Originally posted by: MrDudeManI wouldn't really count hard drives, floppies, etc. as ground paths since they are on the same ground line as the ATX connector. It all meets in one physical location, or at least on every PSU I've ever dissected.
No, there are grounds through the data cables, and grounding of the drive/etc frame as well. Several ground paths.
Also, I've tested multiple motherboards laying around for ground planes around the support screws and I've never had it beep on the multi-meter for continuity.
That's a curious thing you're reporting, since I've found continuity as the norm, and just to double-check I checked continuity on two boards in front of me and both (are).
The grounding pads can also be connected to non-metal stand-offs, which removes them as a grounding loop source.
I don't think anybody claimed a board "had" to be grounded through them, but that IS the intention, why board manufacturers deliberately connect them to ground plane instead of the alternative that would be as easy for them - a break in the copper before the hole.
The shielding around the motherboard peripherals is not a true ground source - it is a shield ground with very little or no current running through it.
It is deliberately ground on the parts I'd mentioned it is a "true ground source". If you mean it is not intended as the primary ground return for supply current or signaling purposes, you are right - but it doesn't in any way change that it IS a ground source by intentional design. To say it doesn't have much current through it is beside the point, we were not talking about fitness for any given load, rather that there ARE multiple ground paths and in the end, if you didn't have the primary ground wire delivery from the PSU, it would then be a higher current ground path.
Lastly, the power from an external video card connection will be at the same ground potential as the ATX connecter ground potential because they are the same physical line.
Wrong. Having an electrical connection is not the same thing as same potential. That's the whole point of our distinctions between multiple grounds. When you have impedance and/or loops, particularly higher current to some devices like a CPU or video card, your ground return currents will result in it being a different gnd potential than another ground line.
Maybe I didn't word the above very well. Take a PSU - forget about the system, just a PSU. Connect a 1 Ohm load to the CPU connector. That's 12A through it. Leave a drive plug empty. You now have different GND potential at the CPU plug than at the empty drive plug. "Ideally" there wouldn't be, in a hypothetical system where there were no constraints like wire size, cost, flexiblity, etc. In a real system there is.
I have a sound card in front of me that is not registering the mounting bracket as part of the ground circuit.
Ok, but I didn't claim it was necessary. The vast majority of sound cards do though, there is typically a ground plane over (most of) the back of the card and (again as with a motherboard) deliberate exposed and tinned copper where the bracket screws to it. It is a curious thing that your parts are so oddly made, the odds are against having more than one that is in a limited # of tests. What makes and models are these? I'm wondering if your multimeter has some problem, either that or at least with the sound card you are measuring a part of the bracket that has coating over the metal. On a non-bare metal surface I put a needle tip on the meter probe so it can pierce the surface layer.