On Nov 7, 7:45*pm, Ronald Raygun
wrote:

Would you guys mind explaining why mains input filtering, or harmonics,
would "leak" current to earth? *Surely *no* equipment should have *any*
connection to earth other than from the enclosure, and there should be
no current path to earth from any of the "gubbins". *

Older equipment, with transformers, doesn't usually do this. The real
problems began with the advent of switch mode power supplies (c.1980):
a way of building power supplies out of complex silicon (which is
cheap these days) and small high-frequency transformers, rather than
the old heavy 50Hz transformers. As these are high-frequency devices,
they make electrical interference that can be transmitted to, and
interfere with, other equipment.

The solution to this (for both interfererence generator and
susceptible equipment) is to fit input filters. These act as short
circuits to high-frequency noise, shorting it out and dissipating it
between live, neutral and earth. In practice they're usually five
components: a pair of inductors or chokes in the live & neutral lines
act as an open circuit to high frequencies whilst a triangular delta
network of capacitors between the three wires, including the earth,
acts as short circuits to high frequencies.

Capacitors are an open circuit to DC and nearly so to low frequency
50Hz AC, so they don't usually trigger RCDs, certainly not when it's
an older RCD of 100mA. However, as John Rumm points out, it's a
cumulative effect and 30mA isn't a huge leakage allowance to begin
with.

And why would interference harmonics cause a leak?

Harmonics isn't the right term. The thing is that if you switch an
inductive load (such as a transformer or motor) and you don't switch
exactly on the zero-crossing point (which only happens with
sophisticated transistor or thyristor switches) then you're attempting
to switch a sizable voltage instantly. With an inductive load, that
would require an infinite rate of change in the current, which an
inductor will resist. So the outcome instead is a "ringing" current,
where the current and voltage oscillate until damped out. The extra
current drawn during these oscillations is also out of phase between
live & neutral (as they're on opposite legs of the inductive load), so
it's seen by the RCD as if it was a leakage current (it isn't, it's
purely between live and neutral, but it's still picked up by the RCD's
differential transformer). This is a relatively slow oscillation,
compared to that of a switched-mode supply's oscillator, so RCDs are
still fairly sensitive to it.