Adding a back up 8k Generator, want to get the N-G situation right

[OffGridForGood]

I got a Champion 8kW back up generator dedicated for on-site power system.
A few details first:
My solar is located in my work shops next door to my home,
The solar is 7kW PV, twin MPP 6048 Inverters (each produces 120/240, each have internal N-G bond only during inverting mode)
Energy storage system is 760Ahr 48v LiFePO4 rack batteries some commercial units and some DIY.
For poor solar Nov-Dec period, I have a dedicated 30A 240vAC supply from the house to supply power for battery recharging off peak TOU,
For powering during on-Peak and during good solar (March - October) the solar supplies a Critical loads panel at the house, 30A 240vAC.
I liver rural in a place with lots of snow, high winter winds and big trees = we can have power outages that last 4 days.

The plan:
In the normal case, if solar is poor, the inverters will trigger charging with the 30A 240 grid supply. To prevent this from taking place during on-peak TOU, I have this power supply running through a 240vAC HWT Timer (Intermatic 30A 240v programable timer) this timer limits utility power from being available to the off-peak hours (7PM to 7AM Mon-Fri, all day SAT SUN). I have the Charge Trigger set high enough that even if the Inverters are 'looking for charging current' early in the morning, the large battery storage will run the full shop all day anyway, and at 7PM when the timer switches on utilty, then (and only then) the inverters go into by-pass mode, and charge the batteries. This has the side benefit of allowing any PV charging available to have already taken place.
My system Never supplies power to the utility.

So what about when the Grid is Down:
I have purchased the generator, a three pole Manual Transfer Switch, and an exterior Generator Plug 30A 240 4-pole.
I read the Generator manual, located and removed the N-G bond wire jumper, and put a label on the generator "Floating Neutral". A test with an ohm meter before removal of the jumper showed N-tied to-G in the generator, and after removal of the jumper, the ohm meter confirms Open between N and G. (the same if running or not).
The plan, is to install the MTS after the timer, to control N, L1, L2 Utility and N, L1,L2 Generator supply to the twin MPP 6048's. The existing Ground bare wire from the utility will pass through the MTS and tie to the generator G. I believe this is key, along with the removal of the N-G in the generator.
My thoughts are, if utility is unavailable and batteries are getting low, I can start up the 8k generator, warm it up, plug it into the exterior wall plug, then turn the MTS from Utilty to Generator and let it run long enough to recharge the battery packs. At 7:00 PM the shops are not running equipment, so loads are low power items mostly lights. The MPP's will use by-pass for any loads while charging up batteries, in by-pass the MPP's break their internal N-G bond, and rely upon the 'grid'. With the Ground from the transfer switch passing through and connecting to the generator wall plug, I believe I maintain a single point of N-G bond (back at the utiltiy panel) and all should be well.

However since this is a complex issue, and asking for comments is less hazardous then blindly connecting 30A 240 (7kW) of power 'to see what happens' I thought perhaps someone would comment on the logic, I am in no hurry, grid is on tonight nothing is connected up yet (MTS and Generator are sitting patiently waiting on the shop floor).

 

[FilterGuy]

OK..... Is this the current setup? Please verify the routing of ground and neutral.


Is this how you would like to add the GenSet?


Assuming I have it drawn correctly, the NG bonding is *not* proper. When the MTS is switched to the Genset and the inverters are in passthrough mode, there will be no NG-Bond. This can be fixed in one of two ways:

1) Add the NG Bond Jumper back into the genset.
2) Leave the NG Bond Jumper out of the genset but don't switch neutral in the MTS.

Also, there are two other problems with both the before and after pictures.

1) When on battery, there will be two NG bonds. The only way to fix this is to remove the bonding screw from one of the inverters, but supposedly that voids the warranty. (MPP leaves the user in a catch 22)
2) There is a big loop in the grounding system. This can be fixed by disconnecting the ground wire from *one* of the inverters inside the critical load box. (Just cap it off)

 

[OffGridForGood]

Nice sketches,
Not quite what I have set up:
The Critical Loads Panel is independent of the timer switch, it is a Reliance 306, and is connected to utiliy on one side, and if utility is lost, each of the six circuits can be manually switched to 'generator' - which is actually power from the solar powered panel, not a generator. Grounds and neutral do not switch in the Reliance.
(Edit: more detail)
MPP 6048's each have N L1 L2 and G terminals on both AC-input, and AC-output, yes, ( bit different than the 6548's)
Starting with the second sketch,
The Reliance CLP should be drawn in parallel with the Timer, and is supplied by the Utility normally (or the MPP-output, using individual manual switches during utility loss) regardless of what the timer is doing. Because the Reliance doesn't switch the Neutral, and it connects to the MPP's independantly of the new MTS for the generator/utility battery charging, would this be correct?

MPP N-G relays:
I have pondered as you say, disconnecting One N-G relay in one MPP, (the slave) to avoid the double N-G in inverting mode.
I was worried to void my warranty in the early days, since it would seem a shame to lose that support day-one. At this stage, one inverter is two years running 24/7 and the other will soon be one year. The two inverters sit 8" appart, and it seemed the 'double N-G bond' hazard would be limited to the short distance betweeen the inverters, and thus be low risk?

 

[Hedges]

Also, there are two other problems with both the before and after pictures.

1) When on battery, there will be two NG bonds. The only way to fix this is to remove the bonding screw from one of the inverters, but supposedly that voids the warranty. (MPP leaves the user in a catch 22)
2) There is a big loop in the grounding system. This can be fixed by disconnecting the ground wire from *one* of the inverters inside the critical load box. (Just cap it off)

As you have drawn it for him, neutral and ground of the two inverters meet at critical load box (and at MTS.)
Running on battery, there can be "objectionable current" in those ground wires to critical loads box due to two neutral-ground bonds.

Would it be better to strap ground of inverters together, strap neutral of inverters together, then run one neutral wire and one ground wire to critical load box?
That way, if the inverters have any objections, it is between themselves. Longer run to breaker panel is left out of it.

Wire it with neutral and ground coming from main breaker panel (where they are bonded) to one of the inverters, on to critical loads panel. That by itself would be compliant.
Connect second inverter with neutral and ground wires stubbed off the first. Don't have conductive conduit or other grounds going from second inverter to anything else, only to first inverter.
This way, the only objectionable current is in ground between the inverters.


Since OP is using Reliance, it is a bit different. Reliance as SPDT and DPDT switches for L1 and/or L2, and breakers (single and dual pole) for when loads are powered by inverter.
Circuits that were originally in main breaker panel get connected by extension wire to the switches, also original breakers to those switches. In one position, switch just receives L1 (and/or L2) from breaker and sends it back. For backup, switch feeds the circuit from breaker in Reliance.

Reliance is supplied with a flexible conduit pre-stuffed with wires to connect original circuits and breakers.
Inverter feeds L1 and L2 (but not neutral) into reliance, and gets L1 and L2, also neutral, from main breaker panel by another user-supplied conduit.
What is a little weird is that neutral isn't routed in the conduit that carries L1/L2 from inverter to Reliance to main breaker panel. Current flows in a large loop through separate conduits.

GFCI circuit don't work as intended with this, but I think you could repurpose a 2-pole L1/L2 circuit for L1,N and make it work.

 

[OffGridForGood]

As you have drawn it for him, neutral and ground of the two inverters meet at critical load box (and at MTS.)
Running on battery, there can be "objectionable current" in those ground wires to critical loads box due to two neutral-ground bonds.

Would it be better to strap ground of inverters together, strap neutral of inverters together, then run one neutral wire and one ground wire to critical load box?
That way, if the inverters have any objections, it is between themselves. Longer run to breaker panel is left out of it.

Correct, the two MPPs' are side by side and feed a normal 100A load centre in the shop a few feet away, a 30A2P breaker from this off grid panel feeds back to the house to one side of the Reliance. The other side of the Reliance is fed from main Utility panel exactly as you describe. To my mind, the double N-G is 'local' to just the 2-3 feet between the MPP's and the panel they supply from their AC-output. I am open to removing One N-G in the slave MPP if needed, however, if this were done, and the master MPP shut down, I would lose N-G bond during inverter mode (yes?) each MPP6048 supplies L1 L2 120/240

 

[Hedges]

Because you use reliance and neutral is not switched, your loads (original wiring to main panel) always has neutral provided by main panel and bonded to ground.
When inverters internally disconnect output neutral from input neutral, and connect inverter's neutral to output, they also bond output neutral to ground (if above diagram of MPP 6048 is correct.)
However, output neutral does not go to Reliance. I think you need output neutral hard-wired to input neutral, or back to main breaker panel. Otherwise, inverter is trying to wiggle Line but has no Neutral connection, so zero current would flow.

(My Sunny Island only switches one line; neutral is unswitched. So I made a single neural connection. Yours switches, so you can defeat that connecting input neutral to output.)

MPP 6048 bonds neutral to ground when operating off battery. You now have a stub of neutral from main breaker panel (where it is bonded to ground) over to MPP 6048 where neutral is again bonded to ground. Objectionable current will flow in that ground, perhaps half the current in L1 because it splits equally between equal size neutral and ground wires? This could cause a slight rise in voltage of inverter chassis. Ensure it is solidly grounded. So long as either neutral or ground is connected, chassis voltage remains low. If both opened, 120V would appear on chassis.

No elevated ground voltage should appear at branch circuits, because they ground back to main panel not to inverter or any point with "objectionable current."

I think separate ground and neutral wires for each inverter to main breaker panel is OK in this case, since they don't supply neutral to a critical loads panel.

If you wire input neutral to output neutral, and wire that to main breaker panel which is bonded, I think you could remove both inverter bonding screws.

If you have not wired input neutral to output neutral (please provide us with a detailed schematic!) then I think you have "Objectionable Current" in ground wire equal to 100% of line current! (or for split phase, 100% of N = L1 - L2)

 

[FilterGuy]

if this were done, and the master MPP shut down, I would lose N-G bond during inverter mode (yes?) each MPP6048 supplies L1 L2 120/240

When stacked for split phase, if one inverter shuts down, the other shuts down as well.

Nice sketches,
Not quite what I have set up:
The Critical Loads Panel is independent of the timer switch, it is a Reliance 306, and is connected to utiliy on one side, and if utility is lost, each of the six circuits can be manually switched to 'generator' - which is actually power from the solar powered panel, not a generator. Grounds and neutral do not switch in the Reliance.
(Edit: more detail)
MPP 6048's each have N L1 L2 and G terminals on both AC-input, and AC-output, yes, ( bit different than the 6548's)
Starting with the second sketch,
The Reliance CLP should be drawn in parallel with the Timer, and is supplied by the Utility normally (or the MPP-output, using individual manual switches during utility loss) regardless of what the timer is doing. Because the Reliance doesn't switch the Neutral, and it connects to the MPP's independantly of the new MTS for the generator/utility battery charging, would this be correct?

MPP N-G relays:
I have pondered as you say, disconnecting One N-G relay in one MPP, (the slave) to avoid the double N-G in inverting mode.
I was worried to void my warranty in the early days, since it would seem a shame to lose that support day-one. At this stage, one inverter is two years running 24/7 and the other will soon be one year. The two inverters sit 8" appart, and it seemed the 'double N-G bond' hazard would be limited to the short distance betweeen the inverters, and thus be low risk?

I am goint to need to see a drawing to understand what is going on.

 

[OffGridForGood]

I will start with an appology, should have started with the sketches.
The question from Heges about Neutral IN and Neutral OUT on the MPP's - I put an ohm meter on the N-in N-out on the MPP's they are closed-loop.

I opened up the panels and Reliance and made two sketches on a note-pad (sorry not very pretty) showing what is connected to what and where. This should make discussion easier. (I don't think it matters for this discussion, but just to be clear, the house has a main and a sub-panel and the reliance 306 is tied to the sub-panel, not the main Utility panel.)

First sketch - the main 200A House panel 'first point of disconnect after the utility' - this is the location of my N-G bond, and where the main earth-ground ties to a 10-foot copper grounding rod outside. From the main panel I show two 2P circuits: a 100A that feeds the sub-panel 35 feet away in a mechanical room, where the Reliance 306 is installed; and a 30A that feeds all the way to the shop 80-feet away, that supplies Utility power to the MPP-IN AC for battery charging and pass-through. The MPP settings limit the pass-through and battery charging combined to a total of 30A.

Second Sketch - the set up in the shop:
The shop runs off-grid most of the time. It's main panel is fed from the 6048MPP-Out AC x 2
One 2P 30A breaker supplies power from this off-grid panel through a buried line to the house, where it supplies one side of the Reliance, in place of what the Reliance would call "Generator".
The 30A 2P battery charging/pass-through power source from the house does not tie to the off-grid panel EXCEPT for Ground. Lets say that more clearly: The Solar off-grid panel is unbonded, and has no earth-ground rod of it's own, it obtains earth-ground from the main panel via the ground conductor cable that is otherwise the Grid-source for charging batteries/pass-through. To be more clear, this ground connection is unaffected by the timer no matter what the timer is doing.

 

[FilterGuy]

Is this correct? (I have a few comments but want to know if it is correct first)

 

[Hedges]

I don't think Reliance has a neutral connection.

The question from Heges about Neutral IN and Neutral OUT on the MPP's - I put an ohm meter on the N-in N-out on the MPP's they are closed-loop.

Is neutral switched between grid and inverter, as well as bonded, like FilterGuy showed?
If so they would only be connected between input and output when the relay was in pass-through.
(but, could also have continuity through ground in inverter mode.)
Make sure you aren't using ground wire as neutral wire.

 

[OffGridForGood]

Is neutral switched between grid and inverter, as well as bonded, like FilterGuy showed?
If so they would only be connected between input and output when the relay was in pass-through.

The inverters were inverting from batteries when I checked the N-in to N-out and confirmed they are closed loop (zero ohms)
The Diagram:
Much appreciate the effort this took to put up, Looking about right, but let me double check the Reliance - this is where things are a bit tricky to follow.
I will go open it up right now, and double check. There are no 2P loads in my Reliance, only because all my critical load items all 120V (even my well pump is 120v)
I will post an update shortly, once I open up the Reliance again. I am bringing the diagram with me...

 

[FilterGuy]

I don't think Reliance has a neutral connection.

It does have a neutral. but it does not switch it.

 

[FilterGuy]

The inverters were inverting from batteries when I checked the N-in to N-out and confirmed they are closed loop (zero ohms)
The Diagram:
Much appreciate the effort this took to put up, Looking about right, but let me double check the Reliance - this is where things are a bit tricky to follow.
I will go open it up right now, and double check. There are no 2P loads in my Reliance, only because all my critical load items all 120V (even my well pump is 120v)
I will post an update shortly, once I open up the Reliance again. I am bringing the diagram with me...

Those inverters will have no internal bond while in pass-through mode. At all other times (including when completely off), they will have an N-G bond.

 

[OffGridForGood]

I don't think Reliance has a neutral connection.

Yes, the Reliance has G and N wires for connection to "Generator" - but in my installation the "generator" is my 30A 240 circuit from the off-grid panel run to the house from the shop. Two Photos, the overall Reliance with the lower cover off, and close up on the ground and Neutral tie ins.

Edit: I forgot to note, far right lower corner - this is the power from the off-grid solar, while the entry in the bottom centre is the connection to the subpanel right beside the Reliance.

 

[OffGridForGood]

It does have a neutral. but it does not switch it.

Correct,
At the MPP's I am wondering about Hedges comment that I may measure zero ohms N-in to N-out while in inverting mode, due to the N-G tie back at the main panel. (am I incorrect on this thinking?) Starting the believe I need Both N-G removed from both MPPs'

 

[Hedges]

Correct,
At the MPP's I am wondering about Hedges comment that I may measure zero ohms N-in to N-out while in inverting mode, due to the N-G tie back at the main panel. (am I incorrect on this thinking?) Starting the believe I need Both N-G removed from both MPPs'

That is my thought.
If measuring ground current isn't convenient (because conduit carrying ground current has "line" inside it, see if you can measure L1, L2, N (both N together), while unevenly loading L1 and L2. If amperage doesn't add up correctly, that suggests ground is carrying current.

If you don't have a clamp AC/DC ammeter yet, they are handy. I bought the Harbor Freight 1000A one, and FilterGuy did on my recommendation (then had to repair it.)

 

[FilterGuy]

Here is the drawing with a 120V load. It does not change the grounding/bonding in any way.

 

[OffGridForGood]

That is my thought.
If measuring ground current isn't convenient (because conduit carrying ground current has "line" inside it, see if you can measure L1, L2, N (both N together), while unevenly loading L1 and L2. If amperage doesn't add up correctly, that suggests ground is carrying current.

If you don't have a clamp AC/DC ammeter yet, they are handy. I bought the Harbor Freight 1000A one, and FilterGuy did on my recommendation (then had to repair it.)

I have a good clamp meter, and even a millivolt level meter if needs-be.
I like where your going: deliberatly imbalance the loads on the off-grid panel then check ground for any current flowing due to the imbalance.

I have to point out something un-obvious, my MPP-Out wires are all insulated, even the Grounds are insulated (green jacket) wires back to the off-grid panel, so, these could have been carrying some current and not be noticed. I will report back, after some imbalanced testing.

 

[Hedges]

If you have conductive conduit, then conduit could carry neutral current and insulated ground wires could carry neutral current, if neutral is bonded to ground in two places.

I'm not sure yet if you've got neutral in a loop with inverter relay sometimes connecting the loop and sometimes breaking it as FilterGuy drew,
or if when relay opens there is no path back except through neutral ground bond and 100% of current flows in ground.

If a loop, I think that can carry "objectionable current", part of the neutral current in ground.

Is his drawing now correct, or can you mark it up?

 

[FilterGuy]

Is his drawing now correct, or can you mark it up?

Yup. That is what I am waiting on before I comment.
(There are several grounding/bonding issues with what I drew).

 

[FilterGuy]

BTW: We have not gotten to the part of adding a Gen-set to the mix.

 

[OffGridForGood]

I have a good clamp meter, and even a millivolt level meter if needs-be.
I like where your going: deliberatly imbalance the loads on the off-grid panel then check ground for any current flowing due to the imbalance.

I have to point out something un-obvious, my MPP-Out wires are all insulated, even the Grounds are insulated (green jacket) wires back to the off-grid panel, so, these could have been carrying some current and not be noticed. I will report back, after some imbalanced testing.

The inverters running on batteries:
The loads on the off grid panel -I put 3A on L1, 13.2A on L2 measured zero Amps on grounds (from each MPP to the panel), there was amperage on the Neutrals that made sense with the imbalanced loads.

 

[OffGridForGood]

BTW: We have not gotten to the part of adding a Gen-set to the mix.

TRUE!
But lets be sure everything is right with the base system before we get to that.
The last drawing you did is correct, although I am not certain how the internal MPP works, I have not removed any internal bonding from either MPP.

 

[FilterGuy]

The last drawing you did is correct, although I am not certain how the internal MPP works, I have not removed any internal bonding from either MPP.

First: Please don't shoot the messenger. You have invested a lot of time and money in your system and probably won't like what I have to say.
However, I think it is important for people to understand potential issues. They can choose to ignore them, but at least it is a conscious decision.

1) If you trace the neutral, you will find that the output neutrals of the inverters are connected back to the input neutrals through the Reliance. (I call this a 'common neutral' configuration. MPP has stated pretty unequivocally that this is dangerous and should not be done. However, as you can tell from your system, this appears to work. I do not know of a way to fix this without using transfer switches that switch neutral. None of the multi-circuit transfer switches like the reliance switch neutral (well, at least none that I have seen).

A lot of people have systems set up this way and I have not heard reports of issues. Furthermore, MPP has not provided info on what the danger is so I can not say how bad the situation is. Personally, I would not build a system in a way the manufacturer says is dangerous.

2) As a side effect of the common neutral, the main bonding jumper at your service entrance is 'seen' by the loads at all times. (You can prove this to yourself by tracing both the neutral and ground from the load to the N-G bond). This means that when the inverters are in battery mode, there are three N-G bonds in the system. This does not meet the code. As long as the system is wired with a common neutral, the only way to fix this is to remove the bonding screws from both systems. However, I have heard this voids the warranty. Even without the problem with the common neutral, when the inverters are in battery mode there are two N-G bonds. If there was not a common neutral, *one* of the two bonding screws could be removed and the system would always have only one NG bond. Since removing the bonding screw voids the warranty, they give you no way to set up the system properly without voiding the warranty.

A lot of people have systems with multiple N-G bonds. The primary danger is when the system is being serviced, but the danger is high enough that the NEC bans it. I can not speak for others, but I would not do it.

3) Loops!!! Lots of Loops in neutral, ground, and netutral-ground.

3.1) Loops in the Ground wire.
In the image below, one of the many loops in the ground is traced in purple.


This is not a code violation, nor is it a safety issue. However, the loops can resonate from noise from the inverters and become big radio noise broadcasting antennas. This can manifest in many ways. Static on AM or FM radio. Dead spots or very slow WiFi. Bad/no cell phone reception, etc.

The good news is that this can be fixed with careful routing of the ground wires. In the diagram below, everything is grounded but there are no loops in the ground system.


Notice that all I did was eliminate the following 3 ground wires.
* inverter 2 -> Off-grid load box
* Off-grid load box -> reliance
* Time switch -> off-grid load box
This breaks all of the loops in the grounding circuit.

HOWEVER. The NEC states that the ground must be run along with the hot wire. The letter of this rule could be satisfied by running the wire but only connecting it on one end. As an example, in the above diagram, there is no ground wire running with the hot wire from the 2nd inverter to the off-grid load box. You could run the wire from the inverter along with the hot, but at the load box just cap it off with a wire nut in the load box.

3.2) Loop in Neutral
With the above diagram, there is still a big loop through the neutral wires. This loop may be even more likely to cause RFI issues because the neutral is carrying current from the inverter. It is not clear this loop can be completely eliminated in a common neutral layout. I have imagined a way to minimize the loop, but it is so weird I hesitate to even mention it. If you want to see it, please PM me.

3.3) Loops in Neutral + Ground.
Anytime there are two or more N-G bonds it creates a loop with part of the loop on neutral and part of the loop on the ground. See above for a discussion of multiple N-G bonds.

 

[FilterGuy]

I am probably getting ahead of things, but here is the circuit with the gen set and MTS inserted after the timer switch.



This would work and would not add to any of the grounding and bonding issues. However, if you fix the Common Neutral issue, the bonding jumper should be added back into the gen-set.