[tootle]

Sboom, connect everything to the pc and retest by backprobing the atx connector.

[Sboom]

First of all I must thank you for your outstanding support !!!

Back on track. Unfortunately I've already done what you're suggesting: it was included in the precious tutorial you gave me. Every voltage was correct with the only two exceptions I highlighted. Not big things, indeed, and I'm getting frustrated...

[tootle]

Sorry wasnt aware if youve tested like that with your devices plugged in youve done a load test what reading did you get for power good? What pins were the exceptions?

[Sboom]

tootle, sorry for my late answer, I've been quite busy today. Let me give you further details about my post #29 in this TD:

1. without load every voltage was in the range with the exception of all (those in the ATX 20pins and ATX 6pins) the pins of +5 V: they all were at 5.31 V (out of range)
2. under load every voltage was in the range; I must underline that I have an ATX 6pins NOT connected to any device. I only noticed a sequence of 12.36-12.45 V for each time I powered on the PC. This voltage was in pin 14 and in the ATX 6pins connector not connected to any device.

I retested today the PC under load and I've found that the sequence a) had a wider range and b) included the pins bringing the 5 V voltage: every time I powered on the PC the 5V pins and the 12V pins had the voltage (5.14-12.45), (5.25-11.91), (5.14-12.45), (5.25-11.91), etc.

All values are still in the tolerance values but the pattern "one OK - one fault" is evident.

My hope is that could be the PSU that is getting worse. I'm planning to find a spare one to run the ultimate test.

PS I really hope that I've not further confused you rather than clarify !!!!!!!!

[tootle]

Please keep in mind you must run tests with the system fully loaded. this means everything connected and you backprobe, Im sorry if the link I provided has caused some confusion as I did not check his method.

Your 3.3 must be within 4% if the power good signal is not between 3-6v you need to replace it. when you test your psu make sure the 6 pin cpu power connector is plugged in and just get the reading by backprobing. do not disconnect anything when you are testing simply backprobe. as far as tolerance goes for poorer quality power supplies 10% is considered acceptable and known as a loose tolerance. Higher quality supplies use a tighter tolerance of 5%, sometimes even lower, although the ATX specification dictates that they should run within 5%. An exception to this rule is the power good signal on pin 8 ( I suspect this to be faulty) and the 3.3V rail which must be within 4%, if any of these conditions are not or any rails tested when backprobed are out of range replace the supply,

[Sboom]

I must sadly report that I run my "ultimate test". I replaced the PSU but I had exactly the same behavior as always: one start was OK, the following was a freeze.

Unless you don't have other suggestion I stop the research: the culprit cannot be anything else than the mother board.

A big, HUGE thanks to whom supported me !!!

[tootle]

You might have a sticky switch disconnect it from the mobo and short the 2 pins exposed.

[Sboom]

Done. No joy: always the same behavior !

What should I do? Assume that is the mobo and give up?

[black-wolf]

It's probably your mobo, but you can try one more thing http://helpdeskgeek.com/how-to/fix-mbr-xp-vista/

[Sboom]

I appreciate the fact that you're not keen to give up (as I am, b.t.w.) but I've already tried to boot without the HDD (connector removed) and with a floppy with a DOS system (in this case the MBR on the HDD is not involved). Same problem as always.

[black-wolf]

OK, motherboard seems to be the only thing left.