits, would need to be done very carefully. Without knowing your level of experience in dealing with electronics, it isn't a course I would recommend you do yourself, particularly without a mfrs cct diagram.

If you paid someone to do it, the cost would probabbly be more than the replacement cost of a new unit. Sadly, such are the economics of our "throw-away" society.

[jamena]

I suppose you have tried the pc with the old 9800 Pro gfx card in, just to see if this affects the boot behaviour? (I really doubt it will, but then you will have tried all you can.)

When I have time I was thinking about trying to put the 9800 back in direct into the agp port (case requires a riser+extender be used) and in fact the problem started when I was installing a newly purchased 6200 (passive). Not sure quite how that could have caused a problem though and given that it all works perfectly with a new psu wired in I figure it's almost definitely the psu alone.

modern switched mode PSU's are a little more complex, although the pribciples remain the same.
<snip>
Sadly, such are the economics of our "throw-away" society.

I should have time to have a proper think about it in a few weeks. I have some experience in elec eng though it's mainly in the software/firmware side of things now (been a while since I built a smps(u) at uni and that was just a simple on/off at the mains affair). If I can find a circuit description (unlikely?) I'll have a go, and I suppose the likely problem based on what you described would be the aux on/off circuitry failing to keep the main psu in the on state for some reason (hopefully not a safety related one...hmmm)

[peterb]

I should have time to have a proper think about it in a few weeks. I have some experience in elec eng though it's mainly in the software/firmware side of things now (been a while since I built a smps(u) at uni and that was just a simple on/off at the mains affair). If I can find a circuit description (unlikely?) I'll have a go, and I suppose the likely problem based on what you described would be the aux on/off circuitry failing to keep the main psu in the on state for some reason (hopefully not a safety related one...hmmm)

That woudl be the area I would start looking at. Unlikely to have safety implications as all the output will be isolated from the input with a transformer.