Unbalanced loads on a 3 phase RCD

[binky]

OK, so both criteria need to be applied.

I've been checking the G83 and G59 settings for SMA inverters loss of mains test. The standards have been altered, trip times are now 2.5s max, however SMA set their inverters for 1 second and according to the test sheets actually trip in 0.3seconds. So the answer is chose the right inverter.

One thing that has occurred to me, if I rodded the inverter and do not connect the AC side earth to the shared earthing for the system how does that affect the requirements of the above?

 

[Sidewinder]

I believe you must rely on specifications, not actual readings for your values.

You will still have an earth path will you not because both the ac side of the inverter and the installation will be connected to earth, i.e. true earth.

Is the SMA suitable for TT or only TN?

 

[binky]

Never seen an inverter specced as being for TT only, they can generally be used for any earthing system. About all the manuals say is 'install to local regs'.... Must be loads of 16th edition TT installations that are incorrect, but in  most domestic settings it isn't hard to get access to the main incoming point.

This site has an earth rod at the distribution boards located about 150m from the main incoming point, I've not checked to see if it connected to the earthing for the sub-main cable, I suspect it probably is, not that that would stop a back-feed to the main incomer should that cable get damaged.

I've attached a cert for a Samil inverter, seems these are set for 0.5s. I get your point about standards, but if I can demonstrate that a particular brand of inverter complies with requirements of BS7671, I don't see a problem, would need to mark up installation that this is the case / cert accordingly.

View attachment SolarRiver 1100TL-S-2600TL-S-Verification test report-G83-2 for UK_EN.pdf

 

[Sidewinder]

I don't think you will find an inverter that is TT only, but you will find them that are TN only.

 

[Rob_the_rich]

Surely if the inverter is referenced to true earth, then an s-type 100mA rcd on the output of the inverter will trip within 500ms, about the same time as the s-type 100mA RCD on the mains side tripping, from the same fault. If both RCDs trip according to standards, disconnection of both supplies should occur in less than a second. Or am I missing something?

 

[binky]

not necessarily as the RCd may not see a supply fault. I prefer to avoid RCDs when installing inverters, the relays tend to chatter on start-up and shut down which has a habit of upsetting RCDs. They are usually OK with a dedicated RCD, but any RCD shared is a mistake.

 

[binky]

I don't think you will find an inverter that is TT only, but you will find them that are TN only.

don't think I've seen TN only inverters ever either, what makes are you thinking of? Only major issue with regards inverters and DNO supplies that I can think of has been micro-inverters needing bonding to frames, and amorphous thin film panels which require earthed -ve on the dc side.

 

[Sidewinder]

Where is @Canoeboy when you need him!

I am pretty sure that I have seen instructions for solar inverters that state for use on TN supplies only.

Not doing solar, I only get to check the documentation side when requested for compliance with the regs.

 

[steptoe]

I'm fairly sure some SMA are unsuitable for TT, I'll have a look when I get home.

 

[binky]

you get off grid inverters, which are different, but I've been checking a few specs and all they say is connect to earth, nothing about the earthing characteristics. However, a high rod resistance ie over about 500 Ohms can cause inverters to see this as an earth fault.

 

[Rob_the_rich]

the RCd may not see a supply fault

I think I see why. On the mans supply, the neutral will be referenced to earth at the transformer. When the mains DP RCD trips, that reference is lost and the live conductors from the solar are only referenced to earth via the earthing at the inverter. This could be of higher resistance, therefore the current flowing to earth would be less, not enough to trip the PV RCD.

 

[binky]

not quite, the RCD at the inverter will not see an imbalance on the submain - leccy will just be off. But the inverter has a delay before auto-tripping out in the event of a power loss, anything up to 2.5seconds, so the inverter could possibly be still backfeeding the faulty submain and killing anyone connected to it.

 

[Rob_the_rich]

Oldish thread, sorry. But 411.3.2.6 says that if TT disconnection time of 1 second can't be achieved then supplementary bonding shall be provided according to 415.2

I still think that if the supply s-type 100mA RCD trips (the earth fault being downstream of it), then another s-type 100mA RCD on the output of the solar would also trip (they "see" the same fault current) very soon afterwards and both within 1 second.