5kva MPII ground relay test failure during LS, not when syncing

Inverters
101

And obviously, all works perfectly, as I cannot test if the changeover was the problem because I need an 8/10 schedule.

image

This comment on another thread …

… came within a hair’s breadth of buying a 2nd 5kva MPII to eliminate the current inverter as the cause. :slight_smile:

Justified it further as the 2nd one is then the backup and have 8kw of power the rest of the time … till I sat myself down, slowly explaining to myself that having 2 inverters connected at the same time to the same DB, is not a “backup” with LS woes.

My cousin had 3 x 3kva’s, not only 3 times the power, but also 3 times the backup … till Eskom LS left him with one kinda crawling along, the bigger the inverter, the bigger the loads become. Learn from that I said.

Still want an older gen MLT as a backup.

102

Changeover switch stripped, the numbers indicating things that looked “off”.

Have maybe changed it over 5-8 times total since it was installed a long time back.

No1, definitely a melt there.
2-3 some residue.
4 looks discolored, maybe it is supposed to look like that some residue on them though.
5-8 some marks on some of the contacts.

image
image997×871 113 KB

Cannot rule out the changeover being the cause … yet.

But dang, the inverter made some weird noise when I tried to change it over. Just really chaffed that I found this problem accidentally.

103

It was probably shorting the input and output together on one of the lines, hence the groaning. Which is a testament to how robust these things are…

Edit: It could also just be that the force applied caused one of the contacts to open a little, causing arcing, and the intermittent connection would naturally change the tone of the inverter.

In my house both Multis starts droning a little when someone starts a hairdryer on half-speed…

104

Let me just state, I’m seriously impressed with Victron’s error notifications, the easily accessible knowledge-based and experienced people one can contact easily.

As previously noted, I’ve remembered when I’ve seen this issue the first time. Forgot about it completely, till the sparkie reminded me on Tursday.

Months ago, out of the blue, my wife had to wake me, was that early (read, I was nowhere near the stuff) because the house was off, UPS’es beeping something awful, even switching off with errors displayed.

I blamed the dang UPS’s … until I tried to switch the critical loads back to the main DB … and all I got was sparks inside the changeover, saw the fash through the box.

So I opened it … and one of the red wires was loose. Reconnected it, and it was working again, until last week. Even did a changeover once or twice.

I planned on replacing it …

I HOPE it was the cause now, it was fooked, but I will not bet on it … yet.

105

One LS after another, nary a problem.

We don’t have 8 am schedules this week, so far.

106

It does feel like you’ve found the problem: the MultiPlus tests it’s ground relay, connecting ground & neutral, and assumes it’s broken if there’s a link when there shouldn’t be (and vice versa).

But you’ve got a known-bad changeover switch, which is probably bridging your pre- and post- inverter neutrals intermittently. Probably due to ambient heat, which rises when your MPPTs are doing their work in that closet of yours. Since earth stays connected, and your grid neutral is connected to earth (somewhere), that link is probably enough to trip the MP detection.

Outside of loadshedding none of this matters since the inverter is itself “bridging” the neutrals – you’re on-grid after all.

Only during LS would it matter, and only if the changeover is making contact where it shouldn’t.

107

“Closet” is open at the bottom and top. On top of that, at 8 am that part of the house is quite cool.
4 pm, it would be hotter, and it never happens then with LS.

I SO wish you are right!

The one part I would then like explained … why only 8 am schedule, and not 4 pm, or evenings? Closet heat is not a factor. And I have tried, at 4pm schedule, to run the MPPT’s hard, as a atter of fact, all LS events after 12, that is when I can 5kw on average out of the panels, recharging the batts during LS times, 8 am, lowest MPPT production due to roof being N/W.

Until I have a 8 am schedule again, I cannot say the problem is solved since the changeover was bypassed.

And IF the problem is still there, the surge protection in the combiner will be disconnected.
And IF that still does not solve the problem, ideally would want to try a MPII 5kva loan unit at 8 am LS event.
If that one works, then I will say the inverter was faulty.

For I would bet my left nut, put this thing on a testbench, and it will work perfectly, just like when I have switched off the main DB, and the breaker to the inverter, a few times, to test as people suggested. Worked perfectly every time I did it.

As a matter of fact, it is 8:45 so within the time frame the error happen, just switched off the DB. Let’s test that now.

108

@mariusm , your brainstorm made me try another angle.

It is not the changeover …

image

I thought, it is within the 8am window, so switch off the DB … I can now replicate the error on demand!!!

YES!!!

Going to remove the surge protection in the combiner box, and switch the DB off again.

109

You might be dealing with a temperature gradient problem somewhere, instead of high temperature specifically: heating up from 08:00 as opposed to being hot at a later time.

Ai, good luck :slight_smile:

1 Like
110

Could be a dry joint, or more specifically, a relay-PCB solder joint that had it’s solder blasted out by a high current fault.

1 Like
111

Nope, not that either.

image

And that would be inside the inverter, yes?

EDIT: My next test, to make sure I can replicate the error on demand, will do a few DB switch off’s to make sure I can recreate it only in the 8 am slot.

112

Yes, but I think it is very unlikely to be the problem. A loose connection on the wiring terminals at the inverter on elsewhere in the installation is much more likely at this point. (or a defective relay)

1 Like
113

I will check all the connections again. Already did everyone on the battery, all was spot on.

Any suggestions where?

EDIT: Just counted, since the last error, we have 8 x LS, not one problem. Not one was at 8am though.

114

What would I test, do, to find that?

115

OK, I have an idea. We suspect the bonding relay is bad, or there is a dry joint.

The idea is to install a TN bond of your own on the output of the Multi, but make it a manually switched joint. EG, use an old breaker you have lying around.

Then first switch off the grid to get the Multi into islanded mode.

Then close the TN-breaker you installed.

The breaker now does the job the relay should be doing.

The bonding-relay check should detect no voltage across TN now… since you have explicitly shorted it out yourself. See if the error returns or not…

Before turning the grid back on, open the TN-bond breaker you installed. Otherwise the relay test will fail.

Disclaimers:

  1. This is for testing purposes only! Don’t make it permanent.
  2. Newer Multis test their bonding relay with an auxiliary contact on the relay, and this test won’t work for them. I have no idea where new starts and old ends.

Looking for the opinions of others on this…

116

Waiting for replies …

In the meantime, what must be done, in laymen baby terms:
Connect a 2nd wire to the AC_Out1 N inside the inverter.
Connect a wire to the DB before the RCD.
Connect the two wires with a circuit breaker, in the off position.

First, make sure the error is happening, DB is then off, and wait for the error.
If the error happens, wait for the inverter to recover, then flick the breaker on, and wait again.
DB stays off all the time.

End of test: Flick the breaker off, then switch DB on.

Only part that I cannot envisage: Connect a wire to the DB before the RCD.

117

I’d simply put the switch between the earth and neutral connections on the output terminal block of the Multi (with the existing wiring) and just let it hang out the bottom… it’s a test setup in the end.

The rest is accurate enough. You’re “helping” the bonding relay with an alternative path. If the error goes away, but returns without the help… it points to a bad relay.

118

MP_GrndRlay_HW_Change
MP_GrndRlay_HW_Change882×356 71.1 KB

Random thought on AM vs PM issue… maybe temperature related as @mariusm pointed at but opposite direction… cooler panels vs warmer panels.

1 Like
119

Not discounting one idea, what throws me a wee bit out, is the 4am and 6am … 2pm, 4pm and 6pm, 8pm slots that work, somewhere in there a temp-related issue must match, not so?

120

Link please?

What throw this error into the Murphy level, nary a think on Google I can find with that parameter, is the fact it only happens when there is LS.

Every relay error mentioned goes back to syncing when Eksom connects, wiring etc.

What further confounds the matter, it is only the morning 8am slot.

And then to add insult to injury, if I switch off the panels at the combiner box, during said LS schedule, no error.

And as some have suggested, damp, wet, early mornings … it has rained a few times.