GFCI (= RCBO) trips when connected to online UPS

[MX-16_fan]

@all,
@timkb4cq,
@m_pav,
@JayM,
@figueroa:

I am currently testing a rarely used, rather new, high quality 2000 VA online (= double conversion) (https://en.wikipedia.org/wiki/Uninterru ... conversion) Uninterruptible Power Supply (UPS) (https://en.wikipedia.org/wiki/Uninterru ... wer_supply), which, according to its boot-up self-test, battery voltage and overall behavior seems to be fully intact.

The problem I am experiencing is that it makes the central GFCI (also known as RCBO, cf. https://en.wikipedia.org/wiki/Residual- ... I_breaker), which is located before the UPS, trip every now and then.

The GFCI (= RCBO) is to trip when earth leakage is above 30 mA and/or current exceeds 40 A.

My first thought that the UPS's filtering mechanism was adding to much earth leakage to the overall earth leakage sum generated by other appliances (to my knowledge, many appliances have about 3 mA of earth leage in standard operation). Hence I installed a second, discrete GFCI (= RCBO) only for the office room the UPS was in, and tested the UPS with no other appliances running. However, even with that second test configuration, that (second) GFCI (= RCBO) trips every now and then.

Any idea what's going on here?

The GFCI (= RCBO) model used is a high-quality double-pole one.

According to its specs, the UPS is to deliver pure sine wave.

Interesting observation is that tripping occurs even when not even a single appliance is connected to the UPS.


Thanks in advance for any hint , Joe

[timkb4cq]

If the UPS is tripping the GFCI with no devices connected then there is obviously a problem with the UPS.
A lot of UPSs have a level of surge protection on the line input (and/or on the output outlets). I would guess Metal Oxide Varistors of the type found in surge suppressor power strips are the most likely product to accomplish this although a high-end unit might use a faster, more expensive semiconductor.

The problem with MOVs is that they degrade each time they absorb a surge, that degradation exhibiting as a decrease in their trigger voltage, and they are connected from both hot & common to ground. You can probably see where I'm going here. Normal variations in line voltage (I've seen areas where swings from 108v - 130v are not uncommon) can then cause leakage through those degraded MOVs. The only solution is to replace them.
They're cheap but you need to be able to solder.

If you put an AC voltmeter on the outlet providing power to the UPS you will probably see that the GFCI trips when the voltage rises above a certain point. If that is not the case then I've misdiagnosed the problem.

[MX-16_fan]

@timkb4cq:

(...)
A lot of UPSs have a level of surge protection on the line input (and/or on the output outlets). I would guess Metal Oxide Varistors of the type found in surge suppressor power strips are the most likely product to accomplish this although a high-end unit might use a faster, more expensive semiconductor.

The problem with MOVs is that they degrade each time they absorb a surge, that degradation exhibiting as a decrease in their trigger voltage (...). The only solution is to replace them. (...)

Thanks so much for that hint.

I just opened the UPS, and indeed I found a group of varistors approximately the size of a thumb nail. They have been marked with permanent marker pen color – hopefully following some check during assembly (this UPS manufacturer apparently assembles every UPS individually, according to the customer's wishes).

Technical error within the UPS wouldn't have been my first guess, since the UPS has, as far as I know, never encountered any serious challenges: As far as I know, it has never before been used under load, and only connected to the grid every now and then in order to be recharged, while the grid in that region is known to be stable with next to no variation in voltage (and frequency).

If you put an AC voltmeter on the outlet providing power to the UPS you will probably see that the GFCI trips when the voltage rises above a certain point. If that is not the case then I've misdiagnosed the problem.

I don't have a voltmeter, but the UPS can show the voltage in the display (don't know the rate at which that information is renewed however). Specs say that the UPS can work with anything from -20 % to +13 % of standard input voltage (= grid standard voltage). Actual input voltage (as shown by the UPS's display) always differed less than 3 % from standard. Don't know if the UPS can measure any "micro-spikes", however.

I saw that APC has some information on GFCIs and UPSs on their website. They say:

You can try to replace the GFCI/AFCI device and see if this solves the problem. Sometimes as GFI/AFCI devices age they will drift out of tolerance. You may also find that some GFI/AFCI devices are more sensitive than others. (...)

If the circuit is not required to be GFI/AFCI by code, you may consider changing to a non-GFI/AFCI receptacle or breaker.
Typically GFIprotection is only required by code in wet/damp locations such as Kitchens, Baths, or exterior receptacles (...).

(...)

There are thousands of successful installations of UPS and computer equipment with GFI/AFCI, however every application and load is different.
For best performance, use your UPS or Surge Protector with a standard non-GFI/AFCI power source.

(https://www.apc.com/lb/en/faqs/FA369034/)

Sounds as if the problem wasn't too uncommon?

They're cheap but you need to be able to solder.

I'm not really a soldering artist, so I somewhat refrain from trying that as my first option.


As a result of all the information above, I would tend to try the following steps (in that order):

1. Try a GFCI of a different brand.
2. Try a "medium sensitivity (MS)" (https://en.wikipedia.org/wiki/Residual- ... ensitivity) 300 mA GFCI instead of the currently used high sensitivity (HS) 30 mA GFCI.
3. Try an Arc-Fault Circuit Interrupter (AFCI, cf. https://en.wikipedia.org/wiki/Arc-fault ... nterrupter) instead of GFCI.
4. Try to exchange the varistors as advised.

Does that make sense?


Thanks so much for helping me to understand this problem. I'll keep you informed.

Greetings, Joe

[timkb4cq]

Is the second GFCI in the electrical panel or is it a GFCI outlet? If it's an outlet then trying another brand is a reasonable option. You may have trouble finding another brand that fits your panel.

One thing I neglected to ask is whether you have checked your grounding rod(s). Electrical panels have to connect to a ground wire separate from the incoming 3 phase AC wiring. That wire typically runs outside and attaches mechanically to a rod pounded into the ground. If that mechanical connection has corroded but not actually broken off (the UPS would show a ground error if it had) then you can get ground current fluctuations sufficient to trip a GFCI. People seldom think of checking it.

[MX-16_fan]

@timkb4cq:

Is the second GFCI in the electrical panel or is it a GFCI outlet? If it's an outlet then trying another brand is a reasonable option. You may have trouble finding another brand that fits your panel.

It's in the fusebox, mounted on a standardized rail. I have indeed found another manufacturer whose products would fit on that rail.

One thing I neglected to ask is whether you have checked your grounding rod(s). Electrical panels have to connect to a ground wire separate from the incoming 3 phase AC wiring. That wire typically runs outside and attaches mechanically to a rod pounded into the ground. If that mechanical connection has corroded but not actually broken off (the UPS would show a ground error if it had) then you can get ground current fluctuations sufficient to trip a GFCI. People seldom think of checking it.

Yes, you're right, people all too often forget about that. However, in this case, the grounding rod is brand new, fully functional, and so is the wire leading to it (hadn't been all too good before), and the connection between the rod and the wire.

Grounding rod has been mounted into the middle of a small natural pond, so there's 24/7/365 conductivity.

I guess GFCI #1 would regularly trip, too, if there was an issue with the grounding itself (it's connected to the same grounding rod, via a common earth busbar). Correct?


Greetings, Joe

[Bartman]

Leakage between the MOV is why the GFCI trips.