Bonding multiple Deye 12kW inverters in Off-Grid Mode

[silverstone]

I managed to start up my system and do some preliminary testing with AC (I could quickly run an air-air heat pump as a load ~ 1.5 kW).

My setup is as follows:
- 3 Inverters connected in Parallel (Master 01, Slave 02, Slave 03)
- Inverters are all Deye SUN-12K-SG04LP3-EU
- 3ph+N Grid (380-400VAC Phase-Phase, 220-230VAC Phase-Neutral)
- Off Grid ("Load" / "Backup") only
- Grid Connection is left open - Couldn't find an authorized electrician to do the job so not much choice there

Each Inverter is protected by a 4P 32A MCB as well as a 300mA Type A RCD.

According to Deye Diagram, 3.10 should be applicable.
3.9 provided as a reference for a possible theory ...




This one looks like my setup:


Attached is my simplified schematics in PDF.

I EXPLICITELY BOND Neutral+Earth in my "AC Combiner Box" (it's essentially a combination of terminal blocks, contactors for switch-over to grid if Inverter fail, EMC filters) by putting a Jumper between thee Neutral Terminal Block and the Earth Terminal Block.

HOWEVER, I just had a realization that I might be doing it wrong: I am bonding AFTER (from the Inverter point of view) the MCB & RCD.

Therefore, if a Breaker/RCD trips or I manually switch it off, the Inverter will effectively be an IT system with Neutral NOT referenced to Earth.

Which is not the IT system that the manual refers to. In a normal European Electrical System, Phase-Phase Voltage is always 380-400 VAC.
For a TT or TN-S system, the Phase-Neutral Voltage is 220-240 VAC. For an IT system, the Phase-Phase Voltage will always be 380-400VAC, it's the Phase-Neutral Voltage which might, in a IT system, go up to 380-400 VAC !


The usual option advocated typically for Sunsynk hybrid inverters (use a contactor to connect Neutral & Earth at the Inverter side) are not really needed, since I never have the Grid connected. However, even if I were to do a permanent Earth-Neutral bonding at each inverter, it still would NOT work, because then I will have problems with ground loops (e.g. current from one inverter looping through the earth bar instead of coming back through its Neutral wire) due to having multiple Inverters.

A potential solution would be to keep the setup as-is, then add a neutral-earth "pull-down" resistor of say 100 Ohm.

Another solution (but more difficult to implement - no space, no alternative busbar on hand) is to use a feedback signal from each Breaker/RCD and based on that, close a Neutral-Earth contactor on the Inverter side. Still no place for a contactor box ...

But given the Schematic of 3.10 from Deye manual, it looks like it doesn't even need a Neutral-Earth bond . Why would they say "do not connect" in Schematic 3.9 (I assume TN-C-S supply) on the "Load" Port otherwise otherwise ? I assume in 3.9 it's forbidden because it would create a ground loop, should you connect the PE wire to the "Load" Port.

But then again my (very limited) testing of ~ 1.5kW (~ 6A) didn't cause any trips. That could be because the RCD combined allow up to 3x0.3A = 0.9A of unbalance but still ...

Does anybody else have similar experiences ? How did you solve it ?

I am NOT afraid of the protection of people & downstream equipment - That's what the bonding in the AC combiner box is for. What I'm afraid is the inverter getting damaged by having an unexpected IT system all of a sudden (if AC Breaker/RCD is open).

I actually could hear a small buzzing noise as I turned off an RCD (only once - didn't happen since then, knock on wood) ... Not sure if it was just current switching from one inverter to the other, some relay operating, ...

 

[silverstone]

I managed to start up my system and do some preliminary testing with AC (I could quickly run an air-air heat pump as a load ~ 1.5 kW).

My setup is as follows:
- 3 Inverters connected in Parallel (Master 01, Slave 02, Slave 03)
- Inverters are all Deye SUN-12K-SG04LP3-EU
- 3ph+N Grid (380-400VAC Phase-Phase, 220-230VAC Phase-Neutral)
- Off Grid ("Load" / "Backup") only
- Grid Connection is left open - Couldn't find an authorized electrician to do the job so not much choice there

Each Inverter is protected by a 4P 32A MCB as well as a 300mA Type A RCD.

According to Deye Diagram, 3.10 should be applicable.
3.9 provided as a reference for a possible theory ...
View attachment 172995



This one looks like my setup:
View attachment 172994

Attached is my simplified schematics in PDF.

I EXPLICITELY BOND Neutral+Earth in my "AC Combiner Box" (it's essentially a combination of terminal blocks, contactors for switch-over to grid if Inverter fail, EMC filters) by putting a Jumper between thee Neutral Terminal Block and the Earth Terminal Block.

HOWEVER, I just had a realization that I might be doing it wrong: I am bonding AFTER (from the Inverter point of view) the MCB & RCD.

Therefore, if a Breaker/RCD trips or I manually switch it off, the Inverter will effectively be an IT system with Neutral NOT referenced to Earth.

Which is not the IT system that the manual refers to. In a normal European Electrical System, Phase-Phase Voltage is always 380-400 VAC.
For a TT or TN-S system, the Phase-Neutral Voltage is 220-240 VAC. For an IT system, the Phase-Phase Voltage will always be 380-400VAC, it's the Phase-Neutral Voltage which might, in a IT system, go up to 380-400 VAC !


The usual option advocated typically for Sunsynk hybrid inverters (use a contactor to connect Neutral & Earth at the Inverter side) are not really needed, since I never have the Grid connected. However, even if I were to do a permanent Earth-Neutral bonding at each inverter, it still would NOT work, because then I will have problems with ground loops (e.g. current from one inverter looping through the earth bar instead of coming back through its Neutral wire) due to having multiple Inverters.

A potential solution would be to keep the setup as-is, then add a neutral-earth "pull-down" resistor of say 100 Ohm.

Another solution (but more difficult to implement - no space, no alternative busbar on hand) is to use a feedback signal from each Breaker/RCD and based on that, close a Neutral-Earth contactor on the Inverter side. Still no place for a contactor box ...

But given the Schematic of 3.10 from Deye manual, it looks like it doesn't even need a Neutral-Earth bond . Why would they say "do not connect" in Schematic 3.9 (I assume TN-C-S supply) on the "Load" Port otherwise otherwise ? I assume in 3.9 it's forbidden because it would create a ground loop, should you connect the PE wire to the "Load" Port.

But then again my (very limited) testing of ~ 1.5kW (~ 6A) didn't cause any trips. That could be because the RCD combined allow up to 3x0.3A = 0.9A of unbalance but still ...

Does anybody else have similar experiences ? How did you solve it ?

I am NOT afraid of the protection of people & downstream equipment - That's what the bonding in the AC combiner box is for. What I'm afraid is the inverter getting damaged by having an unexpected IT system all of a sudden (if AC Breaker/RCD is open).

I actually could hear a small buzzing noise as I turned off an RCD (only once - didn't happen since then, knock on wood) ... Not sure if it was just current switching from one inverter to the other, some relay operating, ...

Just saw the small note on Diagram 3.10: "When the inverter is working in backup mode, neutral and PE on the backup side are connected via the internal relay. Also,this internal relay will be open when the inverter is working in grid tied mode.*

But then I wonder about all of the fuss about Sunsynk inverters etc that need external Neutral-Earth bonding.

What is going on here ?

 

[houseofancients]

I managed to start up my system and do some preliminary testing with AC (I could quickly run an air-air heat pump as a load ~ 1.5 kW).

My setup is as follows:
- 3 Inverters connected in Parallel (Master 01, Slave 02, Slave 03)
- Inverters are all Deye SUN-12K-SG04LP3-EU
- 3ph+N Grid (380-400VAC Phase-Phase, 220-230VAC Phase-Neutral)
- Off Grid ("Load" / "Backup") only
- Grid Connection is left open - Couldn't find an authorized electrician to do the job so not much choice there

Each Inverter is protected by a 4P 32A MCB as well as a 300mA Type A RCD.

According to Deye Diagram, 3.10 should be applicable.
3.9 provided as a reference for a possible theory ...
View attachment 172995



This one looks like my setup:
View attachment 172994

Attached is my simplified schematics in PDF.

I EXPLICITELY BOND Neutral+Earth in my "AC Combiner Box" (it's essentially a combination of terminal blocks, contactors for switch-over to grid if Inverter fail, EMC filters) by putting a Jumper between thee Neutral Terminal Block and the Earth Terminal Block.

HOWEVER, I just had a realization that I might be doing it wrong: I am bonding AFTER (from the Inverter point of view) the MCB & RCD.

Therefore, if a Breaker/RCD trips or I manually switch it off, the Inverter will effectively be an IT system with Neutral NOT referenced to Earth.

Which is not the IT system that the manual refers to. In a normal European Electrical System, Phase-Phase Voltage is always 380-400 VAC.
For a TT or TN-S system, the Phase-Neutral Voltage is 220-240 VAC. For an IT system, the Phase-Phase Voltage will always be 380-400VAC, it's the Phase-Neutral Voltage which might, in a IT system, go up to 380-400 VAC !


The usual option advocated typically for Sunsynk hybrid inverters (use a contactor to connect Neutral & Earth at the Inverter side) are not really needed, since I never have the Grid connected. However, even if I were to do a permanent Earth-Neutral bonding at each inverter, it still would NOT work, because then I will have problems with ground loops (e.g. current from one inverter looping through the earth bar instead of coming back through its Neutral wire) due to having multiple Inverters.

A potential solution would be to keep the setup as-is, then add a neutral-earth "pull-down" resistor of say 100 Ohm.

Another solution (but more difficult to implement - no space, no alternative busbar on hand) is to use a feedback signal from each Breaker/RCD and based on that, close a Neutral-Earth contactor on the Inverter side. Still no place for a contactor box ...

But given the Schematic of 3.10 from Deye manual, it looks like it doesn't even need a Neutral-Earth bond . Why would they say "do not connect" in Schematic 3.9 (I assume TN-C-S supply) on the "Load" Port otherwise otherwise ? I assume in 3.9 it's forbidden because it would create a ground loop, should you connect the PE wire to the "Load" Port.

But then again my (very limited) testing of ~ 1.5kW (~ 6A) didn't cause any trips. That could be because the RCD combined allow up to 3x0.3A = 0.9A of unbalance but still ...

Does anybody else have similar experiences ? How did you solve it ?

I am NOT afraid of the protection of people & downstream equipment - That's what the bonding in the AC combiner box is for. What I'm afraid is the inverter getting damaged by having an unexpected IT system all of a sudden (if AC Breaker/RCD is open).

I actually could hear a small buzzing noise as I turned off an RCD (only once - didn't happen since then, knock on wood) ... Not sure if it was just current switching from one inverter to the other, some relay operating, ...

why a permantent bond ?
these inverter have a "signal island" check box ( btw i have the same inverter abeid with a sunsynk label on it), with closes a relay internally on the master..
safer and better controlable

never seen your issues

permanent bonding is in general a bad idea

 

[LydMekk]

Agree with house.
I don't see the point in having an external bond with mine.

 

[silverstone]

why a permantent bond ?
these inverter have a "signal island" check box ( btw i have the same inverter abeid with a sunsynk label on it), with closes a relay internally on the master..
safer and better controlable

never seen your issues

permanent bonding is in general a bad idea

So I have to enable that option ? From the manual it would appear to be automatic ... I thought this was the same signal that is used to turn on a diesel generator, so it cannot fulfill both roles. And on Sunsynk Youtube, the signal is used to control an EXTERNAL contactor, not an internal bonding relay/contactor.

Well my RCD should have tripped then, if the relay was working correctly .

EDIT: why "Permanent" ? Because I will never have the grid connected to the Inverter, as explained in my first post

 

[houseofancients]

So I have to enable that option ? From the manual it would appear to be automatic ... I thought this was the same signal that is used to turn on a diesel generator, so it cannot fulfill both roles. And on Sunsynk Youtube, the signal is used to control an EXTERNAL contactor, not an internal bonding relay/contactor.

Well my RCD should have tripped then, if the relay was working correctly .

EDIT: why "Permanent" ? Because I will never have the grid connected to the Inverter, as explained in my first post

the 12K's and 8K 3 Phase have a internal contactor

And this is from a sunsynk inverter config page

 

[silverstone]

the 12K's and 8K 3 Phase have a internal contactor

And this is from a sunsynk inverter config page
View attachment 173000

Mmm alright... I guess I trusted more sunsynk videos on YouTube than the manual for my exact model.

BUT how are you dealing with multiple inverters? Are the slaves (automatically) bonded, thus potentially creating ground loops?

Or, on the other side, if only one inverter (e. G
The master) is doing the bonding, and if that inverter's AC breaker or RCD is open, then you lose ALL bonding?

 

[houseofancients]

Mmm alright... I guess I trusted more sunsynk videos on YouTube than the manual for my exact model.

BUT how are you dealing with multiple inverters? Are the slaves (automatically) bonded, thus potentially creating ground loops?

Or, on the other side, if only one inverter (e. G
The master) is doing the bonding, and if that inverter's AC breaker or RCD is open, then you lose ALL bonding?

actually..
i had to explicitly ask keith as i noticed something when playing around...

if you flip that switch , and you have a paralled system ( i have 2 in parallel ) , only the master contactor is set....
if there's no grid connection, it will always be enabled

 

[silverstone]

actually..
i had to explicitly ask keith as i noticed something when playing around...

if you flip that switch , and you have a paralled system ( i have 2 in parallel ) , only the master contactor is set....
if there's no grid connection, it will always be enabled

But again if the breaker / rcd on the load output is OPEN, you'd still lose your bond to the other inverters...

Same problems in my case with external bonding by the way ?.

I could bypass the neutral in the breaker, but I need to feed that through the rcd for sure...

But is the danger of lack of bonding:
- Touch voltage / cannot trip rcd
- Inverter gets damaged

When I tried to set to IT system (neutral not grounded) I got an ominous message saying that it moght damage inverter. Therefore I assumed that all inverters no (longer) bonded get damaged.

Granted IT system for Deye is somewhat different than a normal It system (at least according to the manual)... But..
Is the warning just a case of false alarm?

 

[silverstone]

But again if the breaker / rcd on the load output is OPEN, you'd still lose your bond to the other inverters...

Same problems in my case with external bonding by the way ?.

I could bypass the neutral in the breaker, but I need to feed that through the rcd for sure...

But is the danger of lack of bonding:
- Touch voltage / cannot trip rcd
- Inverter gets damaged

When I tried to set to IT system (neutral not grounded) I got an ominous message saying that it moght damage inverter. Therefore I assumed that all inverters no (longer) bonded get damaged.

Granted IT system for Deye is somewhat different than a normal It system (at least according to the manual)... But..
Is the warning just a case of false alarm?

Actually I'm wondering... Maybe I just completely bypassed the rcd since neutral and earth are bonded on the LOAD side, and not on the SOURCE side (inverter in off grid mode).

It's still bonded alright but pretty useless if there is a phase-earth fault on the load side, the rcd won't trip, and the inverter will stay connected.

Or did I misanalyse wrong (again) ??

 

[timselectric]

The bonding needs to be before the RCD, for it to function properly.
If the RCD is tripped, there's no power, so no safety risk.

 

[silverstone]

The bonding needs to be before the RCD, for it to function properly.
If the RCD is tripped, there's no power, so no safety risk.

The RCD is not really for people protection. 300mA is mainly used for fire prevention or equipment-related damage prevention so to speak.

But I guess, either way, my currently used wiring is wrong .

 

[silverstone]

Alright so an update on the situation ....

I removed the PERMANENT BOND in the AC combiner box.

I checked "Signal Island Mode" on the Master. As soon as I did that, 2 out of 3 selective RCDs (A-Si 300mA) tripped.

I unckecked the Signal, and it "stays" unchecked, but when I try to turn the RCDs back on, they trip anyway.

So it seems like this was a "one way street". I feel some relay/contactor clicking also on the slaves, so my assumption is that they ALL try to bond neutral and earth.

But worse part was that even if I unplug the PE wire from the "LOAD" terminal on the slaves (see schematic 3.9/3.10 in my first post), it still trips.

I can keep the system running on one single inverter and it seems to be happy with that (Master Only connected, Slaves are disconnected through the RCDs), but I'd like to get them all up and running. I left the "Signal Island Mode" CHECKED for now ... Not sure it makes any difference though.

I also disabled Generator / Diesel Start Signal in the Battery settings, as that signal might be used for the same purpose.

I think it's still bonding PE and N on the slaves as well. But not sure how, since the PE wire was (temporarily) disconnected.

What the hell is going on ? SW Version is 2005-1124-1807 with HMI Version 1001-C031 if that helps ...

 

[timselectric]

Ok, now I see the problem.
You have multiple RCD's in parallel. This won't work.
All inverter outputs need to be combined. And then go through a single RCD.

 

[silverstone]

That's going to be a problem, because I have cables in parallel going to the combiner box... And they are not rated to withstand the FL combined current. The easiest I could do is remove The RCDs altogether... Running a 4x16mm2 cable is going to be challenging therwise...

It should work still, as long as there is only 1 bonding point...

It's not a good practice if all inverters bond neutral and earth, as you could have part of the normal neutral current that is now flowing through the earth. That's just wrong...

 

[timselectric]

And they are not rated to withstand the FL combined current.

They don't have to be, if they are under 10' in length.
Or separately protected at or below their rating.
The 10' tap rule only requires that any tap must be able to carry a minimum of 10% of the total current.
You should be much larger than that.

 

[silverstone]

They don't have to be, if they are under 10' in length.
Or separately protected at or below their rating.
The 10' tap rule only requires that any tap must be able to carry a minimum of 10% of the total current.
You should be much larger than that.

Different rules here in Europe I guess .

Either way, I could put the RCD AFTER the EMC filter (AFTER the AC Combiner Box), but that will provide NO PROTECTION in case of faults there. Otherwise, all downstream stuff will be protected as good as possible.

 

[silverstone]

They don't have to be, if they are under 10' in length.
Or separately protected at or below their rating.
The 10' tap rule only requires that any tap must be able to carry a minimum of 10% of the total current.
You should be much larger than that.

Small update after some further investigation ...

I tried to leave a PERMANENT BOND on the Inverter side of the Master RCD, while disconnecting the LOAD/BACKUP + GEN PE Bonding. Still trips.

Then I remove the PERMANENT BOND on the Inverter side of the Master RCD and try again on the Load side of the RCD. Still trips.

After disconnecting EVERYTHING (all inverters disconnected via AC Circuit Breaker and RCD as well), I measure the Neutral-Earth voltage: < 0.2 VAC. It's bonded. On EVERY SINGLE DAMN INVERTER. Even if I disabled the "Island Mode Signal" and could hear something clicking on all Inverters, as I did that.

The system ran quite happily with a PERMANENT BOND on the load side of the RCD (useless as I stated before). Now it refuses to work altogether. 2 out of 3 RCD trip and, if the Master is still able to supply the load, I get a F41 Parallel Setup Alarm. I guess, for whatever reason (why only AFTER I enabled this Signal Island Mode ???), the Inverters detect that there is a load unbalance on each inverter, then decide to trip out.

As you stated @timselectric the solution seems to be a common RCD. Or 3 x RCDs in Parallel.

My current configuration in the electrical Panel is as follows. Note that the 6 devices at the bottom take EXACTLY one ROW (24 Modules width), so the Busbar is very pratical in order to avoid jumper cables etc.



The quick & dirty fix could be to just use a very long busbar, possibly also add PERMANENT Bonding:



Since the contactors are probably not really used anyway (I would get F41 Parallel Error if 2 Slave Inverters are disconnected while the Master is connected), probably the proper solution (besides pulling a 16mm2 cable) would be to replace them with a 32A Breaker each. And get rid of 2 x RCDs using a shorter busbar:



Any thoughts ?

 

[timselectric]

As you stated @timselectric the solution seems to be a common RCD. Or 3 x RCDs in Parallel.

Common RCD.
I don't see parallel RCD's working properly.
The current will never split perfectly across them.

 

[timselectric]

If the inverters are creating the bond.
Don't add a permanent bond.

 

[silverstone]

Common RCD.
I don't see parallel RCD's working properly.
The current will never split perfectly across them.

Yeah except "parallel" in the second schematic would mean exactly that. Potentially multiplying the 300mA threshold. Is completely different from the first schematic where the RCDs are essentially sitting BETWEEN inverters (Series-Connected).

In the second schematic, either I have 300mA or up to 900mA RCD effectively. A bit high but still ...

 

[silverstone]

If the inverters are creating the bond.
Don't add a permanent bond.

I still don't understand WHY they are. And WHY ALL THREE OF THEM. The "Island Mode Signal" has been UNCHECKED. So either way, there will be some current flowing between Earth and Neutral cables, between the different Inverters ...

 

[timselectric]

I still don't understand WHY they are. And WHY ALL THREE OF THEM. The "Island Mode Signal" has been UNCHECKED. So either way, there will be some current flowing between Earth and Neutral cables, between the different Inverters ...

I wouldn't expect much (if any) current flowing between them. I wouldn't be concerned about it.

 

[silverstone]

I wouldn't expect much (if any) current flowing between them. I wouldn't be concerned about it.

Well if they are all doing Neutral-Earth bonding as I suppose they are, the Earth wire could pickup about 75-85% of the Neutral current. Because there are like 4x6mm2 PE wire on each inverter and only 1x6mm2 for Neutral (same for each Phase).

It will NOT overload the PE wire (since it has the same cross-section as the Neutral wire), but it's definitively NOT a good practice. The PE wire is only supposed to carry Fault Current and - to some extent - EMC Current.

You could argue that me choosing multiple RCD (1 per Inverter) was a bad design decision. Besides the multiple Neutral-Earth bonding issue, the EMC currents between Inverters might also be high enough to disturb the RCD.

 

[timselectric]

Why so many ground (PE) wires?
Only one is required per piece of equipment.
It sounds like you have a lot of ground loops in your system.