Ground fault detection

[Shiv Kumar]

Hello all,

I have been reading about ground fault detection requirements based on NEC codes and UL standards. My question is specific to ground fault detection in PV inverters with transformers.

As a reference, I found this excellent link on the requirements of grounding in PV:
http://solarprofessional.com/articles/design-installation/ungrounded-pv-power-systems-in-the-nec?v=disable_pagination

The PV inverters we are building is in the range of a 1000W. Since, the fault current in the case of a ground fault on the PV terminals could be limited depending on PV output, we have thinking of using ungrounded PV system i.e neither of the PV terminals connected to the earth ground and using an impedance detection method to detect ground faults.

But from the link above, I found that most of the ungrounded PV systems are in the case where the inverters do not have transformers. Our inverter however has a transformer. In the case of inverters with transformers, is there a requirement that each electrically isolated subsystem must have a terminal connected to earth-ground? If we do design an impedance measurement technique to detect ground faults while having the PV terminals completely ungrounded and comply with other requirements like disconnecting both terminals of the PV and having overcurrent devices on both terminals would it be a violation to have an ungrounded PV module with a transformer based inverter?

Thank you in advance.

SK

[Rob Beckers]

Hi Shiv,

I'm not aware of any requirement that grounds the negative or positive of a transformer-based inverter, but then again, I don't know UL1741 THAT well. One thing to consider though is that the NEC rules for ungrounded PV systems are much more onerous vs. grounded systems. Over here in Canada (which basically uses the NEC, though rules tend to trickle through a few years after they appear in the NEC) we have to provide a means to disconnect for both conductors within 3 feet of the array, and use armored cable or conduit for all conductors (so no easy install with PVC conduit and cheap wire).

As you noted, most ungrounded inverters are that way because they do not have a transformer, they don't have a choice. If you have a transformer I'd stick with negative grounded, it's a sales argument (easier install vs. ungrounded).

-RoB-

[Shiv Kumar]

Hi Rob,

Thanks for your response. A correction about my post - the NEC only says that ground fault must be detected, signalled and interrupted. The most common way to do this in an isolated system is to ground the negative terminal.

I read an article about a "blind spot" in ground fault detection:
http://www.solarabcs.org/about/publications/reports/blindspot/pdfs/inverter_groundfault-2013.pdf

The above article talks about how in isolated PV systems where the negative terminal is grounded, there is a possibility that a fault on the negative terminal will go undetected as the ground fault is not significant (due to low loop voltage) to trip the GFDI fuse. When there is now another fault on the positive terminal, there is now a direct path between positive and negative terminal which causes a fire.

Have you ever encountered a ground fault that was not detected by the protection by you could find out by visual inspection or other measurement?

Thanks and regards,
Shiv

[Rob Beckers]

Shiv, we don't look for ground faults unless the inverter trips and indicates a ground fault. So, no, I've never seen the case where the inverter doesn't flag the fault.

-RoB-