EG4 18KPV - What not to do in parallel

[exmarkretired]

I do believe so, but maybe @canadianintruder can paint more color on this.

I think it's possible that he didn't understand that all cables that have an RJ45 connector are not equal.

This mistake resulted in the loss of thousands of dollars in appliances.

Thanks for sharing. Now the hard part. If anyone has the integrity to reimburse you. Reinforces my belief as to the cause of damage that occured in my residence.... I chose to eat the loss, as it wasn't worth the time already lost to submit an insurance claim and the vendor was not cooperative.... But I only lost hundreds, not thousands. Best of luck. Again thanks for sharing so someone else may not get hurt....

They will do as the automakers do. Just issue a tech bulletin and not a recall. Save them thousands by doing so....

 

[exmarkretired]

A bit premature there Mr. Eggo? It’s not their fault that a bug slipped through.

Normally I appreciate your insight and knowldge. I actually read what you write... This is more than a bug....Best case you could object to is that it's just a big bug ??? Like the doors that fall of Boeings ?? Meh.. just a bug

 

[Nobodybusiness]

So as many of you may know I posted a thread about my neighbor having an issue with his EG4 18k Inverters.

System Information:
60 Panels
2 EG4 18k in parallel
No Battery

Summary of the issue:
When the grid connection is removed, the inverters would begin to provide 400v to the mains input side of the inverters.
The system was professionally installed and all wiring was correct.
The system was reporting everything was fine. No faults or errors were reported. The inverters were being recognized as being in parallel.

Diagnosis:
EG4 got into the system and checked the configurations and found nothing wrong.
EG4 guided us to first check the parallel cable. (Thank you EG4)
We found that the installer had used a cross-over cable that was likely bought from a Box store.
We made a new cable up that followed the proper standards.

Results:
Systems worked as expected. All commisioning tested passed.

Notes:
The inverters likely should catch that a cross over cable was used and not go into parallel mode.
Also, use the correct pinout if you are buying or making an ethernet cable. Not doing so could be dangerous, until a patch is released you could potentially have this problem.

I would like to call out @Markus_EG4. After checking the config, his very first troubleshooting step revealed the issue.

We plugged the factory cable in, executed the same test and it did exactly what it was supposed to.

Wow that seems like a big miss on the QA.

No sure how a Comm cable could cause this issue but I’m not sure what information is passed between Master / Slave units.

Do you happen to know the pin outs of those connections?

 

[sunshine_eggo]

Do you happen to know the pin outs of those connections?

It's in the manual.

Oh wait... it's not...

 

[Nobodybusiness]

This is the first example I have seen of the 18KPV in parallel.

I wonder if this has been an issue in other parallel operations with this unit?

If they used a straight through cable they might not know it’s an issue.

 

[Brucey]

@EG4_Jared
@Markus_EG4
@Luxpower_Gilbert

Heads up. I submitted a complaint to UL for this issue. You guys need to get this shit fixed and hope you don't get sued.

Wonder if it's a problem with all the luxpower 12ks or just the eg4 version.

 

[DIYrich]

Yea, the communication cable shouldn't cause this type of problem. Especially if it is easily foreseen (using a crossover cable). The units shouldn't go into parallel (not recognize the Master/Slave configuration) without a proper communications cable.

 

[Quattrohead]

These things better ship with their own fucking straight thru cables.

They do come with the parallel cable. I don't know why the installer would use some random cable.
Yes the inverter needs to be more careful with making sure the paralleling is done properly before it starts inverting, definitely a firmware update coming for that one, but the installer is at least 50% to blame too.
But yet again I am mightily impressed with the hardware not blowing up in the most unusual of circumstances, these units are very, very impressive electrically.

 

[Quattrohead]

Heads up. I submitted a complaint to UL for this issue. You guys need to get this shit fixed and hope you don't get sued.

I hope you reported the installer too, they are a good 50% plus of this fuck up as well.

 

[utkvols]

I wonder if it creates some type of phase issue- I bet you the crossed pairs cause the units to align their phases rather than syncing the two legs so they are 180* off phase. Or the protocol is telling both units the voltage is low and to push it higher? This would seem like a hardware issue - I don't see how they could fix that unless they can detect what type of cable is being used. I wonder how many times this has happened to others?

 

[utkvols]

I hope you reported the installer too, they are a good 50% plus of this fuck up as well.

How would they know? Most have no idea patch cables exist - they're dinosaur 90s tech. If the manual doesn't specify a difference, they (nor anyone else - including apparently the manufacturer), that this could have happened.

 

[wpns]

Not assigning blame (the firmware developers from Elbonia deserve some credit here), but this is a "Should Never Happen" event. Don't care if you didn't think to test for it, don't care if your design reviews didn't turn it up, don't care if this is one for the Risks Digest, bad/miswired/wrong cables should never cause that kind of problem.
Yes, they come with cables, and yes it's hard to find crossover cables, but as someone who drove an hour to replace a phone cable with a network cable in a fire alarm system on an emergency basis, I'm not at all surprised.

 

[Cmiller]

Regardless of master/slave comms, "seeing" a parameter outside of grid interactive setpoints should have caused a complete "override" of all operation and a full disconnect! I understand "bugs" slipping through, but this has a few too many different aspects that should have been covered....

On a side note, as I promised to someone in the other thread, I checked how our Sol-Ark 12K reacts under a full clean disconnect from grid. It nearly instantly goes to zero volts on the grid side of the unit, which is what I expected. I took video of a few scenarios and I will edit this post to link a video once I get it put together and onto YouTube. (I tried a few scenarios, so have to put a few videos together into one, and I haven't ever done that. Wish me luck. )

P.S. If anyone thinks I should make a separate informational thread with that content, let me know.

*Link will be here once I get the video on YouTube.

 

[Brucey]

I do believe so, but maybe @canadianintruder can paint more color on this.

I think it's possible that he didn't understand that all cables that have an RJ45 connector are not equal.

This mistake resulted in the loss of thousands of dollars in appliances.

I guess the positive side of things is that you didn't lose your house in a fire after up to 570+ volts of AC were sent to your appliances.

 

[Quattrohead]

It is easy to see how this high voltage could happen and there is absolutely nothing the inverter could have done about it.
With the wrong communications cable there was no communications between the two inverters, fannying around with the master/slave and phase settings to try and get things working could have put the inverters in a situation where as soon as they lose their main synchronization to the grid their outputs are going to drift and go all over the place and could indeed get to the point where the output sine waves are the opposite and the voltage will be accumulative.
Look at the scope image posted, it is nowhere near any kind of sine wave but definitely shows them combining together to produce a much higher voltage.

 

[Brucey]

It is easy to see how this high voltage could happen and there is absolutely nothing the inverter could have done about it.
With the wrong communications cable there was no communications between the two inverters, fannying around with the master/slave and phase settings to try and get things working could have put the inverters in a situation where as soon as they lose their main synchronization to the grid their outputs are going to drift and go all over the place and could indeed get to the point where the output sine waves are the opposite and the voltage will be accumulative.
Look at the scope image posted, it is nowhere near any kind of sine wave but definitely shows them combining together to produce a much higher voltage.

Remember how instead of master/slave he had a different status code on one of the displays?

It seems pretty basic to me, if you try and pair two units, and the comms isn't working (cable disconnected, crossover cable, internal comms fault etc) neither of the units should power up.

 

[SparkyJJO]

Remember how instead of master/slave he had a different status code on one of the displays?

It seems pretty basic to me, if you try and pair two units, and the comms isn't working (cable disconnected, crossover cable, internal comms fault etc) neither of the units should power up.

Yes, BUT....

What happens if you take other inverters that are not supposed to be in parallel, and hook them up in parallel?
Normally we'd say they'll blow each other up. Spark city. Smoked. Fried.
But... what if they didn't blow up?

That's sort of what happened here. They operated independently of each other (sort of), but they're robust enough that instead of blowing each other out, they just stacked on top of each other.

I wouldn't be surprised if other inverters in standalone mode might result in something like this when stacked, if their electronics were tough enough to not just blow up internally.

Am I saying they should operate this way? No. They're supposed to be grid-interactive, and the very fact they did this with the wrong cable is a huge problem. Even in standalone mode, and say they were detecting each other's "grid" and trying to unsuccessfully sync to each other, and fighting each other, they should then go out of range and shut down, not blast out 400V+ in some non-sine wave distorted disaster.

Lux and EG4 are going to learn a lot from this. I think they're on the hook at least partially.
I think the installer is on the hook partially for using some random cable, assuming that all RJ45 ended cables are the same (he should know better). There are plenty of instances with RJ45 cables that just because they fit together doesn't mean they work together. Plug an rj45 cable with POE into RS232 or CAN - yeah it might fit but you're likely going to be sad when there's a poof and things don't work anymore.

Personally, I think much more liability is on Lux/EG4 because the units should never have done this. What would happen if the supplied standard cable was defective or got damaged? Serious issue. But the installer isn't totally off the hook.

Just my opinion, 2 cents, not adjusted for inflation.

 

[Brucey]

So I guess the follow up question, what happens when two paired 18kPVs that are using the correct non crossover Ethernet cable lose connectivity with each other? Do they shutdown?

 

[BuffaloTrace]

how many of these things can be stacked together? My thoughts is this puny 400-600v is nothing compared to what these babies can really do. Maybe add another 1,2,4 of these together in one system with a bunch or a special secret combination of cross over cables and we can have that massive utility grade power we've all been striving for. This is a blessing in disguise.

 

[koppted]

Regardless of master/slave comms, "seeing" a parameter outside of grid interactive setpoints should have caused a complete "override" of all operation and a full disconnect! I understand "bugs" slipping through, but this has a few too many different aspects that should have been covered....

On a side note, as I promised to someone in the other thread, I checked how our Sol-Ark 12K reacts under a full clean disconnect from grid. It nearly instantly goes to zero volts on the grid side of the unit, which is what I expected. I took video of a few scenarios and I will edit this post to link a video once I get it put together and onto YouTube. (I tried a few scenarios, so have to put a few videos together into one, and I haven't ever done that. Wish me luck. )

P.S. If anyone thinks I should make a separate informational thread with that content, let me know.

*Link will be here once I get the video on YouTube.

A proper one-to-one comparison would be to see how the two(2) SolArk 12ks do when someone uses a crossover cable for communication. Please report back if you are able to do the tests with these parameters.

 

[marionw]

I guess the positive side of things is that you didn't lose your house in a fire after up to 570+ volts of AC were sent to your appliances.

Common Ethernet cable colors:

Since there is no direct industry standard for Ethernet cable colors, the ANSI/TIA/EIA-606-B is a voluntary standard showing the importance of labeling and coloring. That being said, there are a few consistencies within the industry worth mentioning here.

The most common colors used are gray, blue, yellow, orange and white.

Gray Ethernet cables are considered “standard” and are often found in residential and commercial networks.

Blue Ethernet cables are commonly used to connect a terminal server to a LAN network. A terminal server provides the ability to connect multiple systems to a LAN network without the need for a modem or other network interface.

Yellow Ethernet cables are generally used for what is called “power over Ethernet” (POE). This means that your cable can supply both electrical power and data in one connection.

Green Ethernet cables are commonly used to classify a crossover connection, which connects different computers and/or devices directly together.

Outdoor Ethernet cables are usually Black and made using outdoor graded black PVC that resists wind, moisture and sunlight.

Ethernet Cables and What Their Colors Mean

When you begin your journey to purchase Ethernet cables, we are here to help you make the best color choice. Whether upgrading an existing network or starting from scratch we can help take the mystery out of color-coding your project no matter how big or how small.It doesn’t matter what color...

satmaximum.com

 

[SparkyJJO]

Common Ethernet cable colors:

Since there is no direct industry standard for Ethernet cable colors, the ANSI/TIA/EIA-606-B is a voluntary standard showing the importance of labeling and coloring. That being said, there are a few consistencies within the industry worth mentioning here.

The most common colors used are gray, blue, yellow, orange and white.

Gray Ethernet cables are considered “standard” and are often found in residential and commercial networks.

Blue Ethernet cables are commonly used to connect a terminal server to a LAN network. A terminal server provides the ability to connect multiple systems to a LAN network without the need for a modem or other network interface.

Yellow Ethernet cables are generally used for what is called “power over Ethernet” (POE). This means that your cable can supply both electrical power and data in one connection.

Green Ethernet cables are commonly used to classify a crossover connection, which connects different computers and/or devices directly together.

Outdoor Ethernet cables are usually Black and made using outdoor graded black PVC that resists wind, moisture and sunlight.

Ethernet Cables and What Their Colors Mean

When you begin your journey to purchase Ethernet cables, we are here to help you make the best color choice. Whether upgrading an existing network or starting from scratch we can help take the mystery out of color-coding your project no matter how big or how small.It doesn’t matter what color...

satmaximum.com

Definitely voluntary.
I've found white and gray to be used where the cables would hide better along trim/baseboards, but any mix of any colors for any usage. Big ugly rolls of dirty yellow and blue cables stuffed under workstations in offices. Red cables just because that's what was on hand. Black because "it looks cool." All regular cables.
Majority of crossover cables I've run across in the field were red or orange.

 

[murmeister]

They do come with the parallel cable. I don't know why the installer would use some random cable.

I might can shed some light on this issue, specifically.

The cable that comes with the inverter is just long enough for their "standard" configuration - Inverter stacked on conduit on top of primary battery. If you deviate from this configuration, the cable will most likely be WAY too short.

My installation has the inverter on one side of a brick wall and the batteries on the other. I had to use a much longer cable to make the connection. I used a very long CAT6 cable that I used before I installed wireless repeaters in my house. I'm REALLY glad I used the proper one. I, too, thought that an RJ45 was an RJ45. I've wired my own cables before but never had a need for a cross-over cable. I've heard the term but never worried about something I didn't need.

My installer suggested that I contact EG4 and request that they provide longer cables when requested. I'm thinking he was spot on.

 

[ricardocello]

how many of these things can be stacked together? My thoughts is this puny 400-600v is nothing compared to what these babies can really do. Maybe add another 1,2,4 of these together in one system with a bunch or a special secret combination of cross over cables and we can have that massive utility grade power we've all been striving for. This is a blessing in disguise.

The maximum is 10 in parallel! Think of the possibilities!

 

[Plum Crazy Rob]

I guess the positive side of things is that you didn't lose your house in a fire after up to 570+ volts of AC were sent to your appliances.

Can anyone substantiate that 600V rated residential wiring hopefully has a safety factor?? Hopefully no melted wiring/ future potential probs...