690.43 EGC Requirements

[BarklieEstes]

This question is based on the 2017 NEC and is similar this previous post.

Is there a requirement to run an EGC from a solar array back to the main service? The racking system we use is UL 2703 listed to bond all the metal parts together, we drive (2) of the supporting pipes 8' deep to meet the requirement of 250.52(5)(a), and the inverter back at the house is ungrounded.

It seems like this method adheres to both the spirit (effective ground fault path) and the letter of the NEC since it bonds the modules per 690.43(A), bonds the racking per 690.43(B), leaves with the circuit conductors per 690.43(C) (albeit gets to where it's going much faster), and complies with 250.134 per 250.134(B) Exception #2. Any other requirements I might be missing?

Regards,
Barklie Estes

 

[jaggedben]

This question is based on the 2017 NEC and is similar this previous post.

Is there a requirement to run an EGC from a solar array back to the main service?

Yes.

The racking system we use is UL 2703 listed to bond all the metal parts together, we drive (2) of the supporting pipes 8' deep to meet the requirement of 250.52(5)(a), and the inverter back at the house is ungrounded. service?

Equipment bonding is always required even when there is no system grounding.

It seems like this method adheres to both the spirit (effective ground fault path) and the letter of the NEC since it bonds the modules per 690.43(A), bonds the racking per 690.43(B), leaves with the circuit conductors per 690.43(C) (albeit gets to where it's going much faster), and complies with 250.134 per 250.134(B) Exception #2. Any other requirements I might be missing?

You haven't mentioned a wiring method between the array and house that complies with 250.134 that provides an effective ground fault current path. The supports in the ground may serve as grounding electrodes for the array but do not provide the effective ground fault path, see the last sentence of 250.4(5). It sounds like you are using a string inverter. The consequence of not having an EGC between array and inverter is that the inverter won't detect the first fault from the PV circuit to the racking. And when, months or years later, a second fault happens on the other conductor, the short circuit will catch the array on fire.

 

[JoeStillman]

I don't see a rule that says you have to run directly from the array to the service. But you have to get there alongside all your other conductors.

It is very important to NOT have two grounding electrode systems. The earth is not a good enough conductor to consider two electrodes connected to each other without an equipment grounding conductor running between them. You may have a "better" (i.e. lower resistance) connection to earth than the utility service with your 8' deep stanchions, but they still need an EGC connection to the grounding electrode system of the rest of the building. Otherwise a lightning strike on or nearby the building or the PV array can cause flashover between metal parts bonded to two different GEC's.

 

[JoeStillman]

Your grounded circuit conductor is allowed to be bonded to an EGC that is bonded to the building GEC. It doesn't have to be run all the way back to the service, if that was what you meant.

 

[jaggedben]

It is very important to NOT have two grounding electrode systems. ..

Actually, a ground mounted array is a separate structure required to have its own grounding electrode system. That system must be connected to the inverter and (eventually) the grounding electrode system 'back at the house.'

 

[JoeStillman]

Actually, a ground mounted array is a separate structure required to have its own grounding electrode system. That system must be connected to the inverter and (eventually) the grounding electrode system 'back at the house.'

You're right. I should have said two separate grounding electrode systems. If you do it right and (eventually) get to the GES back at the house, you've got one grounding electrode system with two electrodes.

 

[jaggedben]

You're right. I should have said two separate grounding electrode systems. If you do it right and (eventually) get to the GES back at the house, you've got one grounding electrode system with two electrodes.

No technically it's two grounding electrode systems, connected by EGC(s).

 

[BarklieEstes]

You haven't mentioned a wiring method between the array and house that complies with 250.134 that provides an effective ground fault current path. The supports in the ground may serve as grounding electrodes for the array but do not provide the effective ground fault path, see the last sentence of 250.4(5). It sounds like you are using a string inverter. The consequence of not having an EGC between array and inverter is that the inverter won't detect the first fault from the PV circuit to the racking. And when, months or years later, a second fault happens on the other conductor, the short circuit will catch the array on fire.

Ungrounded inverters don't utilize an EGC to detect ground faults.

 

[JoeStillman]

No technically it's two grounding electrode systems, connected by EGC(s).

I see your point, but I find it helpful to always remember Article 250.50:

250.50 Grounding Electrode System. All grounding electrodes
as described in 250.52(A)(1) through (A)(7) that are present at
each building or structure served shall be bonded together to form
the grounding electrode system.​

And I consider the stanchions electrodes per 250.52(A)(5):

(5) Rod and Pipe Electrodes. Rod and pipe electrodes shall
not be less than 2.44 m (8 ft) in length and shall consist of the
following materials.
(a) Grounding electrodes of pipe or conduit shall not be
smaller than metric designator 21 (trade size 3⁄4) and, where of
steel, shall have the outer surface galvanized or otherwise metalcoated
for corrosion protection.
(b) Rod-type grounding electrodes of stainless steel and
copper or zinc coated steel shall be at least 15.87 mm (5⁄8 in.) in
diameter, unless listed.

The goal is to ensure that all the metal everywhere is bonded together. Grounding is often mistaken for bonding, but bonding of all the electrodes is what prevents damage from lightning.

 

[wwhitney]

I see your point, but I find it helpful to always remember Article 250.50:

But the emphasis belongs on a different part of that sentence: "at each building or structure." So as the ground mount array and the house are two different structures, then there are two different Grounding Electrode Systems. Which happen to be bonded together by the EGC run with the DC conductors from the array to the inverter at the house.

Not that this distinction is particularly meaningful, as far as I'm aware.

Cheers, Wayne

 

[jaggedben]

...

Not that this distinction is particularly meaningful, as far as I'm aware.

Cheers, Wayne

It's meaningful in as much as it clarifies the following.
- the size of that wire is determined by 250.122
- the two structures GESs only need to be connected by that EGC, not any other bonding jumper

 

[JoeStillman]

It's meaningful in as much as it clarifies the following.
- the size of that wire is determined by 250.122
- the two structures GESs only need to be connected by that EGC, not any other bonding jumper

But you will have a bonding jumper if you have a grounded circuit conductor.

 

[jaggedben]

But you will have a bonding jumper if you have a grounded circuit conductor.

First, he stated the inverter was ungrounded meaning it doesn't have a grounded circuit conductor.

Second, no, typically you won't. See last sentence of 250.32(B)(1).

 

[JoeStillman]

First, he stated the inverter was ungrounded meaning it doesn't have a grounded circuit conductor.

Second, no, typically you won't. See last sentence of 250.32(B)(1).

Sorry, I was getting off topic trying to give an example of when you would have a bonding jumper. Curiously, the connection between a grounded conductor and the equipment grounding system is called a "bonding jumper" for separately derived systems (250.30) but not for photovoltaic systems (690.47(A). ) The informational note introduces a new term for me: "Functional grounded". Any (off topic) grounded conductor is connected to the equipment grounding conductor with grounding electrode conductor and not a system bonding jumper.

No wonder grounding and bonding are so hard to grasp.

 

[jaggedben]

Sorry, I was getting off topic trying to give an example of when you would have a bonding jumper. Curiously, the connection between a grounded conductor and the equipment grounding system is called a "bonding jumper" for separately derived systems (250.30) but not for photovoltaic systems (690.47(A). )

I disagree. While 690 doesn't actually say that you need a system bonding jumper for a solidly grounded system (it probably ought to say that) it also contains nothing that overrides the relevant parts of 250. 690.47 is talking about a grounding electrode conductor. Grounding and bonding are separate things.

Mind you, almost zero PV systems are solidly grounded. I've been working in PV for 15 years and I don't think I've ever worked on a solidly grounded system. A solidly grounded PV system basically cannot be interconnected to any other typical source of power. Most PV systems nowadays don't meet the definition of a separately derived system and therefore are not grounded separately from the AC system they connect to. Older ones that met the definition typically used functional grounding through a GFDI fuse as permitted/required by 690.

The informational note introduces a new term for me: "Functional grounded". Any (off topic) grounded conductor is connected to the equipment grounding conductor with grounding electrode conductor and not a system bonding jumper.

I disagree, see above. Also, a functionally grounded conductor is connected to ground (EGC and/or GEC) by the GFDI device. See 690.42(A).

No wonder grounding and bonding are so hard to grasp.

Well, you've pointed out that 690 is worse than the rest of the code on this. Thank goodness almost no PV systems need user installed system grounding.