Grounding a PV system to a Sub Panel in a detached structure

[PVfarmer]

8awg EGC is fine for a 7700TL. In fact, 10awg is okay.

I'm not trying to argue, but isn't following the manual part of both the inverter warranty and a code rule?
I mean, you can't go ahead and use #1 or 1/0 for L1, L2 or N.

Couldn't an AHJ somewhere potentially say "hey, you didn't use #12 wire there, you can't do that"?

 

[JAZ140]

Thanks again guys! I like that 2014 helps clarify non isolated grounding requirements. I think for now I will ditch the GEC and Bond my A/C EMT to the #8 EGC and call it good. Then just have crimper in my back pocket when the POCO rolls in if they want to see the EGC crimps in the main and sub.

I wonder if they will ever put a spacing requirement on detached structure grounding. IE under 500 feet will not require a separate GE. I still think it is safer to have one earth connection vs two in a residential application when the separation of structures is so close. but that's another topic.

 

[jaggedben]

Isn't the POCO going to say instead of the GEC that they want the EGC to go straight from the inverter to house panel?

There's no reason for them to say that. 705.12 says that systems can be connected at any distribution equipment on the premises.

Shouldn't the N from the inverter also go straight to the house?

No. The neutral and EGC must terminate in the same location as the breaker.

Is there maybe a inverter voltage drop issue with the inverter N and G both connected to the subpanel 150 away from the main?

Not likely with the #3 Cu feeder. Seems like that's already oversized for VD. [EDIT: Looks like it was originally sized to 125A, they downgraded the breaker to 100A for the 120% rule. Anyway, still not an issue...]

 

[jaggedben]

I'm not trying to argue, but isn't following the manual part of both the inverter warranty and a code rule?
I mean, you can't go ahead and use #1 or 1/0 for L1, L2 or N.

Couldn't an AHJ somewhere potentially say "hey, you didn't use #12 wire there, you can't do that"?

You do raise a valid point here, but then again a #12 EGC is not acceptable under the NEC for the 40A breaker that a 7700TL requires. So that's a code Catch-22.

As far as AHJs go, any electrical inspector is far more likely to know the required EGC size for a 40A breaker than to know what it says on pg 36 of the SMA manual.

Having personally landed #10 and #8 solidly on the SMA ground lugs with no difficulty, and being 110% confident that the connection is as good as it would be with a #12, I'm going to go with this: SMA made a typo in the manual. They put the minimum size and forget to include the maximum. Just an experienced guess.

 

[PVfarmer]

1 There's no reason for them to say that. 705.12 says that systems can be connected at any distribution equipment on the premises.

2 No. The neutral and EGC must terminate in the same location as the breaker.

3 Not likely with the #3 Cu feeder.

4 SMA made a typo in the manual.

Thanks for your help.

1 Isn't it possible that the POCO has their own rules which aren't in the NEC? Or isn't it... a fact?

2 Ok, so... the PV breaker is in the subpanel in his drawing- however, neither the N or EGC seem to terminate there. Do I have the definition of "terminate" wrong? This #2 seems to contradict your #1 somehow.

3 I mean voltage drop as in if the POCO is supplying 120V at the main panel, and then there's 150 ft to the sub panel, the voltage will be lower at the subpanel, and the inverter will use that lower V for its' voltage reference, seeing as the inverter N and G are connected to both the N and G bars in the subpanel...do I have that completely wrong, or is it so small as to not be a factor?

To put 3 a different way- how is the inverter going to know what the actual grid voltage at the main panel is if it's feeding the subpanel loads 150 feet away before going to the main panel?

4 I'll buy that. Seems like one size only is wrong for sure, but there should be a maximum, or at least it should say 12AWG *minimum*.

 

[SolarPro]

I'm going to go with this: SMA made a typo in the manual. They put the minimum size and forget to include the maximum. Just an experienced guess.

That's better than when the OEM lists their torque values in foot-pounds when they really mean inch-pounds. :slaphead:

 

[jaggedben]

1 Isn't it possible that the POCO has their own rules which aren't in the NEC? Or isn't it... a fact?

Yes, but generally the POCOs rules can't really contradict the NEC or dictate all the details of an install. Usually the POCO can only enforce rules that are backed up by the legal regulations that govern interconnection agreements. For example, they can require the equipment to be on a certain list, or UL listed, and they can require a disconnect. Sometimes utilities get away with stuff they are not legally allowed to do just because no one has the time or resources to sue them. For example, POCOs that don't allow line side taps.

Nothing in the example above is anything the POCO would have any legal right to make demands about where I work. You get a signed off building permit from the AHJ, the POCO can't dispute that part, and doesn't. It might be slightly different in the OP's state.

2 Ok, so... the PV breaker is in the subpanel in his drawing- however, neither the N or EGC seem to terminate there. Do I have the definition of "terminate" wrong? This #2 seems to contradict your #1 somehow.

You indeed have the definition wrong. The N and G for the solar circuit are distinct from those for the feeder circuit (different sizes, different overcurrent device) and they do indeed terminate in the sub in the diagram.

3 I mean voltage drop as in if the POCO is supplying 120V at the main panel, and then there's 150 ft to the sub panel, the voltage will be lower at the subpanel, and the inverter will use that lower V for its' voltage reference, seeing as the inverter N and G are connected to both the N and G bars in the subpanel...do I have that completely wrong, or is it so small as to not be a factor?

To put it a different way- how is the inverter going to know what the actual grid voltage at the main panel is if it's feeding the subpanel loads 150 feet away before going to the main panel?

Of course I know what you mean.
Yes, the difference is so small as to not really be a factor in this case.
Consider that the 40A output from the inverter only goes a short distance to the sub with #8 wire. Then the rest of the much longer distance to the house is #3 wire. So in effect the output circuit is already upsized for voltage drop. If the entire 150ft run to the house was done with #8 wire then indeed there might be an issue. But that would still probably only call for upsizing to #6, not #3. 150ft is not enough distance to cause major voltage drop issues.

FYI, when the sub is drawing power from the house, the voltage drops from the house to the sub. When the inverter is exporting power from the sub, the voltage rises from the house to the sub, because energy is flowing in the opposite direction. In this case you have to consider voltage drop for 100A flowing to the sub from the house, but only 40A for voltage rise in the opposite direction. Less current equals less voltage drop/rise. This is typical for solar installs and usually ensures that voltage rise from interconnecting to a sub is not an issue.

The inverter does not need to know what the actual grid voltage at the main panel is as long as the voltage rise/drop is not so severe as to fall outside the proscribed window for operation.

Also, it's not the POCOs responsibility to make sure that this part is done right.

 

[PVfarmer]

1 You indeed have the definition wrong. The N and G for the solar circuit are distinct from those for the feeder circuit (different sizes, different overcurrent device) and they do indeed terminate in the sub in the diagram.

2 Yes, the difference is so small as to not really be a factor in this case.

3 FYI, when the sub is drawing power from the house, the voltage drops from the house to the sub. When the inverter is exporting power from the sub, the voltage rises from the house to the sub, because energy is flowing in the opposite direction.

4 The inverter does not need to know what the actual grid voltage at the main panel is as long as the voltage rise/drop is not so severe as to fall outside the proscribed window for operation.

Thanks again.

1 So it means physical termination and not electrical? But I'm still a bit confused- there aren't any overcurrent devices for the N and G at the subpanel.

2 Ok.

3 That's very interesting about the rise- what about when 100% of the PV power is going to the subpanel loads, plus some grid power from the main panel to subpanel loads? Say a big motor or pump kicks in- that wouldn't take the inverter out of its' "window"?

4 I hear you. I'm just thinking that the farther the PV is from the grid xfmr, the more issues you'd have. So if 150 feet is ok, what would you say would be the limit where it would become a problem? 500 feet? 1500?

 

[JAZ140]

A/C range on SMA TL-22 Series is 211-264 when operating at 240 volts. very wide range.

 

[GoldDigger]

1. ...And there is normally never any overcurrent device in the grounded conductor and absolutely never in the EGC.

 

[jaggedben]

Thanks again.

1 So it means physical termination and not electrical? But I'm still a bit confused- there aren't any overcurrent devices for the N and G at the subpanel.

I don't know why we indulge you on this forum when you don't know this really basic stuff. :roll:

A branch circuit is defined by the overcurrent device, in this case the 40A 2-p solar breaker.
Anything upstream of that is considered a different circuit (i.e. the sub feeder in this case).
Overcurrent devices are not allowed in grounded conductors (neutral, in this case), or in EGCs.

3 That's very interesting about the rise- what about when 100% of the PV power is going to the subpanel loads, plus some grid power from the main panel to subpanel loads? Say a big motor or pump kicks in- that wouldn't take the inverter out of its' "window"?

Probably not. Most engineers design feeders for a 2% maximum voltage drop. The inverter can handle plus 10% or minus 12%, so a much bigger window.
Note that by reducing the current that's fed from the grid, the inverter at max output reduces the voltage drop on the sub feeder. So your scenario is not worst case. Worst cases would be:
a) a constant full load on the subfeeder causes voltage to drop below the inverters window in the morning, before the inverter turns on. Inverter never sees a good voltage and never turns on.
b) zero load at the subpanel, full PV output, too much voltage rise from service to inverter.

Any load that turns on after the inverter turns on mitigates the situation.

4 I hear you. I'm just thinking that the farther the PV is from the grid xfmr, the more issues you'd have. So if 150 feet is ok, what would you say would be the limit where it would become a problem? 500 feet? 1500?

The utility has an obligation to maintain the voltage within the window at the service point, so it really doesn't matter if their transformer is on the moon, they are supposed to have engineered their service conductors to handle the drop from there to the service point. So really your obligation is to calculate drop/rise from the service point, and make sure it's about 2% or less.

If your inverter output circuit is 150ft to the service point you should probably do a calculation to check your voltage rise, but I doubt I've ever seen 150ft call for more than one wire size larger, at least in residential work. Thus, as explained above, if most of that output circuit is a sub feeder that is already larger, you pretty much don't need to bother.

FWIW, we always roll with minimum #10 wire for inverter output, even if it's 20A or less. With rooftop temperature adders required, this pretty much means we don't ever have to worry about voltage drop for micro-inverters on the roof.

 

[PVfarmer]

1 I don't know why we indulge you on this forum when you don't know this really basic stuff. :roll:

2 Overcurrent devices are not allowed in grounded conductors (neutral, in this case), or in EGCs.

3 Probably not. Most engineers design feeders for a 2% maximum voltage drop. The inverter can handle plus 10% or minus 12%, so a much bigger window.
Note that by reducing the current that's fed from the grid, the inverter at max output reduces the voltage drop on the sub feeder. So your scenario is not worst case. Worst cases would be:
a) a constant full load on the subfeeder causes voltage to drop below the inverters window in the morning, before the inverter turns on. Inverter never sees a good voltage and never turns on.
b) zero load at the subpanel, full PV output, too much voltage rise from service to inverter.
Any load that turns on after the inverter turns on mitigates the situation.
So really your obligation is to calculate drop/rise from the service point, and make sure it's about 2% or less.

4 FWIW, we always roll with minimum #10 wire for inverter output, even if it's 20A or less.

1 It's gotta be either my outside of the box thinking, or my ability to translate zman's questions. Or possibly that I do make good points every now and then (quoting you )

2 Right. I'm still a little confused by the JAZ's diagram and the EGC/GEC issue. You told him his 1st diagram was cool, and then he deleted the GEC, so...?
In both diagrams he posted, the N and G and bonded at the main panel and not bonded at the secondary panel.
Looking at the second one, the N from the grid is connected to the N bar in the main panel at the same spot as the N wire which goes to the N bar in the secondary panel and the N from the inverter is connected to the opposite end of the sec. panel N bar.
That N line from the grid is also connected to the G bar in the main panel. (Bonded, right?)
So...why is the N from the inverter connected to the secondary panel at all?
If the N from the inverter ran to the N bar of the main panel, wouldn't that satisfy what the POCO is asking for, because the EGC for the inverter would then "terminate" at the main panel?
Please don't get aggravated- just questions here.
I'm not saying JAZ should do this- just hypothetical- why not "terminate" the inverter EGC at the switch in JAZ's drawing?
This is the closest example I can find- yes, I know it's off-grid, but...
http://solar.smps.us/solar-system-diagram-offgrid.jpg

If the G from the inverter went to actual earth at the PV disconnect switch, then the N could pass through the switch, and N and G from PV Inverter bonded at the main panel?

3 Ok- so...a 120V of grid is supplying the loads in the AM. 2% drop is 117.V 12% drop from 120V is 105.6V, JAZ pointed out that: "A/C range on SMA TL-22 Series is 211-264 when operating at 240 volts. very wide range. "
211V at 240V is 105.5V at 120V.
So no problem with your a) scenario, but what about what I mentioned about a bunch of motors turning on while the PV is running at max output and the voltage dropping below 105.6V for a short time, therefore kicking the PV off?

4 Back to that SMA typo- assuming the 12-6AWG for L1, L2 and N outputs thing is NOT a typo- I'm correct that you can't use 4AWG for the L1 L2 and N, because of the physical structure of the inverter output (simply, they aren't lugs?)

A/C range on SMA TL-22 Series is 211-264 when operating at 240 volts. very wide range.

Hey, great drawing btw if I didn't say that already. One thing-isn't the N wire going from the secondary panel to the main panel supposed to "jump" (as in not be connected to) the 2 100A breaker wires?
If you don't mind me asking, did yo use SmartDraw software to do the drawing?

1. ...And there is normally never any overcurrent device in the grounded conductor and absolutely never in the EGC.

Well..ok...but if the PV inverter's N and G are terminating at the secondary panel (where the 40A breaker is)...that doesn't make it a separate EGC system from the main panel EGC?
Like Jben said, if the N and G are distinct...should they be grounded to earth in different places?

jben quote-
"The N and G for the solar circuit are distinct from those for the feeder circuit (different sizes, different overcurrent device) and they do indeed terminate in the sub in the diagram."

 

[GoldDigger]

OK, here are some basics to move things along toward understanding:
A grounded supply of electricity has one grounded conductor (often it is also a neutral, but that is not always true) and one or more ungrounded conductors (hots).
The EGC is a separate conductor (sometimes just the raceway or cable sheath).
The NEC requires that all of the conductors for a branch circuit (the wires after the last overcurrent device in the hot conductors) originate in the same panel.
That means that the physical neutral wire must be connected to some kind of termination at that panel (the neutral bar usually).
In the special case of a branch circuit AFCI or GFCI breaker, if a neutral is used in the circuit it must terminate at a terminal on that breaker and the breaker will extend it through a pigtail to the neutral bar. The OCPD does not interrupt the neutral when it opens. (A GFCI receptacle, OTOH, does interrupt the neutral connection too.)
You do not ever have the neutral "skip" the box where the branch breaker is located and go directly to the upstream panel or disconnect. The line drawing may make it look that way sometimes, but if is not correct.
The grounded connector is bonded to the EGC and GES at exactly one point in the customer side of the service point. Where POCO bonds things on their side is not relevant since customer electrons and POCO electrons have different rules to follow.
For fault current purposes (to allow LV OCPDs to operate properly) the earth connection of the EGC us not important. The connection to the POCO neutral is.

 

[PVfarmer]

a) a constant full load on the subfeeder causes voltage to drop below the inverters window in the morning, before the inverter turns on. Inverter never sees a good voltage and never turns on.

Thanks for hanging in here man. I'll keep this one short.
There are two things I'm not getting about the drawing here.
I don't know what JAZ's loads are- so this is hypothetical.

1. Why is there a green line/wire between the ground bar of the secondary panel and the ground bar of the main panel?
Hypothetically, with all the breakers open, the G wire leaving the inverter is connected to:
(this is all "in order")
1 the PV meter
2 the switch
3the ground bar of the sec. panel (and therefore the earth 4)
and then, ok, "terminated", but still electrically connected to then:
5 ground bar of the main panel (and therefore also the earth 6)
7 the main panel N bar
and then...the ground again at the service transformer (I think? not shown in diagram 8).

Which leads to the second thing-

JAZ's main breaker for the sec. panel is 100A...that's 100A at 120V, correct?
His main breaker in the main panel is 200A- that means 200A of 120V and 100A of 240V.
His PV is shown as 40A- that's shown as a 40A breaker for his 32A of PV...at 240V
But it's really L1 and L2 at 120V, so it's physically two 20A 120V breakers in one panel space.

What I want to know is, what happens when:
Say it's noon and the PV is at max output (32 amps at 240V).
The loads drawn from the secondary panel are cooking along at 50 amps of 240V (which is the same as 100 amps of 120V).

That means there's 32A from PV and 18A from the grid supplying the sec. panel loads. Fine.
Someone decides to to bit of welding or something in addition.
A 20 amp /240V welder, so 40 amps of 120V.

That puts the secondary panel loads over the 100A main breaker, not to mention over the 125A busbar rating.
So! What happens then?

Another 20A of 240V coming from the grid (for a total of 38A) isn't going to trip the 100A breaker.
Or...if it is going to trip the 100A breaker, what about all the other breakers (PV and loads) remaining closed?

Seems like the PV breaker would trip too, but in that case the N and G from the inverter are still connected, all the way to the main panel, and possibly the LV side service ground.
With all the load equipment breakers still closed.

Please...um...advise. Or yell at me because I'm being unclear- it's all good, man.

 

[GoldDigger]

First of all, do not keep converting back and forth between amps at 240 and amps at 120 unless you are putting in a transformer to do it.
A 100A panel for 120/240 single phase three wire will have two buses each capable of carrying 100A and the main breaker will have two poles, each capable of carrying 100A.
If you have a 50A 240V load you are still at only half the capacity if the panel.

Second, when you have PV supply and a local load in the same panel the current flow to/from the bus is in opposite directions for the two and cancels.
If I have 32A at 240V of PV going into the 100A bus and I have a 10A 240V load, there will be 22A outbound on the panel feeder.
If I then add in a 40A 240V welder, I will have 18A coming in on the feeder instead.
If I start with just the 32A if PV at one end of the bus and the feeder at the other end, I can put a 132A load on the middle of the bus (two circuuts at 66A each so I do not overload a single bus stab) and not overload the bus. I will have violated the 120% rule, but it will probably not be unsafe. There is no cross section of the bus carrying more than 100A.

 

[jaggedben]

PVfarmer, if you want to understand the basics of split-phase electricity, or grounding or bonding, there are a lot of other threads here to search through. Or you could start another one under the general electrical forum.

At least that way you wouldn't be garbling up discussions that someone else was hoping to get real answers out of, to save themselves money.

 

[PVfarmer]

OK, here are some basics to move things along toward understanding:
A 1 grounded supply of electricity has one grounded conductor (often it is also a neutral, but that is not always true) and one or more ungrounded conductors (hots).
The 2 EGC is a separate conductor (sometimes just the raceway or cable sheath).
The NEC requires that all of the conductors for a branch circuit (the wires after the last overcurrent device in the hot conductors) 3 originate in the same panel.
That means that the physical neutral wire must be connected to some kind of termination at that panel (the neutral bar usually).

In the special case of a branch circuit AFCI or GFCI breaker, if a neutral is used in the circuit it must terminate at a terminal on that breaker and the breaker will extend it through a pigtail to the neutral bar. The OCPD does not interrupt the neutral when it opens. (A GFCI receptacle, OTOH, does interrupt the neutral connection too.)

4 You do not ever have the neutral "skip" the box where the branch breaker is located and go directly to the upstream panel or disconnect. The line drawing may make it look that way sometimes, but if is not correct.
5 The grounded connector is bonded to the EGC and GES at exactly one point in the customer side of the service point. Where POCO bonds things on their side is not relevant since 6 customer electrons and POCO electrons have different rules to follow.

For fault current purposes (to allow LV OCPDs to operate properly) 7 the earth connection of the EGC us not important. The connection to the POCO neutral is.

Thanks a bunch, man.
As you can see by my previous comments, I'm still a bit confused!

1 In your comment, GES is = to the term GEC, correct? Or more accurately, a GES (Supply?) is made from GECs (Grounded Electrical Conductors?)

2 EGC is Equip. Grounding conductor, and connects the metal frames and boxes of said equip. to earth.

3 This is where I lose it. If the PV inverter is a branch circuit, how do the PV output wires in the diagram terminate and originate at the secondary panel? I don't get how the neutral bar in that panel can be a termination when it continues on from that secondary N bar to the main panel N bar (and main G bar also. Yes, I realize it's through psysically different wires, but they're all connected to the same busses)

4 If, by #5 here, it must be one point only on the customer side, it seems like the N is bonded in two places on the customer side.

5 The N from the PV inverter in the diagram being a GEC, it seems like it's bonded at the main panel and grounded at the secondary? Does that make any sense at all?

6 Aye, there's the rub! :ashamed:

7 So the one point where the PV grounded connector/neutral is bonded to the EGC and GES can be a different point from where the GEC/N connects to the POCO?
That seems weird to me, but I can't quite explain it. Seems like if the POCO N is important, the PV N should be as close to it as possible.


The grounded connector is bonded to the EGC and GES at exactly one point
the earth connection of the EGC us not important. The connection to the POCO neutral is.

 

[GoldDigger]

GES stands for Ground Electrode System. It refers to the complete set of electrodes, GECs and bonding jumpers that ends up connected to the main bond location (often at the main disconnect or main panel).
GEC stands for Ground Electrode Conductor, not Grounded. It connects a Ground Electrode to the bond point. And it cannot contain connectors or reversible splices.

In some cases (often in a PV installation) the same conductor may serve as both a GEC and an EGC. If this is done it must be sized to the larger of the two requirements.

 

[PVfarmer]

jben-
Honestly, I'm worried someone will get upset if I start a new thread- my ? is about this diagram.
Simply-
Where is this "one point" in JAZ's diagram?
If-
The grounded connector is bonded to the EGC and GES at exactly one point in the customer side of the service point.

GD-
Let me correct myself.

But it's really L1 and L2 at 120V, so it's physically two 20A 120V breakers in one panel space.

I meant two 40A 120V breakers- it takes L1 and L2 of 120V, BOTH at 32 amps, to make 32 amps of 240V.

But...
A 100A panel for 120/240 single phase three wire will have two buses each capable of carrying 100A and the main breaker will have two poles, each capable of carrying 100A.

If each pole/bus is 100A, isn't that a 200A panel, not a 100A panel?

I see the diagram as having a 100A secondary panel with two 50A busses (120V each) and two 50A poles, the two poles being in one space, so called 100A.
His main panel is two 100A busses and two 100A poles/circuits (120V) in one space, so called a 200A main panel.

To put it another way, if the loads are either 50A of 240V welder, or instead 100A of 120V sewing machines, that would be 12kVA of load for both.

Or- under where I'm standing at this very moment there's a 200 amp main panel- the transformer I can see out the window is (probably) 25kVA, or 208 amps at 120V- it's two lines of 120V, so L1 and L2 are 104 amps each.
When you put them together to get the 240V they're parallel- the voltage doubles, but the amps stay the same, so 104 amps at 240V comes from the 104A each on L1 and L2 at 120V.

Don't want to start an argument, but I'm pretty confident about this.

If you had a single phase source of 50 amps of 240V power, either 240V L-L or L-N,and you connected it to a 15kVA 240V to 120/240V xfmr, you'd have 100A of 120V on the 120/240V side- it would be L1 and L2 of 120V both being 50A, when you add them together using math you have 100A of 120V, but when you connect them electrically, it's still the same 50A of 240V on the 120/240V side, minus losses, etc.

 

[PVfarmer]

GES stands for Ground Electrode System. It refers to the complete set of electrodes, GECs and bonding jumpers that ends up connected to the main bond location (often at the main disconnect or main panel).
GEC stands for Ground Electrode Conductor, not Grounded. It connects a Ground Electrode to the bond point. And it cannot contain connectors or reversible splices.

In some cases (often in a PV installation) the same conductor may serve as both a GEC and an EGC. If this is done it must be sized to the larger of the two requirements.

Ok! Thanks.

There's a GEC coming out of the inverter, the neutral.
If, in a PV installation, the inverter neutral is acting as the GEC *and* the EGC, which it seems to be in the diagram...and it's supposed to be connected to the main bond location (isn't that what JAZ's POCO is asking for?)
Hold on- where'd the EGC for the inverter go?
The green line that JAZ deleted from his first drawing, which I thought was the ECG for the inverter- where'd that go? :huh: There's still a connection point (lug) there for up to "over 12AWG" wire...

It *was* connected to the ground bar of the sec. panel (and earth), but if he deleted it, the GEC (neutral) of the inverter is going to the N bar of the sec. panel, then to N bar of main panel, then to ground bar and earth at the main panel...
which is what the POCO wants from the OP!

"the POCO wants the GEC to go from the inverter to the house."

But...where should that EGC from the inverter lug be connected as of now? Didn't someone say that the inverter EGC was involved in the DC grounding somehow? If so, seems like that #8 EGC should also go to the main panel.

Of course I could be wrong.
Thanks again.