1
\$\begingroup\$

An ATX motherboard is grounded to an ATX PSU thru the ground wires, from the 24-pin power connector to the PSU.

Is there (inside the ATX PSU) galvanic isolation between the DC output and the AC input?

If so, does this mean that the ground wires (between motherboard and PSU) are not connected directly to the earth prong of the AC inlet?

\$\endgroup\$

2 Answers 2

4
\$\begingroup\$

The input power (North America) is line and neutral. On the same connector is a safety earth connection.

The power supply galvanically isolates the input power from the output power (several voltages with a common).

The earth prong is connected to the metal chassis for safety reasons (on desktop computers), including an internal connection to the metal case of the power supply itself.

Typically, on virtually all desktop PCs and many notebooks, the output power common is also tied to the chassis and the earth prong. On some notebook computers, the chassis is floating.

\$\endgroup\$
4
  • \$\begingroup\$ With "output power common" you mean: ground-plane of the motherboard? \$\endgroup\$
    – user127725
    Commented Nov 13, 2016 at 19:35
  • 1
    \$\begingroup\$ The power supply has outputs of +3.3, +5, +12 (maybe others) and a common. The common will be connected to the ground plane of the motherboard when the power supply is plugged in, and it's usually connected to the chassis of the power supply as well. \$\endgroup\$ Commented Nov 13, 2016 at 19:41
  • \$\begingroup\$ Thanks for explaining, but why would you connect the common to the PSU chassis? The PSU chassis is grounded to the earth prong. The common is the return path from the motherboard to the PSU, or isn't it? The ground-plane is already grounded to the PC case thru brass motherboard standoffs. \$\endgroup\$
    – user127725
    Commented Nov 13, 2016 at 19:48
  • \$\begingroup\$ It's grounded in many places for EMI and safety reasons. \$\endgroup\$ Commented Nov 13, 2016 at 19:52
1
\$\begingroup\$

Try to read something about safety classes of electric equipment.

The basic trick in modern mains wiring appears to be, that the working ground return (the "neutral") is deliberately separate from the protective earth. All the voltage drops caused by working currents happen in the "neutral" wiring, there should be no permanent current flowing through the PE conductors. Thus, any earthed chassises etc. have a nice solid earth potential, the PE terminals etc. are not stressed by the "working" currents and are not directly endangered by faults in the "working grounds". When a bolt gets loose in a working ground terminal that is under load, anything supposedly "neutral" after that failed joint automatically reaches dangerous voltage levels. When a bolt gets loose in the PE wiring, you will get some PE segments unprotected, but at least they do not inevitably start to kick around. Also, this topology allows you to sense that some current "leaks out of the live+neutral conductor pair" = finds some other path back to the source, which means that humans are in danger. This detection is easy to do using a "relay with a common mode current-compensated coil" inserted in the Live and Neutral - called an RCCB or GFI. The overall morale from this: anything yellow-green must be kept in perfect shape, first and foremost.

Here in Europe (240V / 50 Hz), we have some standards-based electric safety classes, the most important ones being:

  • "safety class I" = metal chassis connected to the PE prong in the three-pin wall socket
  • "safety class II" = double isolation, plastic shell, PE prong unused in the mains cord (the cord is two-pin)

On several occasions (stereo systems powered by SMPS, notebook computers) I've encountered electric devices nominally in class II, where the output "residuals" of the switch-mode PSU were causing problems: noise in analog audio interconnects, tingling in your fingertips if you touched some part of the chassis (often related to connector shields) etc. - and, earthing the culprit device to PE often helped.

Take a look at this very conceptual sketch:

switch-mode PSU, parasitic RF blocked by a cap, residual AC current

I mean to argue that just earthing the secondary is still a compromise (tradeoff). Apart from the residual current, another possible downside is the emergence of ground loops involving signal grounds in networks of multiple devices, interlinked by direct interconnects with shared signal reference grounds, especially if single-ended transmission lines are used (as opposed to balanced). To live with ground loops in networks, you need balanced inputs, and to get rid of ground loops, you need galvanic isolation elements...

See also the differences between SELV, PELV and FELV = different styles of powering electric equipment by so-called "extra-low voltage". In a dry and generally low-risk environment, anything below 50 V AC or 120 V DC (under 60 V DC in moderate risk environments) is considered "pretty safe", probably in the sense that it's unlikely to kill you upon a random touch. Thus, the secondary of a PC PSU qualifies as the pretty safe "extra low voltage" category. This is possibly why the electric safety norms do not object against earthing the output common terminal of such power supplies.

The standard style of powering office/home PC internals is PELV. It is a design tradeoff taken many years ago and to date it still makes perfect sense :-)

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.