Plug in solar and GFIs

[pv_n00b]

UL has published a good white paper on this subject, Safety Considerations for Plug-In Photovoltaic (PIPV) Systems. I recommend reading it to understand some of the issues. It points out the problems with GFCI circuits.

Safety Considerations for Plug-In Photovoltaic (PIPV) Systems

Understand the unique risks of plug-in photovoltaic (PIPV) systems and key safety considerations for residential use in this white paper.

www.ul.com

 

[wwhitney]

It points out the problems with GFCI circuits.

The most interesting point to me was that GFCI receptacles disconnect both circuit conductors when they trip, while GFCI breakers don't disconnect the neutral conductor (although obviously they could have been designed to do so).

So if you have a ground fault with PIPV on the load side of a GFCI receptacle, when the receptacle trips, the neutral current path is cut off and the PIPV can no longer supply power to the ground fault. While if the branch circuit is protected by a GFCI breaker, the PIPV can continue to provide current to the ground fault until its anti-islanding feature detects the loss of the grid. The UL standard for that allows up to 2 seconds to disconnect.

Cheers, Wayne

 

[jim dungar]

The most interesting point to me was that GFCI receptacles disconnect both circuit conductors when they trip, while GFCI breakers don't disconnect the neutral conductor (although obviously they could have been designed to do so).

Probably because you cannot misfire a GFCI breaker by swapping Line and Load connection as you can with a feed thru receptacle.

Opening the receptacle neutral is a relatively recent requirement, considering GFCIs were introduced more then 50 years ago.

 

[kwired]

If we look at the UL 943 standard that tests GFCIs we see that they are not tested for operation with a power source plugged into the load side. They are tested for miswiring where a source is wired to the load side and they are supposed to shutdown. So GFCIs are not listed for back feed. I would not plug a power source into a GFCI in my house.

UL 943 | UL Standards & Engagement | UL Standard

UL 943 | UL Standards & Engagement | UL Standard | Edition 5 | Ground-Fault Circuit-Interrupters | Published Date: May 17, 2016 | ANSI Approved: August 25, 2025

www.shopulstandards.com

The Legrand GFCI receptacles will not reset if power is applied to load terminals. They will reset if power is applied to the line terminals.

I have done a little experimenting with them before, once it is reset it stays set even if power is lost and restored again. You could swap the input power while it is "set" to the load terminals and it will work fine. Will even trip when a ground fault condition is introduced but will not reset again until you have power on the line side terminals. Since they do function this way I presume they still meet listing requirements. IIRC they also disconnect the receptacle from both line and load terminals when they are tripped, which would still keep the receptacle from having voltage if feed is to load side terminals. I know older GFCI receptacles would still be functional while "tripped" if you had supply to the load terminals.

Have not done similar experimenting with other brands.

 

[kwired]

I'd bet a milkshake if there are two power sources a utility and small inverter then a cord and plug connected load on the same branch , with one source behind the GFCI and one in front, you would no longer have GFCI protection, as the second source could feed current in to a fault equally with the primary source. Easy experiment would be a plug in one of these inverters to GFCI duplex , replace the PV panel with a regulated DC bench supply to max out the inverter, place a 15k ohm 5 watt ceramic resistor from line to equipment ground on the other receptacle, GFCI wont see the entire load. If a branch circuit has more than one source the alternate source should be plugged into a single receptacle that is not 'down stream' of the GFCI. (Same issue with overload, you'd need a 12 AWG wire and 15A breaker with a max 5A inverter )

I'd say GFCI protection is compromised but not necessarily eliminated. Source and circuit impedances would factor in how much current flows from where. Plus since GFCI receptacles have neutral to ground detection features I would think that component would still cause a trip as that feature would inject a signal across line and load side and you would have a circuit through the utility source at very least to cause trip imbalance. The small inverter would have to essentially be unbonded or you would have tripping every time you plugged it in presuming premises side EGC is in good condition.

 

[wwhitney]

I'd say GFCI protection is compromised but not necessarily eliminated. Source and circuit impedances would factor in how much current flows from where.

As discussed previously, the current from the inverter to a live-ground fault contributes to imbalance in the GFCI's detection coil additively with the current from the utility. For a GFCI receptacle, there's no compromise to GFCI protection for such faults.

Cheers, Wayne

 

[tortuga]

Yeah Wayne is correct, I think the only issues are when its placed on a multi outlet branch circuit if its a 5.5 amp source added to the first receptacle in the chain of a 15A general purpose dwelling unit lighting and receptacle branch circuit then its effectively a 20.5 amp circuit.
I know there are probably some clever tricks that can be done to estimate the load on the branch circuit, so if it had some type of 'energy management system' it might be fine. Or if it was added to the last receptacle it might also be fine.
Besides as an AHJ there is really nothing you can enforce past the receptacle.

 

[Master Nater]

it seems you guys have a good grasp on the issue of double-ended feed on a circuit, and the potential for overloading said circuits. it looks like you call this phenomenon "breaker masking." has anyone brought up the issue of what i refer to as, reverse voltage drop? i really don't know what the technical term for this is, but that's what i have been calling it.

i discovered during my time running grid-tied solar arrays that when production is high, and usage is low, what is usually a voltage drop across conductors, still occurs, but in the opposite direction towards the grid. so the idea that "current is current, the direction doesn't matter," IS FALSE. for those who do not know what i'm talking about, let me explain.

in order to run power back to the grid and match the grid's voltage, the alternative production system must raise it's output production voltage, in order to have it's power drop as it goes across the conductors of the system on it's way back to the source of the sine wave and voltage it's trying to match. that would usually be the utility transformer. that transformer is usually around 250v at zero usage and zero voltage drop. i've seen output on inverters as high as 262v, trying to match the 250v that the utility transformer is producing at it's terminals. this means, the voltage applied to the entire service, and all of the devices plugged into it, is much higher than typically applied, and probably rated for.

just some food for thought...

 

[jaggedben]

We call it voltage rise. It can also be thought of as normal voltage drop in the other-than-normal direction when net power at a measured point is flowing from the grid-tied inverter instead of the grid. It is normal and expected and problems with that are extremely rare.

But also as far as GFCI goes, voltage rise doesn't matter. While the direction of current matters (at any instantaneous moment of the AC waveform), as Wayne alluded to in post #46, the direction of power shouldn't affect a GFCI that is engineered to trip on any current imbalance between line and neutral. Because it works out that whether a fault is fed by the grid-tied source or the grid, there is a current imbalance.

The issue with defeating overcurrent protection is a separate issue from GFCI. One way to deal with it might be to set a low current limit for plug-in solar and figure it's okay, kinda like the long standing "120% rule" in article 705. Another way is to require an energy management system to monitor the circuit, but then it's really no longer 'plug-in'.

 

[winnie]

'Current is current, direction doesn't matter' is correct with respect to the response of a transformer measuring residual current on a circuit.

But you are absolutely correct that a backfed circuit may display 'voltage rise' from the resistance between an inverter creating the backfeed and the utility source. The resistance of the circuit conductors does introduce a certain directionality into the scenario.

 

[pv_n00b]

The only way I see to make PIPV safe is to require a dedicated circuit with only one outlet that the PIPV is plugged into. If the circuit requires GFCI/AFCI then the protection device has to the listed as bidirectional. There is movement to add a bidirectional test to the GFCI standard and develop a new PIPV standard, but it can take years for this to hit the street and manufacturers to list products to it. The only way to enforce it would be to require the use of a different NEMA plug than the standard 5-15 or 5-20 so consumers can't take the product out of the box and plug it into a standard socket. But PIPV is being sold today with a standard plug and hopefully that will not kill someone or burn down a house while the codes and standards people spend years working this out.

 

[ramsy]

The only way I see to make PIPV safe is to require a dedicated circuit with only one outlet that the PIPV is plugged into.

Any kind of radio frequency telemetry may trip out xFCI, as signal noise propagates thru Panelboards.

If the circuit requires GFCI/AFCI then the protection device has to (be) listed as bidirectional. There is movement to add a bidirectional test to the GFCI standard and develop a new PIPV standard, but it can take years

Collecting the required signatures for legislated code amendment, is the KISS that prohibits AHJ adoption of xFCI junk.

 

[wwhitney]

The only way I see to make PIPV safe is to require a dedicated circuit with only one outlet that the PIPV is plugged into.

Seem to me it would be equally safe to ensure that the branch circuit is a single straight daisy chain from the circuit breaker (no branches) and that the PIPV is plugged into the last outlet.

Cheers, Wayne

 

[ggunn]

Seem to me it would be equally safe to ensure that the branch circuit is a single straight daisy chain from the circuit breaker (no branches) and that the PIPV is plugged into the last outlet.

Cheers, Wayne

Anything else connected to that circuit would have access to the maximum current from the breaker plus the current from the PV. Not good.

 

[wwhitney]

Anything else connected to that circuit would have access to the maximum current from the breaker plus the current from the PV. Not good.

Why does that matter? The breakers are just to protect the wiring, not the loads.

I guess there's the issue of the receptacles themselves. So maybe you'd need to use a 15A breaker and limit the PIPV to 5A, so the receptacle can't see more than 20A. Under those restrictions, then if the branch circuit is #12 Cu, you wouldn't even need to restrict the PIPV location. If the branch circuit is #14 Cu, you'd still need to restrict the PIPV to the end of a single daisy chain branch circuit.

Of course, I expect the PIPV advocates would say that while these restrictions are necessary for safety matching the status quo, the theoretical problems discussed here do not occur that much in practice, and so are an acceptable tradeoff for the "good" of PIPV. I don't yet have an opinion on that myself.

Cheers, Wayne

 

[ggunn]

Of course, I expect the PIPV advocates would say that while these restrictions are necessary for safety matching the status quo, the theoretical problems discussed here do not occur that much in practice, and so are an acceptable tradeoff for the "good" of PIPV.

Yeah, if it burns a house down, it's only one, right?

 

[analog8484]

Yeah, if it burns a house down, it's only one, right?

By that logic ... Christmas lights too?

 

[ggunn]

By that logic ... Christmas lights too?

No logic there, either.

 

[jaggedben]

The whole supposed benefit of PIPV is that you wouldn't have to hire an electrician or pull a permit. So proposing rules like those mentioned is pretty pointless. UL could limit the output, that's I feel the best that can be done. And while I suspect that limiting output to something like 4amps would eliminate danger in 99.99% of cases, if there's going to be 10,000s of these things sold...

 

[brycenesbitt]

has anyone brought up the issue of what i refer to as, reverse voltage drop?....
i've seen output on inverters as high as 262v, trying to match the 250v that the utility transformer is producing at it's terminals. this means, the voltage applied to the entire service, and all of the devices plugged into it, is much higher than typically applied, and probably rated for.

Yes.
The utility in my area even publishes data. For any line segment I can see the "solar hosting capacity" which
is some measure of the peak voltage on the line segment after some voodo math that can't be verified.

In short they publish how much solar a given line segment can take, before there's a ? percent risk of a +5% voltage event.