0
\$\begingroup\$

An inline GFCI rated at 15A 120VAC with 4-6mA trip level is used to provide 120VAC to an AC-DC converter. The converter outputs +12VDC with max 1A current. This is then used to power a high voltage DC-DC converter that takes +12VDC input and outputs 1000VDC with maximum 500uA output current. The GFCI and DC components share ground*.

*Edit: The GFCI does not have a "ground." I am referring to the third prong (green wire) here which is connected to chassis/ground.

The 1000VDC output is shorted to the ground of the AC-DC converter output. Will the GFCI trip? Please let me know if more information is needed. This is totally hypothetical. I'm simply trying to better my understanding of GFCIs.

Here is a bad schematic depicting this scenario: enter image description here

\$\endgroup\$

3 Answers 3

2
\$\begingroup\$

A typical GFCI/RCD cares about one thing - whether the line and neutral currents are equal and opposite. When equal, there's no leakage. When unequal, there's leakage somewhere (maybe to ground, maybe to another circuit, maybe to a damp windowsill, and maybe through a person), and the GFCI/RCD trips to protect life and property.

For your circuit, the answer depends on the details of the AC-DC converter, specifically whether its AC (line-neutral) side is isolated from the DC (+12VDC/ground) side. I'm going to assume that insulation throughout the system is good enough that current doesn't flow through any undrawn pathways.

If it's isolated in that manner, then the line and neutral currents via the GFCI will be balanced (loop 1 in the following figure):

enter image description here

Note that I assumed the AC-DC converter was isolated as well, with separate loops (2) and (3). If the AC-DC converter were not isolated, the two loops would be connected, with some current also flowing out of the HV DC-DC converter's ground back to the ground connection of the isolated AC-DC converter.

Note that the short circuit is immaterial - we could have any load, regardless of whether it's a short circuit, normal load, etc.

Most consumer AC-DC converters are isolated in this manner; the DC loop is separated from the AC loop (e.g. by an internal transformer).

The ground connection to the GFCI is also incorrect. The GFCI circuit doesn't actually use ground as part of its protection circuitry; only the current mismatch between line and neutral is of any importance.

On the other hand, if the AC-DC converter were somehow un-isolated1 and you inadvertently bonded earth ground with GND/chassis and neutral at the converter, the RCD could trip even without the HV DC load having a short-circuit fault:

enter image description here

While the actual current flow is more complex than the simplification drawn below, the outcome is the same as drawn - line and neutral currents are no longer equal (just like if a normal ground fault occurred), which causes the RCD to trip.

Of course, if the GND/chassis were floating relative to earth ground, the current marked by (3) would not flow. The only return path for line current is again through neutral. However, that chassis could now be hot, and a person touching it could get shocked, while creating the ground fault that the GFCI should trip for.

1 I'm not aware of specific examples of commercially available power supplies for this, but there are certainly ways to make such a topology in principle (e.g. capacitive dropper)

\$\endgroup\$
1
  • \$\begingroup\$ Thank you for this answer. To clarify, the GFCI "ground" in my diagram is simply the third prong of the connector going to the wall. The third prong is not connected to anything within the GFCI and looks to just pass through it. \$\endgroup\$
    – earl
    Jul 7, 2023 at 20:26
1
\$\begingroup\$

Whoops! That drawing shows safety ground going to the GFCI. It does not.

GFCIs/RCDs do not need or use safety ground. As implemented in North America, GFCI breakers have no access to safety ground. GFCI deadfronts designed to mount in a standard junction box, take safety ground only to ground the cover plate screws. GFCI receptacles take safety ground only to ground the cover plate screws and deliver it to the socket ground pins.

The GFCI has no relationship to safety ground, so stray current on safety ground will be of no concern to it... as long as it's isolated from the AC source it is monitoring.

It is super important to understand that AC mains safety ground, despite sharing a name, is completely unrelated to electronics 0V reference and current return aka Vss aka "GND". It is faulty thinking to presume Vss/GND should be connected to AC mains ground. There may be a reason to do that, but if there isn't, there isn't.

\$\endgroup\$
1
\$\begingroup\$

No the GFCI will not trip.

The GFCI monitors difference between mains live and neutral.

Under normal conditions, they should be identical when supply has two-prong mains plug. If supply has three-prong mains plug, some few hundred microamps of current leaking from live to ground will happen via mains input filter, which is intended. Only if the supply develops a fault (e.g. moisture or component damage) and mains live leaks more than few hundred microamps to ground, maybe a few milliamps, then the difference between neutral and live is too much and it will trip.

\$\endgroup\$

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.