Reverse polarity when PV inverter switched on!?

[mrrkspark]

Hi all

Very strange situation this evening. I installed a PV system around a year ago and has been working fine.

Called out today as the generation meter had stopped increasing. Customers been away and only arrived back today.
Western Power have recently put up a new pole in the customers garden but not yet connected cables to it. New pole is touching cables feeding the old pole slightly.

When I close the AC isolator to the inverter, the polarity switches for the whole house!
Inverter was working when I arrived and generating, but no increase in generation meter reading. I assume it is because the power is being fed back through the neutrals of the meter now and not registering.

Somehow I feel this is related to the tn-c-s or cne. The consumer unit in the house was replaced by someone a few years ago, I am informed he was ex western power employee. He has installed a earth electrode and not used the TN-c-s available at the cut out.

The generation meter reading is 2645. And the inverter total generation reads 2665. Western Power installed the new pole 3 days ago, and we have had decent weather in Cornwall (for October) for the last 3 days. 3kwp system. My 4kwp system has generated around 10kwh per day for the last 3 days.

Any advice much appreciated

Sorry if this is a bit vague. Please ask if any more info required, I am really interested in this one!

PV system has been left isolated for now!

 

[UKMeterman]

Stick a meter between a true outside earth stake and the N at supply, if you get more than about 10 volts call the DNO emergency number.

 

[mrrkspark]

Stick a meter between a true outside earth stake and the N at supply, if you get more than about 10 volts call the DNO emergency number.

Thank you for your advice. Could this potential cause the reverse of polarity?

 

[UKMeterman]

If you are geting more than about 10 volts between N and TRUE E then do a quick visual check to make sure that the N is in the middle two terminals on the meter and into the nonfused side of the cutout, if all correct then the supply transformer has lost it's earth connection and there is a dangerous situation for the DNO to fix.

 

[mrrkspark]

If you are geting more than about 10 volts between N and TRUE E then do a quick visual check to make sure that the N is in the middle two terminals on the meter and into the nonfused side of the cutout, if all correct then the supply transformer has lost it's earth connection and there is a dangerous situation for the DNO to fix.

Ok I see, I did visually check the neutrals and their path earlier and all is ok. Will check the potential to the electrode tomorrow morning. Many thanks

Having done a bit of research and speaking to a friend, it could be possible that the inverter is trying to synchronise with the supply/faulted supply, therefore aiming a potential between unconventional points. I may also try disconnecting the cpc from the supply to the inverter for testing purposes and see how that affects the potential.

 

[UKMeterman]

NO do not disconnect ANY CPC and then apply a supply. Do the basic checks as I suggest, with a decent load on, ask the customer to boil a ketttle whilst you do the check I suggest. If the inverter made a mess of syncronising with the supply there would be a loud bang, blown fuses and so on.

Do the test I suggest to a seperate electrode several meters from the exisiting electrode and and several meters from the DNO pole.

 

[mrrkspark]

NO do not disconnect ANY CPC and then apply a supply. Do the basic checks as I suggest, with a decent load on, ask the customer to boil a ketttle whilst you do the check I suggest. If the inverter made a mess of syncronising with the supply there would be a loud bang, blown fuses and so on.

Do the test I suggest to a seperate electrode several meters from the exisiting electrode and and several meters from the DNO pole.

Ok cool, I know what you are saying. I will certainly do the checks you mention as you say.
I meant that if the potential between N to true earth is ok, then possibly proving it as a fault at the inverter by disconnecting the cpc for a test. From what you say the protective devices would have picked this up so I will not bother tomorrow and let you know how I get on.

Thanks again UKMeterman

 

[mrrkspark]

Right so very strange today. I did as you suggested and checked voltage to true earth and sure enough had 30 volts N to earth, only with the AC isolator switched on though!
Then found 217 volts between L and true earth, and 217 volts between CPC and true earth!
I was asked by a colleague out of interest, to disconnect the AC plug to the inverter and check again. Sure enough it was still there. Disconnected plug from cable and made safe conductors and still there. Stripped back cable, and found the line conductor insulation melted right through to the earth conductor!!
Why the hell this did not operate the MCB or RCD I do not know and am very worried.

Having dug deeper it could be that dc voltage to true earth is causing heat build up in the AC?!
Protective devices both failed to operate?

Massive coincidence that it happened the day Western Power were there!?

With the cable replaced there is no voltage between N and true earth, and all as it should be.
I have left it disconnected and will be returning to replace the protective devices and look further into a DC earth leakage possibility.

Sure I am still missing something though

 

[UKMeterman]

Ok, when you turn on the AC isolator you connect a capacitor between L and E within the inverter. There is a fault if you only got 217V between L and true earth, or 30V N to earth It sound like the neutral earth link is suspect at the substation, this is a dangerous situation, you should have understood this as an electrician and called WPD there and then....

If the N E link is missing at the substation then any earth fault will not have any current flowing and will cause the E on the installation rise to L potential, this would then make it appear that the polarity has swapped.

Do your customer (and anyone else on the same substation) a favour and call WPD NOW before there is a fire.

 

[mrrkspark]

So the fault was present when the inverter was disconnected completely. As I mentioned the fault was within the cable feeding the inverter, from the AC isolator. So when I replaced the faulty cable, the fault completely disappeared.
Now no voltage between N and true earth, or the cpc to true earth. This is now the case with the inverter connected, and disconnected.
I understand what you are saying, if the fault was still there after the cable was replaced, and the inverter reconnected it would have been the first thing I would have done.

 

[UKMeterman]

Sorry to be blunt. There are two faults, you have found and corrected the 2nd fault, the L to E short on the inverter side of the AC inverter. The Lto E short should have tripped a MCB or RCB, this did not occur. The installation is still UNSAFE until you find the cause and either rectify if on the customer side or pick up the phone and call out WPD if it is on the supplier side. The evidence is that it is on the supplier side and as a professional electrician you need to make that telephone call.

 

[Marvo]

^^As above, you still have a fault. Isolate the installation and test the incoming supply Ze, Zn etc.

 

[mrrkspark]

These were the first checks I did and all ok. Did them again after fault found and same again.
Called WPD this morning and they turned up within the hour. Said everything is fine and specifically the N-E link is fine.
Failed RCD, no trip on MCB as TT, 21 ohms Ze so low fault current. I hear the fail rate for RCDs is 50% within 6 months. Not sure about this though.
Fault on the inverter cable could have been done to a loose connection in the plug, or likely DC leakage from the PV.
Will be returning asap to do a thorough overhaul.

Thanks for your comments guys

 

[UKMeterman]

Ok, what you do need with a PV array on site is an RCD that can detect pulsating DC faults.

 

[SibertSolar]

What inverter was this? And what make/model/type of RCD was fitted? I know it's a couple of weeks since the thread, but I am curious and haven't been on here for a while, thanks.