First point of disconnect, isolated vs bonded, separately derived vs not

[The_Archer]

Recently I have been getting conflicting information from solar professionals and inspectors: Here is a scenario 1: AC side: two solar inverters wired up to a load center, from the load center to a fused service rated disconnect. In the load center grounds and neutral are isolated ( no bond screw) in the svc rated disc the grounds and neutral are joined together and bonded, from the svc disc to to a tap in the main service panel, the EGC from PV array is then crimped to the GEC. Is there a scenario where you isolate grounds and neutrals in both the load center and the svc rated disconnect? also, in scenario 1, if the EGC from the array passes through the load center and AC svc disc and is split bolted to the two pieces of equipment, do you still need to carry a separate ground to the msp with you tap feeders? IE: 6/3 Romex from the line side of the svc disc to the tap, the ground with the Romex is redundant if the EGC is crimped to the GEC inside the MSP, what of the EGC is crimped outside the panel or landed on a water clamp to water main? The Soares book on grounding has a great illustration showing 3w vs 4w, separately derived vs not separately derived. This illustration confuses me a little, I guess the main question is what is the distinction with separately derived vs not separately derived with regards to grounds and neutrals, first point of disconnect, bonding etc.

 

[Carultch]

Recently I have been getting conflicting information from solar professionals and inspectors: Here is a scenario 1: AC side: two solar inverters wired up to a load center, from the load center to a fused service rated disconnect. In the load center grounds and neutral are isolated ( no bond screw) in the svc rated disc the grounds and neutral are joined together and bonded, from the svc disc to to a tap in the main service panel, the EGC from PV array is then crimped to the GEC. Is there a scenario where you isolate grounds and neutrals in both the load center and the svc rated disconnect? also, in scenario 1, if the EGC from the array passes through the load center and AC svc disc and is split bolted to the two pieces of equipment, do you still need to carry a separate ground to the msp with you tap feeders? IE: 6/3 Romex from the line side of the svc disc to the tap, the ground with the Romex is redundant if the EGC is crimped to the GEC inside the MSP, what of the EGC is crimped outside the panel or landed on a water clamp to water main? The Soares book on grounding has a great illustration showing 3w vs 4w, separately derived vs not separately derived. This illustration confuses me a little, I guess the main question is what is the distinction with separately derived vs not separately derived with regards to grounds and neutrals, first point of disconnect, bonding etc.

I've wondered this myself. This is in the gray area, where it is unclear whether a supply-side interconnection's "tap disconnect" is or is not a "service disconnect". And what it precisely implies in terms of how you are supposed to build it, in contrast from a generic feeder disconnect. 705.12(A) allows a supply-side interconnection, but doesn't say how you are supposed to do it. Is it a tap? Is it service conductors? Is it a service disconnect? Etc.

As I understand it, "separately derived system" implies that it is an independent transformer secondary, and therefore would need to have its neutral-to-ground established somewhere, in order to have neutral be at the same absolute voltage as ground. In this instance, where you have a fused disconnect on the line side of the main service disconnect, nothing is "separately derived" because they are both part of the same voltage system. As long as neutral and ground of the PV system are eventually connected to neutral and ground of the service, you have the voltage reference you need, and I believe that a neutral-to-ground bond would be redundant. I always keep neutral and ground isolated, unless otherwise required by the NEC or AHJ.

 

[jaggedben]

I agree it's a gray area, but only in so far as the code doesn't make it crystal clear that the disconnect is a service disconnecting means. I'm of the opinion that it is a service disconnecting means, but we've had some long, inconclusive arguments on that point in this forum. Published articles take different stances. However, once you make a decision on that question, the rules are relatively clear...

a) If it is a service disconnecting means, then you may bring the neutral from the tap to the disconnect and bond the neutral to the enclosure. No EGC is required from the tap (although depending on code interpretation and installation details, you may have a GEC going back through the same raceway). The EGC for anything toward the inverter side starts from the same point where the neutral is bonded, just like any other service disconnecting means.

b) If it is not a service disconnecting means, then bring an EGC from the location of the existing neutral-to-ground bond, and bond the disconnect to it. Do not bond neutral to ground at the disconnect.

Regardless, neutral and ground are not bonded again anywhere on the inverter side of the disconnect.

As Carultch said, we are not dealing with a separately derived system if we are only looking at the AC side of the inverter. And even if the DC side would meet the definition of an SDS (and most new inverters don't), 690.47 has its own rules for PV system grounding that would supersede anything in article 250.

 

[Carultch]

I agree it's a gray area, but only in so far as the code doesn't make it crystal clear that the disconnect is a service disconnecting means. I'm of the opinion that it is a service disconnecting means, but we've had some long, inconclusive arguments on that point in this forum. Published articles take different stances. However, once you make a decision on that question, the rules are relatively clear...

I generally consider it a service disconnect, if there is a standalone utility service that is established for the PV system, therefore with its own utility account and its own utility meter independent of the building service.

Whereas if it is a behind-the-meter system that just happens to be on the line side of the service disconnect (only possible with hot sequence metering), I consider it "not a service disconnect", and avoid bonding ground and neutral thusly. I still try to keep it within the "six handle rule", and I still provide a pair/set of labels enumerating all the "handles" that you need to throw to shut off the system.

 

[ggunn]

I generally consider it a service disconnect, if there is a standalone utility service that is established for the PV system, therefore with its own utility account and its own utility meter independent of the building service.

Whereas if it is a behind-the-meter system that just happens to be on the line side of the service disconnect (only possible with hot sequence metering), I consider it "not a service disconnect", and avoid bonding ground and neutral thusly. I still try to keep it within the "six handle rule", and I still provide a pair/set of labels enumerating all the "handles" that you need to throw to shut off the system.

Even among AHJ's there is disagreement. CPS (San Antonio) considers a line side connection to be a service entrance, so in that jurisdiction we bond neutral to ground in the disco and break the EGC at that point. Austin Energy and ONCOR, OTOH, do not consider it a service, so in those jurisdictions we leave ground and neutral unbonded and carry the EGC back to the service.

 

[jaggedben]

Probably the best course of action is to ask the AHJ "do you consider this a service disconnect?". If you try to make the decision for them you may never win.

On small systems I have frequently run both a neutral and a ground back to the tap and bonded the disco neutral to the enclosure. That way if the AHJ doesn't want something it's easier to remove whatever that is (especially if 'it' is that green bonding screw that has long since disappeared! :happyyes. That's a pretty good strategy up to a 60A disco; above that the stakes start to get much higher.

 

[Carultch]

Probably the best course of action is to ask the AHJ "do you consider this a service disconnect?". If you try to make the decision for them you may never win.

On small systems I have frequently run both a neutral and a ground back to the tap and bonded the disco neutral to the enclosure. That way if the AHJ doesn't want something it's easier to remove whatever that is (especially if 'it' is that green bonding screw that has long since disappeared! :happyyes. That's a pretty good strategy up to a 60A disco; above that the stakes start to get much higher.

There's always the bonding sharpie, and an ordinary screw. (Joking)
http://d15bv9e9f3al6i.cloudfront.ne...harpie-Permanent-Marker-Pen-Fine-Green_P2.jpg

 

[jaggedben]

A long time ago I met an electrician who gave me a 10 minute lecture on green nail polish.

 

[ggunn]

Probably the best course of action is to ask the AHJ "do you consider this a service disconnect?". If you try to make the decision for them you may never win.

On small systems I have frequently run both a neutral and a ground back to the tap and bonded the disco neutral to the enclosure. That way if the AHJ doesn't want something it's easier to remove whatever that is (especially if 'it' is that green bonding screw that has long since disappeared! :happyyes. That's a pretty good strategy up to a 60A disco; above that the stakes start to get much higher.

But isn't that a violation either way they look at it? Either you bond the neutral and break the EGC, or you leave neutral and the EGC separate and run the EGC back to the service.

 

[jaggedben]

But isn't that a violation either way they look at it? Either you bond the neutral and break the EGC, or you leave neutral and the EGC separate and run the EGC back to the service.

I'm not sure it's a violation to have an EGC even though it isn't required, especially if I call it a GEC instead of an EGC. What code section would it violate? That is, as long as there's no current on the neutral; but this is a solar system.

 

[Carultch]

I'm not sure it's a violation to have an EGC even though it isn't required, especially if I call it a GEC instead of an EGC. What code section would it violate? That is, as long as there's no current on the neutral; but this is a solar system.

If there is current on the neutral, it will be shared between the neutral and EGC, if you bond the neutral and ground at multiple points. In the event that the EGC is significantly smaller than the neutral, and there is significant neutral current, then there is the possibility that the EGC carries more current than is permissible.


The EGC and system of non-current-carrying conductive materials should not be carrying current, except during fault conditions. This is why there should always be as few EGC-to-Neutral bonds as possible, except where otherwise required. So this is not a situation where you should just overbuild out of ignorance. This is a situation where it is important to get a clear understanding of what the rule is.

 

[ggunn]

I'm not sure it's a violation to have an EGC even though it isn't required, especially if I call it a GEC instead of an EGC. What code section would it violate? That is, as long as there's no current on the neutral; but this is a solar system.

I don't think you can bond the same EGC to neutral in two places.

 

[ggunn]

If there is current on the neutral, it will be shared between the neutral and EGC, if you bond the neutral and ground at multiple points. In the event that the EGC is significantly smaller than the neutral, and there is significant neutral current, then there is the possibility that the EGC carries more current than is permissible.


The EGC and system of non-current-carrying conductive materials should not be carrying current, except during fault conditions. This is why there should always be as few EGC-to-Neutral bonds as possible, except where otherwise required. So this is not a situation where you should just overbuild out of ignorance. This is a situation where it is important to get a clear understanding of what the rule is.

And I don't think it makes any difference as far as code compliance is concerned that one does not expect any unbalanced current on the neutral.

 

[jaggedben]

Okay, so as I said, change the name to GEC and suddenly I'm not only allowed to bond it to all the same enclosures, I'm required to. Note in the post that started this I didn't call it an EGC, I just called it a 'ground.' And again, if the inspector doesn't like it ... easier to take it out.

As for whether the code makes a distinction for EGCs depending on whether actually current flows on the neutral... Well, 250.6 simply says to arrange things to avoid objectionable current. If the solar source doesn't put any current on the neutral, how have I not met that requirement? Also, if I don't install the wire ground but bond the neutral at both places, then any current that could flow on the neutral could also flow on the connected metal enclosures and raceways. That is allowed if using the neutral to bond stuff at the service, and is not considered objectionable current. Having an extra ground wire in there isn't any different and doesn't make anything less safe.

 

[Carultch]

I don't think you can bond the same EGC to neutral in two places.

Bingo! That's the heart of the issue.

Two places per what, specifically so?
Per service?
Per service disconnect?
Per transformer secondary?

What is the flow chart for determining neutral-to-ground bonding?