Pv on rcd

[GEOFF S]

Why is an RCD required between the inverter and final circuits?
Inverter is connected to the incoming supply from the grid via SY cable through trunking to a meter then to an isolator , then to fuses (can't remember the ratings) then via SWA to to a busbar chamber in the intake room.
Not sure if there is d.c./AC separation at the inverter, though there is an indicator led for if there is an earth fault on the d.c. side?

Hi

Without simple seperation between d.c and a.c, a fault on dc could transfer to the a.c side, so a type B RCD must be installed.

Cheers

 

[Deleted member 26818]

Seems a bit complicated when just two RCDs will suffice.

 

[uksparks]

Seems a bit complicated when just two RCDs will suffice.

Not two normal ones?

 

[GEOFF S]

What's to explain?
Earth and neutral are connected together at the inverter.
Don't know what you mean by seperation and needing an RCM?
Why would it require an RCD?
Would existing RCDs (if there are any) in the installation, require changing from A/AC to B type?

Hi

How you know the N-E link is internal and its purpose?

Cheers

 

[Deleted member 26818]

As far as I am aware, these particular inverters do not require type B RCDs.
I know that the earth and neutral are linked by visual inspection.
I was interested in why there is a neutral for 3 phase inverters, when to my mind there would be no need, as there is no transformer to connect a neutral to the star point.
As no one I asked could provide an answer as to whether a neutral is required, I thought I would investigate the next inverter I got hold of.
There is a wiring block to connect the output conductors, then some short lengths of conductors to over voltage surge arresters, with blank modules for the earth and neutral, the other side of the surge arresters, is where the earth and neutral are linked, with only the earth and phase conductors continuing into the innards of the inverter.

Something that puzzles me, is how the inverter knows there is a connection to the grid?
Does the inverter pause the output from the inverter to measure the grid, or is there some manner that the inverter can differentiate between the grid voltage and the inverter voltage?

 

[davesparks]

I have to say this is a really interesting thread!

if we need to disconnect both ends, how about this.

from the main db, circuit starts with an RCBO, then connected to that would be a 6A MCB on a shunt and that circuit holds open a NO contactor which the inverter end is connected to (the one that runs back to the main db).

so by doing this, should there be overload or earth fault on that circuit, thevRCBO will trip and disconnect the line at the board end and open the contactor disconnecting both line and neutral from the inverter. But would this disconnect or open the contactor in the right time to comply?

Just a thought?? No idea if it would be ok or not

How is a 6A MCB with a shunt trip going to hold a contractor open?

 

[davesparks]

Why is an RCD required between the inverter and final circuits?
Inverter is connected to the incoming supply from the grid via SY cable through trunking to a meter then to an isolator , then to fuses (can't remember the ratings) then via SWA to to a busbar chamber in the intake room.
Not sure if there is d.c./AC separation at the inverter, though there is an indicator led for if there is an earth fault on the d.c. side?

It is required if the inverter does not provide suitable separation between the DC and AC.
The basic outline of the theory is that if DC can get into the AC circuit then it will prevent a A or AC RCD from operating as the DC saturates the sensing coils these preventing the RCD from operating.

Injecting DC into an AC RCD is one method which can be used in 'anti-trip' Zs testing to prevent an RCD from tripping.

I'd be surprised if a non-domestic unit wasn't provided with appropriate separation, but it really isn't my field so I don't know.

 

[davesparks]

Seems a bit complicated when just two RCDs will suffice.

Two RCDs are unlikely to suffice, but we've strayed so far in this topic I'm not certain what you are saying the two RCDs would doing?

This started with the provision of 30mA RCD protection to a PV inverter circuit (in a domestic setup if that makes any odds) in a situation where regs require it for the usual shock protection reasons. But now it's strayed into type B RCDs and N-E links so I'm getting a little lost!

 

[Deleted member 26818]

If the cable from the inverter to the mains requires RCD protection as you said because it may be concealed in a wall less than 50mm.
Then two RCDs, one at each end of the cable would suffice.
One will disconnect the power from the inverter, and the other the power from the mains.
I mentioned the fact that the neutral and earth are linked because someone asked if all inverters are separated from earth.

 

[davesparks]

If the cable from the inverter to the mains requires RCD protection as you said because it may be concealed in a wall less than 50mm.
Then two RCDs, one at each end of the cable would suffice.
One will disconnect the power from the inverter, and the other the power from the mains.
I mentioned the fact that the neutral and earth are linked because someone asked if all inverters are separated from earth.

Two RCDs may not suffice, if the inverter is earth referenced through a solid link then it should work, but if it's earth referenced through a resistance or through fortuitous connection it may not.

The bottom line that most of us were getting at is that the circuit should be designed so that RCD protection is not required, which isn't exactly difficult really.

 

[uksparks]

How is a 6A MCB with a shunt trip going to hold a contractor open?

It it will hold it closed sorry, then when triggered it will release the contactor and shut the PV end off.

 

[GEOFF S]

I have to say this is a really interesting thread!

if we need to disconnect both ends, how about this.

from the main db, circuit starts with an RCBO, then connected to that would be a 6A MCB on a shunt and that circuit holds open a NO contactor which the inverter end is connected to (the one that runs back to the main db).

so by doing this, should there be overload or earth fault on that circuit, thevRCBO will trip and disconnect the line at the board end and open the contactor disconnecting both line and neutral from the inverter. But would this disconnect or open the contactor in the right time to comply?

Just a thought?? No idea if it would be ok or not

Hi

Im lost, whats this disconnect both ends about? The inverter must disconect within 5 secs, as we cant connect to Final circuit the a Distribution circuit it is, 5 Sec disconection.

Cheers

 

[uksparks]

Two RCDs may not suffice, if the inverter is earth referenced through a solid link then it should work, but if it's earth referenced through a resistance or through fortuitous connection it may not.

The bottom line that most of us were getting at is that the circuit should be designed so that RCD protection is not required, which isn't exactly difficult really.

Even if it's on an MCB you will still have the issue of two supplies.

 

[uksparks]

Hi

Im lost, whats this disconnect both ends about? The inverter must disconect within 5 secs, as we cant connect to Final circuit the a Distribution circuit it is, 5 Sec disconection.

Cheers

but if you create an earth fault or even dead short or whatever, the PV could still be feeding the circuit that you should hope would be dead.

 

[davesparks]

It it will hold it closed sorry, then when triggered it will release the contactor and shut the PV end off.

You don't actually know what a shunt trip is do you?

A shunt trip is a unit which you can fit to an MCB to allow it to be tripped by a remote signal. They can be used to add RCD protection to larger mcbs or mccbs/acbs by using an earth leakage relay connects to a shunt trip to operate the breaker.

This is why I suggested that a DP MCB at the inverter fitted with a shunt trip which is operated by an auxiliary contact on the RCD at the CU end might provide adequate protection.

 

[davesparks]

Even if it's on an MCB you will still have the issue of two supplies.

What issue of two supplies?

 

[davesparks]

Hi

Im lost, whats this disconnect both ends about? The inverter must disconect within 5 secs, as we cant connect to Final circuit the a Distribution circuit it is, 5 Sec disconection.

Cheers

It came up as a result of the mention of a PV circuit being provided with RCD protection.
I suggested the only way a PV circuit which requires 30mA RCD could be provided with it is if the RCD disconnects both ends of the supply.
But I thought it would illustrate why a PV circuit should be designed to not need RCD protection,

 

[uksparks]

I didn't mean a shunt, I mean one of those things you use to fix an indicator to an MCB for example, I can't think what they are called, I call them a shunt, it's probably not but Hager do them anyway.

 

[Deleted member 26818]

Two RCDs may not suffice, if the inverter is earth referenced through a solid link then it should work, but if it's earth referenced through a resistance or through fortuitous connection it may not.

The bottom line that most of us were getting at is that the circuit should be designed so that RCD protection is not required, which isn't exactly difficult really.

Don't see why an RCD would be affected by the inverter being referenced to earth?
Difficult, I'd say virtually impossible with a TT installation.

 

[GEOFF S]

As far as I am aware, these particular inverters do not require type B RCDs.
I know that the earth and neutral are linked by visual inspection.
I was interested in why there is a neutral for 3 phase inverters, when to my mind there would be no need, as there is no transformer to connect a neutral to the star point.
As no one I asked could provide an answer as to whether a neutral is required, I thought I would investigate the next inverter I got hold of.
There is a wiring block to connect the output conductors, then some short lengths of conductors to over voltage surge arresters, with blank modules for the earth and neutral, the other side of the surge arresters, is where the earth and neutral are linked, with only the earth and phase conductors continuing into the innards of the inverter.

Something that puzzles me, is how the inverter knows there is a connection to the grid?
Does the inverter pause the output from the inverter to measure the grid, or is there some manner that the inverter can differentiate between the grid voltage and the inverter voltage?

Hi

All inverters with no simple seperation require RCD type b protection.

I can only assume the inverter reacts the same as a tx, three phase unbalanced load, neutral current, connecting the neutral to earth to tie star point down etc.

The inverter will need to synchronise with the grid voltage, id assume at this point it can differentiate.

Cheers

 

[GEOFF S]

It came up as a result of the mention of a PV circuit being provided with RCD protection.
I suggested the only way a PV circuit which requires 30mA RCD could be provided with it is if the RCD disconnects both ends of the supply.
But I thought it would illustrate why a PV circuit should be designed to not need RCD protection,

Hi

Only one rcd would be required at teh output of the Inverter.

Cheers

 

[Deleted member 26818]

Not for a fault in the cable between the inverter and mains.
As has already been posted, if an RCD were to cut the power from the inverter, it would leave the power from the mains connected.
There would need to be one at each end of the cable, one for the mains and one for the inverter.

 

[davesparks]

I didn't mean a shunt, I mean one of those things you use to fix an indicator to an MCB for example, I can't think what they are called, I call them a shunt, it's probably not but Hager do them anyway.

There's a multitude of devices which can be added to breakers to perform any function you want

 

[davesparks]

Hi

Only one rcd would be required at teh output of the Inverter.

Cheers

But then the grid supply would not be disconnected in the event of a fault.
This is a cable with a source of supply at both ends.

 

[davesparks]

Don't see why an RCD would be affected by the inverter being referenced to earth?
Difficult, I'd say virtually impossible with a TT installation.

If the inverter N is solidly referenced to E then the RCD will work. If the inverter N is not referenced to earth then the RCD is unnecessary as you have an electrically separate supply so it cannot deliver a shock.
But what if there is a N-E link which is fortuitous?
(I don't know the answer and am probably overthinking now)

 

[davesparks]

Don't see why an RCD would be affected by the inverter being referenced to earth?
Difficult, I'd say virtually impossible with a TT installation.

In a TT you would still want to design the circuit to not need 30mA RCD, only the fault protection RCD which is likely to required, and ensure that the inverter used does not include a N-E link so providing a n electrically separate supply when not connected to the grid.


In theory (and yes it is an unconomical and borederline daft theory) you could avoid any RCD protection completely by using double insulation as the protective measure for the circuit instead of ADS, or even electrical separation. Or of course by getting the Ra low enough to not require RCD

 

[Leesparkykent]

Just in city's at the mo and flicking through the PE magazine and napit have just written an article on this and says the rcd must be located at the inverter end. Page 37 of Febs PE magazine if any ones interested.

 

[gnselectricals]

Just in city's at the mo and flicking through the PE magazine and napit have just written an article on this and says the rcd must be located at the inverter end. Page 37 of Febs PE magazine if any ones interested.

Off to the salers now will have a look cheers

 

[Worcester]

Just in city's at the mo and flicking through the PE magazine and napit have just written an article on this and says the rcd must be located at the inverter end. Page 37 of Febs PE magazine if any ones interested.

Now we're going to get even more confused lol

 

[GEOFF S]

But then the grid supply would not be disconnected in the event of a fault.
This is a cable with a source of supply at both ends.

Hi

Im not sure i follow, so lets break things down.

We must not connect to final circuits, so 5 Sec rule applys.

If there is no seperation on the inverter or the PV cable a.c requires rcd prorection then this is placed at the inverter.

Unlike a tx the inverter is not connected to the general mass of earth at the star point.

The PV a.c cable needs to be provided with fault protetcion at the connection to the mains a.c.

If a fault occurs on the a.c system, the power will disconnect within 5 secs, the loss of power will shut the inverter down in milliseconds.

If a fault on the PV cable occurs then the RCD at the inverter operates, thus causing loss of power and inverter shuting down.

If we look at ADS and earting the TX is earthed, so contact with live parts and general mass of earth will give rise to a risk of shock, this isnt the case with the inverter.


Cheers

 

[davesparks]

If a fault on the PV cable occurs then the RCD at the inverter operates, thus causing loss of power and inverter shuting down.

But the faulty cable is also connected to the grid supply, so yes the rcd at the inverter will disconnect the inverter but it will not clear the fault

 

[GEOFF S]

But the faulty cable is also connected to the grid supply, so yes the rcd at the inverter will disconnect the inverter but it will not clear the fault

Hi

But the cable will have protection at the other end as i mentioned, 712.434.1, so with the inverter isolated via the RCD, is it any different to any other circuit not needing rcd protection.

Now if we bury the cable less 50mm, and used standard cable, then yes two rcds at each end would be required, but this can be enginered out using swa etc.

Cheers

Cheers

 

[davesparks]

Hi

But the cable will have protection at the other end as i mentioned, 712.434.1, so with the inverter isolated via the RCD, is it any different to any other circuit not needing rcd protection.

Now if we bury the cable less 50mm, and used standard cable, then yes two rcds at each end would be required, but this can be enginered out using swa etc.

Cheers

Cheers

That was the whole point of the damned discussion, what can be done if the PV cable needs rcd protection!
That is exactly the situation I have been saying you would need two RCDs for.

 

[SibertSolar]

Fascinating...

 

[davesparks]

Fascinating...

It's the first really interesting discussion I've seen on here in ages!

 

[Leesparkykent]

I think the easiest way would be to use a reference method that doesn't require additional protection by means of an RCD ;-)

 

[davesparks]

I think the easiest way would be to use a reference method that doesn't require additional protection by means of an RCD ;-)

Yes, I completely agree, but the discussion has been interesting all the same. It's got me thinking about something properly for the first time in ages.

 

[Leesparkykent]

Yes, I completely agree, but the discussion has been interesting all the same. It's got me thinking about something properly for the first time in ages.

Its defiantly something I want to get to the bottom of as no doubt eventually will end up coming across the situation when surveying/doing a board change.

 

[Energetic]

How about approaching this from the other direction, are there any 3.6kW inverters that are compatible with a 30mA RCD? I think Power One used to do a HF transformered version, at a cost. Was there ever an SMA 3600HF? I once emailed them asking for one...

Seems a big hole in the EU market if there isn't one given the ever increasing requirements for RCD protection.

 

[CBR600]

How about approaching this from the other direction, are there any 3.6kW inverters that are compatible with a 30mA RCD? I think Power One used to do a HF transformered version, at a cost. Was there ever an SMA 3600HF? I once emailed them asking for one...

Ginlong Solis state max 20ma DC injection. We use them on 30ma RCD's without tripping issues.

 

[SibertSolar]

First thing, I'll reiterate what several others have already said - for transformerless inverters, always try to "design-out" or mitigate the requirement for an RCD in the first place.

All transformerless inverters will have a degree of inherent leakage current. I think that G83/2 requires that the tested/reported leakage current is no greater than 0.25% of the max AC current. For a 16A/3.68kW (G83 limit) inverter, this equates to 40mA, most decent units will rarely leak more than 5-10mA.

A 30mA RCD can trip from an AC leakage current value greater than 15.1mA (I stand to be corrected on that exact figure though).

As has been stated already, a TypeAC RCD's trip coil will saturate (and never trip under AC ground fault conditions) if any component of the leakage current is pure or pulsed DC. A TypeA RCD will still operate under AC ground fault conditions even if a pulsed DC component of the leakage current is present. It's trip coil will saturate if any pure DC component is present. However, if the amplitude of the pulsed DC leakage current is greater than 6mA then the Type-A RCD cannot be guaranted to operate within its certified scope of operation - this is an important point (can you or the inverter manufacturer guarantee that the 6mA limit will never be exceeded?). A TypeB RCD will operate under AC ground-fault conditions without being compromised by either pulsed or pure DC components of the leakage current.

Hypothetical (but also very real) scenario - TL inverter, requirement for RCD protection under BS 7671 (for whatever reason), TypeAC RCD installed. RCD never "nuisance" trips, just sits there looking pretty. Subsequent AC ground fault occurs (for whatever reason) but the RCD still doesn't trip. Too late to realise that the reason it never "nuisance tripped" is that the trip coil was saturated with DC and became about as useful as a chocolate fireguard. In that scenario, the end customer has been sold/supplied with a circuit protective device that will not operate as intended. It is worse than not fitting an RCD in the first place.

If there is no simple separation (ie, a galvanic transformer) between the DC and AC circuitry within the inverter, and an RCD is required on the AC output of the inverter (due to consideration of BS 7671) then a TypeB RCD is the required solution to provide ADS/ground-fault protection. However, many TL inverter manufacturers certify their product to VDE 100-712 in that it is, by design, unable to inject DC to ground. If the inverter manufacturer states this, then they typically go on to say that if an RCD is required then it need not be a TypeB (ie, a TypeA is ok to use).

There are some transformered inverters still around, although not many. The difference it makes is that, if an RCD is required on the AC circuit, then the inverter alone will not cause nuisance tripping of the RCD, and there is no scenario where a DC leakage current component might compromise the normal function of such an RCD, regardless of what type of RCD is fitted.

Completely agree with the various previous comments/advice about never sharing an existing RCD - a big no-no. "Nuisance tripping" and DC impact on RCD selection/function are two completely different considerations.

Just my tuppence...

 

[davesparks]

How about approaching this from the other direction, are there any 3.6kW inverters that are compatible with a 30mA RCD? I think Power One used to do a HF transformered version, at a cost. Was there ever an SMA 3600HF? I once emailed them asking for one...

Ginlong Solis state max 20ma DC injection. We use them on 30ma RCD's without tripping issues.

And that 20mA DC injection will saturate the RCD coil and prevent it from tripping in any situation if it is not a type B RCD.
So the most likely reason that you have experienced no RCD issues is that the inverter is actively preventing it from tripping, even under a >30mA AC fault.

 

[SibertSolar]

Exactly Davesparks....although, CBR600 didn't explicitly state that it was a TypeAC RCD fitted.....benefit of the doubt and all that...

 

[Deleted member 26818]

The literature for the ABB Trio inverters, state that they do not continuously inject fault current, so type B RCDs are not required.

 

[Gavin A]

All transformerless inverters will have a degree of inherent leakage current. I think that G83/2 requires that the tested/reported leakage current is no greater than 0.25% of the max AC current. For a 16A/3.68kW (G83 limit) inverter, this equates to 40mA, most decent units will rarely leak more than 5-10mA.

dunno like, I seem to quite regularly see 12-15mA leakage on the SMA 3600TLs particularly in winter, though I'm not entirely clear if this is leakage that would show on the AC side or DC side leakage via capacitance to earth, it's always much worse in cold damp conditions so probably the latter.

 

[davesparks]

The literature for the ABB Trio inverters, state that they do not continuously inject fault current, so type B RCDs are not required.

To my suspicious mind that suggests that they do/can inject non-continuous or occasional DC fault currents

 

[Deleted member 26818]

They may well do.
The literature recommends providing RCD protection for the output, and states the differential protection required is A/AC.