I removed an SCR, 2N5061, with this behaviour:
connecting an HD (or a load) doesn't shut off the PSU anymore, but
voltage drops from 5 to 4.2, and it's not enough.
I moved the trimmer with slight advantage (4.5) then the PSU shutdown
again.
I had to wait some minuites, discharge all caps before it started too
work again.
Under the SCR there's a fast switching diode 1N4148F, with a constant
reverse voltage of 1.40. When everything goes bad, it's the first to
receive 0 voltage.
Across A-K of the SCR I read about 4.34 V, G and K are equipotential.
May be this is the right zone to investigate ?
A friend of mine owns a similar PSU (same PC), working, but he'll be back
after easter.
Having a similar model could help me tracing down the fault...