Roy Lewallen wrote:
No, the GFCI doesn't need the safety ground. Like the device you're
describing, it measures the difference between the "hot" and "neutral"
wires, by running the pair together through a transformer. A GFCI is
supposed to trip in 1/40 of a second at 5 mA.

In our system, like yours, the "hot" and "neutral" wires should always
carry equal currents, and the safety ground shouldn't carry any.
However, the safety ground is sized the same as the current-carrying
conductors so it'll handle a fault current large enough to kick the
circuit breaker.

That is pretty much the same as our RCDs, right down to the large-sized
'protective earth' wire.

Its primary function is to protect against something like the "hot"
wire making contact with a metal appliance frame, which should be
connected to the safety ground.

An interesting thing happens when you switch off to isolate part of the
house wiring, and then cut the "dead" three-wire cable. (We call it
"flat twin and earth". It has PVC-insulated hot and neutral wires with a
bare ground wire in between, all in a flattish grey or white PVC sheath.
Is this what you call "Romex"?) If you cut the hot and ground wires
together, nothing happens; but if you cut the neutral and ground wires
together, the RCD trips. That's a puzzle the first time it happens, but
it's because that cable isn't quite as "dead" as you think. the
isolating switches interrupt only the hot wire, so the neutral and earth
are still being shared with everything else that's switched on in the
house.

Does that happen in the USA, or do you have two-pole isolating switches
for individual sub-circuits?

This isn't totally idle curiosity. If there's an electrical problem
while we're staying with the in-laws in Atlanta, I'm the one who's
supposed to know about these things...

When it comes to a GFCI, no assumption is made about how much of the
imbalance current is returned via the safety ground and how much via
the Earth or some other path.

Exactly the same as our RCD, then... just a slightly ambiguous name.


--
73 from Ian G3SEK 'In Practice' columnist for RadCom (RSGB)
Editor, 'The VHF/UHF DX Book'
http://www.ifwtech.co.uk/g3sek