I had to remove ALL the surge protectors in our house before my Multigrid stopped tripping the main breaker on the Always on DB ... AFTER we fixed earth leakage problem on the house.

4 hours ago, plonkster said:

I read somewhere that the reason why we have 30mA RCDs is precisely because if they made the limit any lower people would stop using them due to all the nuisance tripping.

That may be one reason but i think it is far more scientific than that.

The reference curve for potentially lethal currents applicable to the current path from the left hand to both feet obtained from IEC 479-1 Figure 14 is shown below.

From this graph it is clear that at 30 mA the probability of ventricular fibrillation is 0%.

Currents > 30mA  can have serious consequences.

 

 

1 minute ago, Carl said:

From this graph it is clear that at 30 mA the probability of ventricular fibrillation is 0%.

I actually wrote that and noticed after hitting "Save" that it was no longer there. Oh well... how terribly fortuitous that the human body has a nice 30mA gap so we have some room to mess around 😛

 

On 2019/09/16 at 1:21 PM, Pipes said:

Hi guys,

I have an Axpert MKS 5k inverter with Pylontech batteries (US3000 x 2) and eight 330W panels. From the main supply, power goes through an earth leakage to the inverter input and the geyser. There is also a stove connected which is bypassing the EL, but that must have been done 20 years ago. The output of the inverter also goes via another earth leakage to the live trip switches (lights and plugs) and an isolated neutral bar. I am using a common earth.

The problem I have is that the secondary earth leakage trips randomly. It trips quite a lot when the inverter switches from utility power to solar/battery. I have never seen a trip when it switches from solar to utility at night. Recently it has also tripped when plugging in the treadmill and some other exercise machine. Again this

doesn't always happen.

This weekend I made sure I didn't mess up the neutral connections when splitting off the stove and geyser during installation and all looks good.

Any ideas what else I can try?

@Pipes

I went trough similar experience back when i just installed mine, tldr is that you would have to wire specifically outside your inverter to overcome your issue.

https://powerforum.co.za/topic/2880-axpert-earthing-issue/

 

12 hours ago, Padwan said:

I went trough similar experience back when i just installed mine, tldr is that you would have to wire specifically outside your inverter to overcome your issue.

Thanks @Padwan, but that article seems to have many theories, but no indication what the final fix was you've implemented. I really don't want to end up putting in a change over switch.

Someone earlier suggested adding the bonding relay accessory. I think in the past people have also said their issues improved by bonding earth and neutral on the inverter output. I have to weigh in on this again...

You really have to understand this bonding stuff and why we even do it. You can't have two bonds in the same system (it's against the regulations), so you cannot bond the output of the inverter UNLESS the system is purely off-grid.

In this case, the tripping happens during switchover. The bonding relay (if you have one) must be open while you're connected to the utility and must remain open until AFTER the utility is disconnected. I therefore doubt very much that this has anything to do with bonding.

Where bonding does play a role, is when the RCD on the OUTPUT of the inverter trips when the bonding relay closes (after disconnecting the utility).

Furthermore, bonding is not required for an RCD to work. Bonding is done for safety, to reference the entire system to ground (so you know the neutral is at the same potential and the live is at 230V, without bonding both lines could be thousands of volts above earth even if relative to each other they carry only 230V). Bonding does tend to make RCD action more reliable, but is rarely the reason for nuisance tripping.

An RCD works merely on imbalance. Six goats went in and only five came out the other side... there must be a hole in the fence somewhere. The trouble is that there can be momentary imbalances, for short periods of time things might go a little out of kilter, so-called transients, and that's what all this "immunity" is about, you want an RCD that gives you a little grace.

Edited September 19, 20196 yr by plonkster

8 hours ago, Pipes said:

article seems to have many theories, but no indication what the final fix was

I also have a Axpert but clone type, what I can add is if you don’t bond neutral and earth while you are drawing power from inverter in off grid mode, than your output from inverter has like a floating neutral. If you test with a earth leakage tester you will find it shows you have no earth. Also the chances are great that your earth leakage downstream won’t trip in event of earth fault because you don’t have a bonded neutral/earth. 

My Axpert is completely off grid and do not take power from grid so therefore I installed a bridge between neutral output and earth point. (I am not using the bonding relay but are considering installing it). When testing with earth leakage tester this shows all polarity and earthing is correct and earth leakage trips when it registers a leakage current.

Because I am off grid I can bond the inverter neutral and earth output because it is considered my supply and do not rely on municipal earth bonding point. I also have my own earth spike earth purely for my inverter although it is bonded to my main municipal earth for equipotential bonding reasons.

My Axpert use to trip my earth leakage intermittently when switching solar to grid opposite to @pipes but since bonding i have not experienced trip again. Axpert floating neutral have something to do with it the “details” I don’t know.

Only other concern is check those exercise machines maybe that has 50% chance of faults.

If all fails consider buying a Victron🤣🤣🤣

 

21 hours ago, Gerrie said:

Also the chances are great that your earth leakage downstream won’t trip in event of earth fault because you don’t have a bonded neutral/earth.

Exactly. Not having the bond is more likely to decrease the sensitivity (because it increases the overall resistance of the loop).

Uhh, no. Not having an earth-neutral bond doesn't decrease sensitivity, it stops the protection from working altogether. This is unsafe, and should never be allowed.

15 minutes ago, Stanley said:

Not having an earth-neutral bond doesn't decrease sensitivity, it stops the protection from working altogether

It could do that, yes, but "rendered useless" is technically covered under "decreased sensitivity". My argument here is that bonding something usually doesn't stop nuisance tripping. It leaves it the same... or makes it worse. And additionally, adding extra bonds without knowing what you're doing is not a good idea either.

I will however disagree with you a little bit. The protection isn't completely nullified. If the system is not bonded (eg IT system), a person touching one of the conductors would not receive a fatal shock, since the overall impedance of the loop is just too high and no real current is flowing. But if there is an earth fault anywhere (lowering that impedance), and a person touches the other conductor (thereby creating a second earth fault), an RCD still provides protection since it cares little about the bonding... it only cares that 30mA is finding an alternative path.

3 minutes ago, plonkster said:

But if there is an earth fault anywhere (lowering that impedance), and a person touches the other conductor (thereby creating a second earth fault), an RCD still provides protection since it cares little about the bonding... it only cares that 30mA is finding an alternative path.

This would only be true if the first fault were on the other side of the RCD. If the first fault were on the load side of the RCD and you touched the other conductor then the fault current wouldn't cause any difference in the Live and Neutral currents flowing through the RCD, so it won't trip.

2 minutes ago, Stanley said:

This would only be true if the first fault were on the other side of the RCD

Yeah, there are many corner cases in an IT system, that is for certain. I'm not an expert on this either. The German regulations apparently call for an RCD if you're feeding sockets, but to really make it effective it needs a lower rating (15mA), which isn't really readily available. And IT systems are rare, and the point of IT systems is high availability. It's a red herring. My point is, I doubt the issue has to do with bonding. Not directly.

What I have seen is that because the bonding is momentarily interrupted on the output during changeover, the transients that creates can throw RCDs. But that happens when you switch BACK to the grid (for the RCD before the inverter) or it affects the RCD after the inverter. In this case the RCD before the inverter trips when the grid is disconnected. That is unlikely to be because of bonding.

22 hours ago, Gerrie said:

If all fails consider buying a Victron🤣🤣🤣

Fully agree, doing that solved a lot of problems for me... 😬

On 2019/09/17 at 6:56 AM, Pipes said:

The hunt is on for a better RCD😀

This is quite a complex issue and looks like your hunt will be on for a while to solve your earth leakage tripping problem but this is what makes it fun.

The best is probably to do as the guys recommended.

1. Fit a high quality earth leakage like schneider or ABB but these won’t be cheap and tripping might still be there, a few $ down the line.

Than if that did not work.

2. Check those neutrals in your DB with a megger if not already done, and see if any neutral might give a reading to earth. You should get infinity or Mega ohm readings to be sure its clear. It might be a utility load causing the problem that is not part of your solar setup and when switching back to utility it trips. Also test with an earth leakage tester and see what the mA trip is. (Should be about 20mA-26mA range). I normally test with normal loads plugged in and than test with everything plugged out, that will give you a indication what is your average use leakage current). You might already be on say 12 mA leakage that only leaves you with about 8mA tolerance before a trip occurs.(Lots of Cellphone chargers plugged in can be culprits of earth leakage)

3. Bonding of the inverter output relay, I would rather not do as Plonkster also mentioned because your inverter are also feeding from grid where you probably already have a grid neutral earth bond. If you want to do the bonding thing than I think it would be better to purchase the correct relay for your inverter with the options setting in your inverter and it will bond the correct sequence. To know if your inverter bonds automatic, just test from neutral output to earth point and it should probably read <0.1V. Than you know that part is sorted.

hope this help in some way

 

 

Thanks for all the advise guys. I started with the option to first replace the ACDC earth leakage with a model from Schneider and have not experienced any trips since last week Thursday.

I did however check the Neutral to Earth voltage at the output of the inverter even before making this change and it was around 0.5V, so it looks like my inverter already does the bonding.

Type B RCCBs/RCDs are specified for use with transformerless Inverter installations due to the harmonics see attached Schneider guidance https://www.schneider-electric.com/en/download/document/998-2095-12-31-13AR0_EN/

16 hours ago, Peukert said:

Type B RCCBs/RCDs are specified for use with transformerless Inverter installations due to the harmonics

They are also insanely expensive so you'd be nuts to buy one unless you know you need it. A so-called type-F is cheaper (still very expensive), but according to at least the ABB documentation I read, the APR models (ABB's moniker for the transient resistant types) really has everything you'd need for residential AC protection.

To summarise, the pecking order is:

1. Type AC
2. Type A
3. Type A-with-high-immunity, ABB calls it APR, Hager calls it HI (high immunity), Gewiss calls it IR (impulse resistant).
4. Type F, where you have VSDs and other things with the potential higher frequency leakage currents.
5. Type-B, can also detect DC leakage currents.

You should not have to venture higher than level 3 in a normal residential setting. Your wallet won't really like it anyway 🙂

 

LVE do a type B 40A 30ma for 95€ Hagers is 1000€ (retail price 2000€) the others are about 350€ Retail price (1000€). Moeller have the best range.

50 minutes ago, Peukert said:

LVE do a type B 40A 30ma for 95€ Hagers is 1000€ (retail price 2000€) the others are about 350€ Retail price (1000€). Moeller have the best range.

All good and well if you don't pay retail price then! That was exactly my experience. Around 1000 Euro and up (16k ZAR). There are people in this country who would kill to make that sort of monthly income. The ABB RCD I bought in the end was around 250 Euro.

A type B RCCB is not less sensitive, it just affords additional protection in that it can also trip for DC earth faults.

Obviously, the standard CT differential arrangement of an AC current  in vs AC current out is defeated by DC current.

That is why you see two differential measurement circuits inscribed on the face of a type B as opposed to the single toroidal cct. of a type A.

DC earth faults are possible from the array side depending on the type of inverter.

If you're willing to use Aliexpress or Alibaba, you will get a Type B under 100 Euro.

In the past 24hrs I've been quoted:

type B rccb  40A 30mA 2P: $48/pc             40A 30mA   4P:  $50/pc

on Alibaba.

(Which I grant you is still 10x the price of a Type AC, A in the same market).

I also want to say that in my experience the vast majority of earth leakage trips are due to earth leakage faults. I'd spend my money on a Megger before I bought a new EL.

 

4 minutes ago, phil.g00 said:

A type B RCCB is not less sensitive, it just affords additional protection in that it can also trip for DC earth faults.

True, but usually they incorporate a bit of logic against nuisance tripping. These units have little electronic brains... not just the dumb-but-effective residual coil+amplifier type thing 🙂 And resistance against nuisance tripping is what we're after. My argument (be it right or wrong) is that type-F and type-B adds preciously little of that above the transient resistant type-A units, so it's usually not worth spending the extra dosh on them.

4 minutes ago, phil.g00 said:

I also want to say that in my experience the vast majority of earth leakage trips are due to earth leakage faults. I'd spend my money on a Megger before I bought a new EL.

In my case I bought an earth leakage clamp meter, noted that my standing loss was quite low, but during the switchover the peak measured quite high. Clearly I was dealing with some kind of transient 🙂

 

Edited September 25, 20196 yr by plonkster

@plonkster, Respectfully until you've meggered your circuits clear of earth, you won't convince me.

 

53 minutes ago, phil.g00 said:

until you've meggered your circuits clear of earth, you won't convince me.

I've thought a bit about what you said, and I think there is one case I have not thought of where an insulation test would be a good thing to do: Leakage from a neutral conductor to earth.

When neutral is bonded, the potential between the neutral conductor and earth is very low (less than 1V in most cases, and certainly below 30V), so even a weak insulation will pass very little current. As we have seen however, when that bond is temporarily removed, the potential between neutral and earth rises very quickly to about the halfway point, and a weak insulation somewhere will now pass significantly more current.

Again, I was fairly certain that this was not my problem. There was practically no difference in the measured leakage current whether bonded or not bonded. What I saw was definitely a transient that occurred precisely when the bond was re-established (the backfeed relay closed, pulling neutral and earth on the inverter output back together). My case was further strengthened in that I could see the transient change as I plugged appliances in/out.

Overall I am going to agree with you though. I too have found that in general, if an RCD trips... it's because there is a leakage somewhere. First rule that possibility out completely. Doing an insulation test is a good (and comparatively cheap) way to rule out a lot of problems. An earth leakage current clamp (again, expensive, so pay someone else) is also a good thing, if you have a standing loss of close to 15mA... then don't be surprised that you get nuisance trips: Find the leak!

In my house I actually had quite a substantial standing loss. It was close to 10mA. Again... I traced it to filters and surge arrestors. Even with everything unplugged I still have a few milliamps of leakage (maybe 2mA or so for the whole place). That's probably due to insulation 🙂

16 minutes ago, plonkster said:

I've thought a bit about what you said, and I think there is one case I have not thought of where an insulation test would be a good thing to do: Leakage from a neutral conductor to earth.

It still is. Your bond is before the earth leakage/RCD. Your insulation test are done with the RCD in an off position, or in some cases electricians even disconnect the Load side of the RCD before they test. That way you can do the complete test even with the bond still in place. Just ensure all the appliances are unplugged. 

52 minutes ago, Jaco de Jongh said:

That way you can do the complete test even with the bond still in place

Exactly. You can see if there is an insulation failure on the neutral side too (cause the bond is not there). Weak insulation will cause higher leakage from neutral to earth if the earth/neutral potential difference is higher, as it is for those few seconds while the transfer switch connects/disconnects. So an insulation test will help rule this out. In short, I now agree with Phil. Never really disagreed either... we're just talking about two issues that both affect this.