I have broken PSU I've worked on before, and had most of it apart. Today I finally measured the 3.3Vrail while it tries to start (w&w/o PC).
So with a Tek2430 scope, and found it only gets to about 1.5V hooked to the PC, 1.7V w/o PC, and just sits there. The supervisor chip waits 75ms before shutting it down.
The trace of this tho does not look like a normal cap charging, it looks more like a parabola. It only takes 1ms to get to about 1.3V, and then it takes even dips a bit for 10ms, then by 30-40ms it at 1.5V (or 1.7Vw/o load)
The 5 and 12V rail look like normal exponential charging curves. The 3.3V trace starts slow, like a parabola, with a normal vector pointing to the upper-left. Sounds strange to me but I'm pretty green.
I actually I had most the cap's out and did ESR tests, so IDK. I'm about to map the 3.3V section.