I bought a house that had had solar previously but the owner had removed part of the system. What was left was a subsistence panel in a utility closet on the 2nd floor, a 2-pole transfer switch for selecting the souce of power for the subsistance panel as either a multiwire 60A branch circuit(2-pole breaker) in a nearby subpanel, or from inverters that were originally in the basement along with a solar charged battery.
The original inverter output wire was just left dangling in the basement - it is a #8 x 3 plus ground (black white red and green).
So I attempted to reconstruct this system and bought a couple of Xantrex SW4048 inverters to generate the two hotlegs. This was over 10 years ago when the thinking that has since gone into the code hadn't yet happened. It has gradually dawned on me through working on PV that there are some issues that were never thought about and maybe still not properly adddressed in the code. Here they are:
(1) Since there was only a 2-pole transfer switch, and it switched the hots, the neutral was unswitched. The neutrals of the inverter outputs were therefore bonded to the utility neutral using the neutral busbar in the transfer switch.
(2) However, are the inverters not a separately derived system, that needs to have its neutral output bonded to ground at source?
If so, I could no longer connect the neutrals together in the transfer switch, as I would now have a code violation of having the same neutral bonded to ground at two separate points, one in the main service panel, and again at the inverter outputs. So I would have to switch out the the 2-pole transfer switch for a 3-pole transfer switch that switched neutral too, as the code now requires for "separately derived systems"
(3) But that's not the end of the story; the Xantrex inverters also have a battery charging mode if the sun don't shine. They have a utility input comprising a neutral and both hots, and that neutral is connected straight through internally to the AC output neutral. That's how they work - they are bidirectional, converting DC to AC or AC to DC, and many modern inverters (e.g Outback) to the same thing. So, if I wire up the battery charging input, I still have my neutrals bonded together and grounded at two points in the system and that is a code violation
(and you will find out why if you have an electric guitar with a magnetic pickup - you will never get rid of the hum due to the 60Hz field enveloping your whole house! You will hear it in your phone and your stereo system too!))
(4) Therefore, it appears that you cannot ground the neutral of the inverter outputs separately; they HAVE to get grounded only via the utility neutral that supplies the battery charging input for their charger mode. (Unless you want to install a 10KVA isolating transformer!)
(5) Now my 2-pole transfer switch would get its utility neutral from the subpanel on the second floor, which is grounded all the way back at the main service panel, while my inverter neutral gets its ground via a separate branch circuit feeding it in the basement directly from a breaker in the main service panel. Now if I join the neutrals in the transfer switch, I have a huge neutral loop. Although there is only one point in the system where neutral is bonded to ground, there are two neutral return paths for both utility and inverter current - one through the second floor subpanel and one through the inverter in the basement direct to the main service panel. So my guitar buzzes like hell again!
(6) So it seemed like I had to have the three-pole transfer switch even though my inverters no longer have their neutrals grounded separately from the point the utility neutral is grounded.
(7) Moreover, the inverter system is totally reliant for its neutral-to-bond ground on the neutral from its battery charging input back to a 2-pole breaker in the main service panel. But there is NO ALTERNATIVE to that with inverters that have a battery charging input from the utility having a common neutral with their AC output.
So here are my conclusions:
a) A secondary power source whose neutral output is connected for what ever reason to the grounded neutral of a primary power source, is not a separately derived system. Its neutral output therefore does not need to be and in fact may not be separately grounded.
b) If any device is connected to both the secondary power source and the primary power source via different routes (i.e. wire in different conduits, or cables that come from one place in a building for one source and another place in a building for the other power source, the device may not connect their neutrals. Example: A transfer switch for selecting the primary or secondary power source must switch the neutrals even though we don't have a separately derived system.
c) If however a transfer switch is supplied by the primary and the secondary power source from a common location that allows a common neutral to be used, (e.g. one neutral and the two pairs of hots in the same conduit) then since there is only one neutral, the transfer switch does not need to switch the neutral
Perhaps, when a secondary power source INEVITABLY has to get ints neutral-to-ground bond via the neutral of a primary power source that is already bonded to ground, there should be a warning label at the location that the pimary source's neutral is bonded to ground that the neutral-ground bond is also the neutral-ground bond for a secondary power source, and that it should not therefore be disturbed without first de-energizing the secondary power souce as well as the primary power source.
There may be other solutions too, but I don't believe any of them are addressed in the code right now.
The original inverter output wire was just left dangling in the basement - it is a #8 x 3 plus ground (black white red and green).
So I attempted to reconstruct this system and bought a couple of Xantrex SW4048 inverters to generate the two hotlegs. This was over 10 years ago when the thinking that has since gone into the code hadn't yet happened. It has gradually dawned on me through working on PV that there are some issues that were never thought about and maybe still not properly adddressed in the code. Here they are:
(1) Since there was only a 2-pole transfer switch, and it switched the hots, the neutral was unswitched. The neutrals of the inverter outputs were therefore bonded to the utility neutral using the neutral busbar in the transfer switch.
(2) However, are the inverters not a separately derived system, that needs to have its neutral output bonded to ground at source?
If so, I could no longer connect the neutrals together in the transfer switch, as I would now have a code violation of having the same neutral bonded to ground at two separate points, one in the main service panel, and again at the inverter outputs. So I would have to switch out the the 2-pole transfer switch for a 3-pole transfer switch that switched neutral too, as the code now requires for "separately derived systems"
(3) But that's not the end of the story; the Xantrex inverters also have a battery charging mode if the sun don't shine. They have a utility input comprising a neutral and both hots, and that neutral is connected straight through internally to the AC output neutral. That's how they work - they are bidirectional, converting DC to AC or AC to DC, and many modern inverters (e.g Outback) to the same thing. So, if I wire up the battery charging input, I still have my neutrals bonded together and grounded at two points in the system and that is a code violation
(and you will find out why if you have an electric guitar with a magnetic pickup - you will never get rid of the hum due to the 60Hz field enveloping your whole house! You will hear it in your phone and your stereo system too!))
(4) Therefore, it appears that you cannot ground the neutral of the inverter outputs separately; they HAVE to get grounded only via the utility neutral that supplies the battery charging input for their charger mode. (Unless you want to install a 10KVA isolating transformer!)
(5) Now my 2-pole transfer switch would get its utility neutral from the subpanel on the second floor, which is grounded all the way back at the main service panel, while my inverter neutral gets its ground via a separate branch circuit feeding it in the basement directly from a breaker in the main service panel. Now if I join the neutrals in the transfer switch, I have a huge neutral loop. Although there is only one point in the system where neutral is bonded to ground, there are two neutral return paths for both utility and inverter current - one through the second floor subpanel and one through the inverter in the basement direct to the main service panel. So my guitar buzzes like hell again!
(6) So it seemed like I had to have the three-pole transfer switch even though my inverters no longer have their neutrals grounded separately from the point the utility neutral is grounded.
(7) Moreover, the inverter system is totally reliant for its neutral-to-bond ground on the neutral from its battery charging input back to a 2-pole breaker in the main service panel. But there is NO ALTERNATIVE to that with inverters that have a battery charging input from the utility having a common neutral with their AC output.
So here are my conclusions:
a) A secondary power source whose neutral output is connected for what ever reason to the grounded neutral of a primary power source, is not a separately derived system. Its neutral output therefore does not need to be and in fact may not be separately grounded.
b) If any device is connected to both the secondary power source and the primary power source via different routes (i.e. wire in different conduits, or cables that come from one place in a building for one source and another place in a building for the other power source, the device may not connect their neutrals. Example: A transfer switch for selecting the primary or secondary power source must switch the neutrals even though we don't have a separately derived system.
c) If however a transfer switch is supplied by the primary and the secondary power source from a common location that allows a common neutral to be used, (e.g. one neutral and the two pairs of hots in the same conduit) then since there is only one neutral, the transfer switch does not need to switch the neutral
Perhaps, when a secondary power source INEVITABLY has to get ints neutral-to-ground bond via the neutral of a primary power source that is already bonded to ground, there should be a warning label at the location that the pimary source's neutral is bonded to ground that the neutral-ground bond is also the neutral-ground bond for a secondary power source, and that it should not therefore be disturbed without first de-energizing the secondary power souce as well as the primary power source.
There may be other solutions too, but I don't believe any of them are addressed in the code right now.