Wacky Problem, 36kW 6 SMA Inverters Ground Fault Issue

[zenmurphy]

Hi people. I'm new to this forum and just looking for a place to gain some insightful knowledge as through my career I've surpassed most of my sources for knowledge and experience.

So I'm just over 1.5 MW installed and I've run into my first problem where I'm truly stumped. Ground Fault experts please read on!

I have six SMA Inverters (despite 1.5MW this is my first time using SMA) powering 152 Sharp 240watt modules on the roof of a 10 story high rise. 108 of those are on a large rack and power 4 of the inverters. Their are 4 combiner boxes on the top of the rack easily accessible by a pent house roof. My equipment ground runs across the tops of my rails and split bolts into each combiner box, leaving the equipment ground headed back to the inverters and the system ground in 4 places. the Two combiners on the left and right side feed two SMA 5000US inverters (2X11 strings) and the two middle combiner boxes feed two SMA 7000US inverters (3x10 strings).

So yesterday during commissioning, I booted up the two SMA 5000US inverters and one of the 7000's and they all operated normally. The 2nd SMA 7000US blew it's ground fault fuse almost immediately upon turning it on. Going up to the combiner boxes, at first I registered two strings that wouldn't show voltage potential from positive to negative. We fixed what we thought were bad MC-4 connector terminations and the strings came up reading from Positive to Negative. However, NONE of the six strings total in those two middle combiners will register any voltage to ground, period, and one of those combiners feeds an inverter that maintained properly operating conditions for the 90 minutes it was on. The ground is clearly not open in both combiners, as it is connected both to the system ground through the inverter and also to the system ground through the solid #8 running across the rails and into the other combiners.

I am stumped, and the only other ground fault issue I've had in the past I was able to fix by the standard test the voltage to ground and find the fault in the series string... difficult to do when they array is saying six strings have no potential to ground!

Any information anyone can provide would be a lifesaver. This system has a scheduled inspection of Monday the 8th and this inspector is 4 weeks out for inspection rescheduling.

Thanks in Advance,

Matt Murphy
NABCEP Certified PV Designer
Peck Solar - A Division of Peck Electric
South Burlington, VT

[zenmurphy]

Additional Notes

Oh, just so it is not confusing. The remaining two inverters I mentioned are on racks located on top of the penthouse above the main rack. I didn't include them in the problem because they are on a separate equipment ground because they are on different racks, so those two inverters are completely separate systems.

[Sunking]

I didn't include them in the problem because they are on a separate equipment ground because they are on different racks, so those two inverters are completely separate systems.

Matt this statement catches my attention as what you describe is a clear code violation and might be causing your problems.

Every electrical system in the facility must be bonded to a COMMON GROUND ELECTRODE SYSTEM aka GES as defined by NEC . You cannot have one system electrically isolated from the other using separate ground electrode systems. They must share a common ground electrode system.

Moving on; GFCI circuit protection does not need an equipment or earth ground to operate properly, as that is what it is designed to do. GFCI is a ZERO SEQUENCE circuit protection device. What that means is it monitors the current in ungrounded and grounded circuit conductors (In this case positive and negative polarity circuit conductors) for any imbalance in load currents. In other words the current in positive circuit conductor must be equal to the current in the negative circuit conductor . If not, you do not have ZERO SEQUENCE and a fault exist so it trips. You do not want any current flowing on any ground conductors. It is dangerous and clear code violations.

That tells me you more than likely have your negative circuit conductor bonded to ground in more than 1 place. This diverts (parallel circuit) some of the load current on the negative conductor onto the ground conductor around the GFCI device resulting in unbalanced or ZERO SEQUENCE current.

[zenmurphy]

Re: Code Violation

Thanks for the response.

I think in trying to clear this up I made things unclear. They are on the same grounding system. They are on separate equipment grounding conductors. There are three racks, one holds 104 modules, and there are 24 each on the two upper racks. The building steel, module racking, and modules are all bonded, the equipment grounds go through the combiners, down to the inverters, and into the system. They are not separate systems, but they are separate PV systems as they are different arrays located in different places. All inverters are bonded to the Common Ground Electrode in the subpanel I installed in the penthouse, where it then feeds into the 1200 Amp switchgear that runs the building. So the ground on the main racking runs across the top of the rack and split bolts into each combiner. The grounds on the upper two racks run straight into their respective combiners. So the 4 lower combiners have the grounding conductor connected together before the AC side of the system, where as the upper racks/combiners/inverters don't join together with that system until the AC side.

I am in agreement that somewhere the grounding conductor and the grounded conductor are bonded, but I'd be damned if it was in an obvious place. I will conduct some more testing in the morning but I wanted some feedback.

I mentioned the condition of measuring the potential to ground in those combiners as 0 volts because it is very anomalous and I would love some brainstorming on why that could possibly be, especially when one of those combiners is feeding an inverter that was operating normally, and double especially because the two combiners that are attached to that same ground wire running across the big rack are registering potential to ground properly.

As we get busier and busier and I am running around between job sites more and more these problems tend to come up. All fixable, but ridiculous. I only really have 1 foremen that I can trust to run a commercial job of any scale. All of my other installers are extremely well trained electricians who have undergone PV training but the greater details tend to elude them. We also have a weird phenomena around here where most electricians are mystified by DC.

Thanks again, and good looking on pointing out the code violation. I am fixing 6-12 pv systems a year riddled with them. The last one was a 240V system non-operational for six months because the installers didn't include a neutral wire... they ran a 240v system back on 10-2 romex.

[Sunking]

I am in agreement that somewhere the grounding conductor and the grounded conductor are bonded, but I'd be damned if it was in an obvious place. I will conduct some more testing in the morning but I wanted some feedback.

That is the only way for it to happen. The grounded circuit conductor has to be bonded once, but only once just like your service neutral. It cannot be bonded again or else you will get load current flowing on the equipment grounds. I would look at the negative polarity conductors real close and start using an Ohm meter to find a unplanned ground. On a system this large it is real possible you nicked the insulation or pinched a conductor coming into contact with building steel or raceway. Isolate the conductors by disconnecting, then use a MEGGAR to find the fault.

Good Luck and let us know what you find. Happy hunting.

[zenmurphy]

Sometimes you can't find good help...

So on Friday I was talking with SMA tech support (I have SMA combiners, disconnects, and inverters) for 3 hours before they told me good luck.

At about 1:30 with my best installer becoming really frustrated and meggar tests giving perfect readings, I remember a couple weeks back I checked some MC-4 connections, was displeased by the work done, and told the Master Electrician on the site to go and check all of them.

So I undid a negative homerun and as soon as the plastic was off the metal end of the connector fell off with no effort. We checked all six homeruns in those two combiners, we found connections not even crimped, just slid on, we also found connections so weak the wire that was supposed to be crimped looked like it was stripped but still from the factory (no strand flattening, no indent). 12 of 12 MC-4 connectors looking terrible.

We checked the remaining 16 connections in the system and had to replace 14 more for a grand total of 26 of 28 bad MC-4 crimps. Thankfully I had figured this out now, as I would have been going back to that site over and over as ground faults popped up.

When we fixed them the combiner boxes were reading proper voltage to ground and we were all set.

Master Electrician with 40 years experience can't crimp a flipping wire. Unacceptable. Fortunately I had already laid him off because he took about 80 hours too long on a 300 hour job.

Thanks for your suggestions.

[Sunking]

Were the MC4 connectors factory made or field terminated?