Checking for Ground faults prior to Inverter connections

[jerryls]

One of my major concerns with installers of large PV roof top systems is that they make all the connections to the inverter on the DC side before doing any testing of ground faults.. So I always request on the drawings that they check that the equipment ground is not electrically connected to the negative side of the PV array prior to inverter connectios. This connection is done inside the inverter usually through the GFI circuit. However, if there is another electrical connection between the two outside the inverter, the GFI circuit becomes useless. Everything will work fine, but if there is a short in the hot side somewhere in the future, as in the case of the Bakersfield project, it can be a dangerous situation. I was wondering if anyone agrees with this logic. I think it should be a standard procedure and maybe even mentioned in section 690 of the code.

 

[SolarPro]

Yes, 690.42 states that if the system ground connection is made via a 690.5 ground-fault protection device, which is almost always the case, then this bond "shall not be duplicated with an external connection."

You write: "So I always request on the drawings that they check that the equipment ground is not electrically connected to the negative side of the PV array prior to inverter connections. This connection is done inside the inverter usually through the GFI circuit. However, if there is another electrical connection between the two outside the inverter, the GFI circuit becomes useless. Everything will work fine, but if there is a short in the hot side somewhere in the future, as in the case of the Bakersfield project, it can be a dangerous situation."

A couple things. First, not all PV arrays are negatively grounded—or even grounded at all. Rather than refer to negative or positive poles, you want to refer to grounded or or ungrounded current carrying conductors. That is technically correct regardless of the system grounding convention. Second, the blind spot does not occur any time there is a connection between a grounded conductor and ground outside the inverter. If a ground fault fault occurs in a grounded PV output circuit conductor, say between a string combiner and the inverter input combiner, then the 690.5 GFP device absolutely will trip. If the fault current exceeds the fuse rating—typically 5 A—then the GFP system will work as intended. "Blind spot" refers specifically to a fault between a grounded PV source circuit conductor and ground; in this case, if the fault current in EGC is below the trip threshold of the GFP fuse, then the fault will go undetected.

Note that a proper system commissioning process will include insulation resistance testing for the PV source and output circuits. This is what protects against an undetected first fault in a source circuit at start up. You might want to call this out in the notes on your plan sets if you are concerned about a hidden fault between a grounded PV source circuit conductor and ground.

You can also review these SolarABCs resources RE: the ground-fault detection blind spot:

http://www.solarabcs.org/about/publications/reports/blindspot/index.html

http://www.solarabcs.org/about/publications/meeting_presentations_minutes/2013/06/webinar-0625.html

Depending on the inverter and the design of its GFP system, you may be able to further reduce the chances of having an undetected ground fault on a grounded source circuit conductor by replacing the 5 A GFP fuse with a 1 A GFP fuse. Other approaches are listed, as well. Ultimately, this will be addressed by the inverter equipment standard.

Last but not least, you can design ungrounded PV power systems, per NEC 690.35, and eliminate this potential problem altogether:

http://solarprofessional.com/articles/design-installation/ungrounded-pv-power-systems-in-the-nec

 

[jaggedben]

In a previous thread I explained the way that I do source circuit testing at commissioning time, which is the way I train the rest of our employees to do it. If done for every source circuit, it is a fairly comprehensive means of catching faults across all inverter and combiner topologies. The method tests for faults on both sides of all source circuits.

http://forums.mikeholt.com/showthread.php?t=142928&p=1371399#post1371399

Note that if the inverter or combiner has the typical common grounded bus, then it isn't really possible to safely test for ground faults on both sides of the circuit, or at least not to isolate them, unless you install whips. You can also use combiners that come with pre-wired connectors, I think that should be a feature of all combiners.