Electronics > Beginners

Puzzled about Online UPS + GFCI Behavior

(1/6) > >>

DrDeke:
The following question came up in conversation with a friend recently:

"What if you took an online (double-conversion) UPS, plugged it into a GFCI (RCD)-protected receptacle, turned it on, plugged a GFCI tester into one of the UPS' output receptacles, and pressed the button on the tester? Would the GFCI on the receptacle supplying power to the UPS trip?"

After discussing how we think GFCIs, double-conversion UPSes, and GFCI testers work, we both came to the conclusion that nothing would happen when the test button was pressed. Then we went and actually tried it and, to our surprise, the GFCI trips every time.

The UPS we are using (a Liebert UPStation GXT1500RT-120) has a block diagram in its manual:



We're in the United States, so the utility power at the receptacle comes in on three wires, "hot", "neutral", and "ground". The neutral and ground wires are bonded together at the building's service entrance, and the hot wire has 120 V RMS on it.

The wall/utility receptacle has a built-in GFCI. My understanding of how a GFCI works is that it measures the difference in current flowing through the hot and neutral lines in the receptacle and, if this difference exceeds a few mA, trips, interrupting power to the circuit.

My understanding of how an online/double-conversion UPS works is that it it draws AC utility power, converts that power to DC, and uses the resulting DC power to maintain some batteries in a charged state and to power an inverter. The inverter's output is connected to the output receptacles on the UPS. The output receptacles of the UPS are powered by the inverter during normal operation (when utility power is available) as well as during utility power failures. The only time there would be a direct AC path between the [hot wire of the] UPS input and output receptacles would be if the UPS is in bypass mode. This particular model of UPS has an LED indicator on the front panel which illuminates when the UPS is in bypass mode, and it was not in bypass mode during any of our testing. As far as I understand, all of this is consistent with the block diagram from the UPS' manual.

My understanding of how an external (meaning not built into the GFCI device) GFCI tester works is that when the button on the tester is pressed, a resistor is connected between the tester's hot and ground prongs. This causes less current to flow through the tester's neutral prong than is flowing through its hot prong, which should cause the GFCI the tester is plugged into to trip.

I don't have a detailed understanding of how an inverter works, and the block diagram is a bit light on details about how the neutral and ground lines are connected (if at all) to the inverter.

It seems like the neutral line would have to be connected to the inverter for it to work. If this is the case, and if the neutral and ground lines were bonded to each other inside the UPS, then i can see how current would flow from the inverter's hot line, through the tester, out the tester's ground line, then back to the inverter through the ground/neutral bond inside the UPS. But I don't see how this would result in different amounts of current flowing through the hot and neutral lines in the receptacle that is feeding power to the UPS, and thus I don't understand why the GFCI in that receptacle would trip.

If any of you have an idea as to what might be going on here, I will be very interested to hear your explanation!

sokoloff:
Given that’s the block diagram, it seems clear that the neutral and ground are pass through and the tester that causes some of the current to return on the ground leg will trip the GFCI.

(I’d have reasoned the same way you did before testing and prior to looking at the block diagram.)

DrDeke:
But since the inverter is the "source" of the current flowing out of its hot wire, in order for any current to be able to flow "out" of the tester's ground wire at all, wouldn't that current need a path back to the inverter? I guess I'm not seeing how that path would involve the hot or neutral wires of the receptacle supplying power to the UPS' rectifier.

DrDeke:
Oh! I guess if the neutral wire is connected to the inverter (which it seems like it must be), but the neutral and ground wires are *not* bonded inside the UPS, then the current would flow out the inverter's hot wire, into the tester, out the tester's ground wire, through the UPS on the ground line, into the wall receptacle's ground line, back to the building's service entrance at which point it is bonded to the building neutral, then back down the line to the receptacle, through the receptacle's neutral wire, into the UPS, where it would return to the inverter.

If this is correct, then the current coming "toward" the UPS on its supply neutral would offset some of the actual load current returning toward the utility from the UPS, which would definitely cause the GFCI to trip. Perhaps that's what's going on...

sokoloff:
Yes. That is my assumption.

Navigation

[0] Message Index

[#] Next page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod