Generator feed to 2L+1N system

[kevindd992002]

So if my neutral-earth bonding happens on the service panel (house side of c/o switch), where do I need to install an RCD/GFCI?

 

[Simon47]

OK, from your description, I think this is what you have before adding any changeover switch etc :

And after a bit of thought, this is what I think you need :

For your changeover switch, note that the N-E bonding is done BEFORE the switch - that avoids having a switch in your earth connection which (apart from some very specific situations) is expressly prohibited in our UK regs. Just to be clear, your earthing must not be reliant on any switch working properly.

And while you correctly point out that your two boards have their earth terminals connected together via the neutrals, I would explicitly bond them with an earth wire. Actually, I'd consider splitting the earth off the neutral at your meter (so L1, N, L2, and E from meter to each board) and avoid having the shared PEN (protective earth and neutral) internally - again it's something that's expressly prohibited in our regs. The issue is what if someone comes along and starts working on the system, on the assumption that (having pulled the supplier fuses) it's OK to disconnect the neutral. You now have an installation where part of it is supplied by a generator, and the earthing is split across two separate earth electrodes. Under fault conditions, that disconnected neutral could now carry a hazardous voltage relative to other neutral/earth conductors/terminals.
This is either-or - you don't need to run a separate earth wire from the meter location to each board AND also run a dedicated bonding wire between them. Either way, you'd have a solid single earth reference for the whole installation regardless of what's being done to the wiring. Also, this can be done just with conections to the earth bars in each board - you don't need to dig anything out to access the earth electrodes.


Now the fault protection.

As you can see, the neutral from the auto-transformer is connected to the earth bar. This means that (within the ability of the genny and transformer), an L1-E or L2-E fault within the installation will trip it's respective breaker. But the genny and/or transformer may not have the "oomph" to trip anything more than a smaller breaker so you could overload one or the other for an extended time. An RCD (GFI) would trip on the imbalance and disconnect the whole board.
With the assumption that the installation is otherwise adequately protected, this RCD is only to protect the equipment - so it can be of a larger trip current and time delayed (time delayed being the most important part) to give you some discrimination between that and what you have in the distribution board. E.g., if you already have an RCD/GFI on a circuit with a fault, then the one in the board would trip and disconnect the fault, while the time delay on the supply from the transformer will mean that it won't trip as well. But if you have a fault on a non-RCD/GFI protected circuit, then the RCD/GFI in your transformer supply will disconnect the whole system - inconvenient, but better than burning out your equipment.

You may have noticed an additional breaker on the output of the genny. This needs to be a true 2-pole (i.e. senses in both poles) breaker to protect the genny.
The genny only has a single pole breaker in one line, if there is an internal fault to earth, then a large current could flow with nothing to trip - and again there's a risk of burning out the genny or transformer. Consider fault between the genny winding on the side with the internal breaker and earth - then draw the path the fault current would take and you'll see what I mean.
Selecting this breaker could be tricky. Too low a tripping current and you'll get nuisance tripes, too high and the genny won't be able to drive enough current through it to trip. Another RCD/GFI might be more appropriate.

I have to say that this is somewhat outside of my comfort zone. It does really need someone familiar with your local practices and available equipment to do the detail design work - wire sizes, breaker ratings, etc. Hopefully I've explained the "why" of each part and you can see how the pieces fit together.

 

[kevindd992002]

@Simon47

OK, from your description, I think this is what you have before adding any changeover switch etc :
View attachment 63362

Yes, without a changeover switch, this is how it is setup. The N and G are tied in the main panel of both buildings.

And after a bit of thought, this is what I think you need :
View attachment 63363

For your changeover switch, note that the N-E bonding is done BEFORE the switch - that avoids having a switch in your earth connection which (apart from some very specific situations) is expressly prohibited in our UK regs. Just to be clear, your earthing must not be reliant on any switch working properly.

For the most part, I understand but I thought that (at least for the US and the Philippines) the standard is to bond N-E at the main panel alone? The main panel (distribution board as you call it) is AFTER the changeover switch. You also mentioned earlier to bond both B1 and B2 earth rods through the main panel ground bus bars. How does the grounding/earthing be reliant on the switch if N-E are bonded on the main panel?

This is what I meant:

https://www.electriciansforums.net/attachments/gen-changeover-jpg.61804/

@pc1966 do you have any comments on this BEFORE or AFTER changeover switch N-E bonding?

And while you correctly point out that your two boards have their earth terminals connected together via the neutrals, I would explicitly bond them with an earth wire. Actually, I'd consider splitting the earth off the neutral at your meter (so L1, N, L2, and E from meter to each board) and avoid having the shared PEN (protective earth and neutral) internally - again it's something that's expressly prohibited in our regs. The issue is what if someone comes along and starts working on the system, on the assumption that (having pulled the supplier fuses) it's OK to disconnect the neutral. You now have an installation where part of it is supplied by a generator, and the earthing is split across two separate earth electrodes. Under fault conditions, that disconnected neutral could now carry a hazardous voltage relative to other neutral/earth conductors/terminals.
This is either-or - you don't need to run a separate earth wire from the meter location to each board AND also run a dedicated bonding wire between them. Either way, you'd have a solid single earth reference for the whole installation regardless of what's being done to the wiring. Also, this can be done just with conections to the earth bars in each board - you don't need to dig anything out to access the earth electrodes.

Understood and I think this is what's happening to my installation now because the generator does not have a neutral. So when the B1 main panel is switched over to gen power, the B1 socket connections are:

220V -> L1 and L2 from genny, E from B1 ground rods and bonded to supplier N
110V -> L1 or L2 from genny, E from B1 ground rods and N from supplier (E and N bonded)

Now the fault protection.

As you can see, the neutral from the auto-transformer is connected to the earth bar. This means that (within the ability of the genny and transformer), an L1-E or L2-E fault within the installation will trip it's respective breaker. But the genny and/or transformer may not have the "oomph" to trip anything more than a smaller breaker so you could overload one or the other for an extended time. An RCD (GFI) would trip on the imbalance and disconnect the whole board.
With the assumption that the installation is otherwise adequately protected, this RCD is only to protect the equipment - so it can be of a larger trip current and time delayed (time delayed being the most important part) to give you some discrimination between that and what you have in the distribution board. E.g., if you already have an RCD/GFI on a circuit with a fault, then the one in the board would trip and disconnect the fault, while the time delay on the supply from the transformer will mean that it won't trip as well. But if you have a fault on a non-RCD/GFI protected circuit, then the RCD/GFI in your transformer supply will disconnect the whole system - inconvenient, but better than burning out your equipment.

What kind of RCD should I be looking at and should it have a larger trip current than the changeover switch breakers? I thought RCD's should be very fast acting? With a time-delayed RCD, are those just used to protect equipments? If we're talking about RCD's in subpanels, these should be very fast in case a person touches a live wire, isn't it?

Also, how high of generator changover switch breaker should I be looking for my current genny?

You may have noticed an additional breaker on the output of the genny. This needs to be a true 2-pole (i.e. senses in both poles) breaker to protect the genny.
The genny only has a single pole breaker in one line, if there is an internal fault to earth, then a large current could flow with nothing to trip - and again there's a risk of burning out the genny or transformer. Consider fault between the genny winding on the side with the internal breaker and earth - then draw the path the fault current would take and you'll see what I mean.
Selecting this breaker could be tricky. Too low a tripping current and you'll get nuisance tripes, too high and the genny won't be able to drive enough current through it to trip. Another RCD/GFI might be more appropriate.

So is it going to be like this?

Genny outputput -> RCD -> breaker -> changover switch breaker

This part I'm confused.

 

[pc1966]

@pc1966 do you have any comments on this BEFORE or AFTER changeover switch N-E

It seems this thread has run and run!

The critical point is you should NEVER be without a ground connection no matter what the switch is doing (even if stuck half-way).

Typically here we separate N & E from the incoming supply before you do anything, so in that sense the bond is before a changeover switch, but then we don't haver the link in the DB panel as you do which would normally be after a change-over switch.

 

[pc1966]

To some extent if you are linking N-E before the panel you might as well have only L1 & L2 switched as N & E are always common, and that would also be the auto-transformer centre tap (after any protection).

I see a 3-pole MCB appears above, it could be for the 100% load with the transformer rated at 50% as then the tap is good to that current anyway. So a 5kVA transformer on a 10kVA generator and a 40A or 45A 3-pole MCB is a sane choice, etc.

 

[kevindd992002]

It seems this thread has run and run!

The critical point is you should NEVER be without a ground connection no matter what the switch is doing (even if stuck half-way).

Typically here we separate N & E from the incoming supply before you do anything, so in that sense the bond is before a changeover switch, but then we don't haver the link in the DB panel as you do which would normally be after a change-over switch.

I apologize if the thread has run longer as it should.

As for the critical point, that is what I thought. I guess it's just really a matter of country specification like you and simon have mentioned. With N and E bonded at the DB panel (after the changeover switch), then I satisfy the "never be without a ground connection" criterion as it is never part of the switching mechanism.

To some extent if you are linking N-E before the panel you might as well have only L1 & L2 switched as N & E are always common, and that would also be the auto-transformer centre tap (after any protection).

I see a 3-pole MCB appears above, it could be for the 100% load with the transformer rated at 50% as then the tap is good to that current anyway. So a 5kVA transformer on a 10kVA generator and a 40A or 45A 3-pole MCB is a sane choice, etc.

I see what you're saying. Now circling back to what you mentioned in the beginning of this thread, I thought that we should NEVER let the neutral of the generator touch the supply neutral? If N-E are linked before the panel, then both neutrals are linked through the ground connection. Wouldn't that have the same undesirable effect?

So 45A for the 3-pole MCB for the transformer/generator combo, ok. Can I keep the 200A MCB that I currently use for the supply coming from the meter?

How about for the RCD, what rating should I be looking at? And with N-E linked on the DB panel on my setup, I'm assuming the RCD should also be installed on the DB panel itself possibly replacing the main MCB on that same panel?

 

[Simon47]

So, I meant to get back to this earlier ?

I'd agree with pc1966 - don't switch the neutral at all, use a 2 pole changeover switch.

It's not really a problem having the genny share the neutral with the supply - the supply is after all grounded and means you are sharing the ground. The genny is fully floating (best check that actually !) and you were going to earth the centre tap anyway.

As to breaker ratings, TBH it's too far outside my comfort zone to determine what's suitable - at least, not without spending time with my head in the books. I was hoping someone else would chime in - there are people on here who deal with gennys and portable setups as part of their normal job. Nothing like practical experience of what these things actually behave like. I have a 3kVA genny, and I know that it really really can't handle overloads - even transitory ones like trying to start a motor.

The RCD can be omitted IFF you can guarantee that under fault conditions either the load breaker will trip OR the genny breaker will trip. The problem there is that small generators are notorious for having poor overload capacity - so probably just isn't capable of producing enough current to trip even a moderately sized breaker on it's fast magnetic trip. For example, you have an 8kVA genny, so that's nominally 36A full load. A curve B MCB needs between 3 and 5 times it's rated current for the magnetic trip to operate - so a (say) 20A MCB will need between 60 and 100A. The genny won't produce that, so a fault on a 20A circuit might simply overload the genny until something else trips - while possibly exposing a user to an unsafe voltage on an "earthed" item. Even a 6A MCB (commonly used for lighting circuits here) needs between 18 and 30A (so assume it needs at least 30A) - and if there are a few amps of other load, again the genny might not be able to trip it.
So I'd say the RCD is probably essential for safety of people and animals. The only downside to fitting it is the possibility of nuisance tripping of the whole supply if "everything combined" has a bit of leakage. In general it's frowned upon here to have the whole installation on one RCD for this reason - but running from a small genny is a different situation.

The MCB on the genny output is to protect the genny from certain faults. With a winding-earth fault in the genny, "significant" current could flow round the circuit but NOT through the breaker that's on the genny panel. There's potential for this to overload the genny and the transformer until the genny burns out. There's an argument that as soon as such a fault occurs, the genny is effectively scrap anyway - but you don't want the risk of a fire, nor of burning out the transformer.
Again, I suspect that the genny would be incapable of tripping a 45A MCB even on it's thermal trip (which might not trip below 90A). Given the max rating of the genny equates to 38A, I suspect that it might struggle to even trip a 32A breaker.
And unless you are particularly agrresive with electricity usage, I think 32A (or even 25A) would be more than enough to keep the house running. That's 32A (or 25A) each on both the 110V sides, or 32A (25A) on 220V, or some combination in between.

That's my reasonings. I rather hope someone with first hand experience in this area can give their input on what rating devices they'd use.

 

[kevindd992002]

Ok, that's interesting. With a 2-pole changeover switch and bonding N and E before it, I think the modifications I need to make will be so much easier than originally planned. If you remember, the first diagram in your post here is my current setup. With that I mind, should I be doing all these?

• Disconnect N and E bonding both DB main panels?
• Connect ground bus bar of B1 DB main panel to ground bus bar of B2 DB main panel?
• What's the best way of bonding both Neutrals (supply and generator) to ground? Remember, my earth rods are already embedded into the ground under cement.

And yes, my generator has a floating output, that has been tested in the earlier parts of this thread here.

Also, I'm still concerned on @pc1966 's posts here and here where he specifically said:

"The autotransformer to create the neutral point must be on the generator side of your change-over switch. It must never be connected to the utility supply in that manner as that will likely destroy it."

"Actually if the local regulations don't require N to be on change over then best not to change things. Messing about with the board supply is definitely not a DIY or similar activity.

What is essential is you never put the generator or the autotransformer on to the utility supply, not at any time."

What is the reason for that? How will sharing the neutral of the supply with the generator destroy the supply?

As for the breaker ratings, I'm still trying to wrap my head around your reply I need to draw my own schematics as my head is hurting trying to understand just by reading. I'll circle back with you on this breaker rating topic.

 

[pc1966]

The point about the auto-transformer is it must not be directly on the supply L1-N-L2 at any point as any imbalance will cause a huge current as the transformer attempts to maintain it's winding ratio voltages.

When it is on the generator the source of power is only L1-L2 and the resulting transformer current is partly from that magnetising the core (and losses) but mostly it is from the load imbalance. The load in your house is not capable of such a huge current as your incoming supply, as fundamentally it is limited by the generator output.

However, as you are likely to have a transformer that is 50% or so of the generator output it really requires some protection in case of an overload that results in too much current on the transformer. Here a 3-pole MCB could be used to protect it and the generator, though as you have already realised they often just stall if seriously overloaded and don't deliver anywhere like the prospective fault current that a test would suggest (same as many UPS whose regulation suggests 2kA fault but in reality drop at 50A or so load).

So a 10kVA generator is about 42A at 240V. With a 240V-120V transformer for the centre tap at 5kVA is is the same on the 120V end (5000 / 120 = 41.67). Hence something like a 3-pole (linked operation) 45A MCB would provide reasonable protection for long-term thermal overloads, even if a short is likely to just stall the generator.

In terms of switching, in the UK the N-E link is on the supplier's side and it is prohibited to link N-E in the installation, so here we would have another N-E link in the generator and use a change-over switch that swaps neutrals as well as the line conductor(s). That way the E is always connected, but at any one time there is only on N-E link and it is always before the DB (panel).

But you are using USA style rules and there the N-E link is in the panel. In this case it really does not make sense to switch the N as you will have it linked in any case, and the less messing around with supply lines the better. That is why I later though you would be as well just switching the L1/L2 pair between the incoming supply and the 3-pole MCB, but leaving the N & E there all the time.

 

[pc1966]

This sort of idea: