Hmm I have some ATX PSUs lying around. What was putting me off was adding the +28v needed for the GE. Having said that, I could live without powered firewire and I don't have an ADC monitor so I wonder if Bunsen's tip about wiring the +5VSB line into the +28V pin would work. Using a wall adapter to supply the +28V sounded like a messy solution.
I pulled out two of the PSUs I have. Both are 300W which is lower than the 338W the original AcBel one had. The first one has both power connectors on the back, while the other only has one and an on/off switch. This makes the first one a better fit but it has a lower amp rating on the +5VSB (1.5a vs 2a on the second PSU). Would that make the second one a better candidate for the +28V re-wiring job? Or are both out of the pciture because they're lower power than the the original.

Click to expand...

If you're feeling energetic, you can build this:

http://www.io.com/~trag/28v_vreg.pdf

Note, that the contents of that file are not my work. You won't get enough 28V amps to drive an ADC monitor, but it will be enough for Firewire devices.

Since I found that file I've looked into "Boost" and "Buck-Boost" regulators at length (for aquarium LED lighting). There are a bunch of other choices besides the LM2577, but that one is already spec'ed out and seems good enough.

I don't know if it is still true, but in the distant past, the 5V-SB would tie to the regular 5V supply after the power supply activated. You could also do this with a rectifier and a current limiting resistor, so that the 5V can supply the 5V-SB but the 5V-SB will not supply the 5V If this is the case, then the current limit on the SB doesn't matter much, as there won't be much draw unless the computer is turned on. Of course, one will still be limited by the capacity of the boost circuit which is probably well under 1A total.