Design Review - Growatt SPF 5000 ES Grid Backup & Neutral Bonding

[Desert_AIP]

I'm not understanding the desire for an unattended cold start (restart).
If there is a failure that causes one of these safety mechanisms to activate (deactivate), I would think you would want to require manual intervention to identify the cause of the failure before restoring power.

 

[??????]

I'm not understanding the desire for an unattended cold start (restart).
If there is a failure that causes one of these safety mechanisms to activate (deactivate), I would think you would want to require manual intervention to identify the cause of the failure before restoring power.

@Desert_AIP - the reason for potentially wanting an unattended cold start is because the system must be attended for a cold start even if the system is operating normally. It's arguable if the AT disconnect contactor is even needed (creating the cold start problem in the first place), since the AT is protected by the 30A CB on the inverter disconnect service panel. There's also the (untested) use case of the grid-to-offgrid transfer within the GW internal ATS. The manual says this is 10-30ms transfer time. That might be enough to lose the electric field in our contactor coils and force us into a cold start scenario to get our 120V load center back.

Here's my proposed solution to the cold start problem. This design (should):

• Disconnect L1/L2 input to the AT in the event the AT does not produce 120V +/- tolerance on its two legs
• Isolate the 120V load center from the cold start process (solves issue number 2 of #150)

But there's more - this proposed design can have the cold start contactor functionality extended with a few different options:

1. Attended (least risk, pictured in attached diagram): A plain-jane 2PDT HVAC contactor - the operator physically actuates the throws of the contactor (there's a button on the unit for this purpose) - no electricity required - to bootstrap the system from a cold start. Operator must be on-site.
2. Remote-attended (medium risk): A wifi switch that runs on the GW AC OUT 240VAC and actuates the coil of the cold start contactor. This would allow a remote operator to 'push the cold start button' from anywhere in the world via wifi/cellular/etc.
3. Fully autonomous (highest risk): A logic controller that effectively emulates the remote operator of #2 above. It attempts to bootstrap the system n number of times, and throws an error (optional) if bootstrapping fails. No operator required.

 

[Desert_AIP]

That makes a bit more sense.
I would think you would want a fail safe (require manual intervention) for a loss of neutral situation.

The other scenarios, switching, etc. autonomous.

I've used cascading relays before where you have to press a momentary button to "arm" the primary relay, and it's wired into a self latching configuration.
The primary relay's contacts then feed one side of the coil to a secondary relay that controls a load.
And then the switching of the load is controlled by the other input to the secondary coil.

The primary coil is dependent on some critical factor, and if that fails, the entire system shuts down and requires manual reset.
But as long as that critical factor is maintained, the load can be readily switched on and off.

You're using some of the same logic in your diagram.

If you wanted to auto restore a situation initiated by a lost neutral, you'd HAVE to verify the neutral was restored first.

 

[??????]

If you wanted to auto restore a situation initiated by a lost neutral, you'd HAVE to verify the neutral was restored first.

Yep - exactly right. If we wanted autonomous cold-start (option 3 of #152), the neutral verification is performed by the two over/under voltage protection relays - protecting the 120/240V load panel at all times. The only thing the cold-starting potentially risks is the AT, which is still protected by a 30A DP CB at all times. If anybody has transformer failure experience, that'd be valuable to have here for perspective.

We should keep in mind that the OP intent was to solve the David Poz problem of a trip of the 30A CB for the AT and losing a neutral due to that. I'd say #133 does that already, and #152 is possibly overkill. However, we don't monitor AT temp, which is something SolarEdge does do (#145), so maybe #152 isn't a bad idea, just-to-be-safe.

My personal opinion is that - if @automatikdonn can test the grid/offgrid ATS, and it holds the contactor coils (does not need a new cold-start), I'll probably implement #152 with option 1 (attended cold start). Grid going down for a period that wouldn't be supported by battery - and I wouldn't be present for - is very unlikely in my case, and the added AT safety is nice to have. People's use cases/risk tolerances may vary and they should implement accordingly.

 

[Desert_AIP]

I'm thankful for this thread and discussion. Relocating in a few months and want to install a system like this with a ground array.

 

[automatikdonn]

@automatikdonn - this looks great! Really cool that you're birddogging this effort with the real world setups to validate the schematics - thanks for taking all this time to share.

Feedback:

1. Cold-start with a 3-position switch is a great idea - but I think it can be simplified further. If you can manually actuate one (or more) of the coils of your three contactors, you could just manually close the contacts for the ~2s that the voltage protection relay would need to 'pick up' the 120V monitoring and take over the job of electrically holding the coils on those contactors. I'll noodle this idea this weekend - I think I have an idea to remotely simulate a manual push button for folks who may not have physical access (i.e. cabin in the woods)
2. I think there may be a significant design flaw here with placing the 240V AT feed on the same panel as the 120V loads. If you do have a faulty AT (let's say it puts out 195V on one leg), the bootstrap process would expose the loads on that '120V load panel' to this voltage - which is exactly what we're trying to avoid. I think we need to place the ATs upstream of the contactors at all times to avoid exposing any 120V loads to stray voltages even during bootstrapping.
3. I think the wiring you have on the three contactors exposes them to the imbalanced load breaking problem that @LeRoyK brought up in #129 and #130 - if we've got two switched phases (L1, L2) on the 3PDT contactors you're using, I think we're limited to a max of two contactors, each contactor switching 150A (50A x 3 poles). Each phase would be distributed across the three switched poles of each respective contactor. Unless we find different hardware I think that's our physical design limit.

1. Yea I was thinking a simple push button momentary switch would have worked really nicely, but I didn't have one. I would also like to wire in a green light so I know when the system is active. (mainly for the wife who may have to operated it from time to time). And sure a screwdriver works to active the contactor as well, I was just out of hands for my video.

2. You are 100% correct and there should only ever be 240V loads coming from this panel. There even needs to be a sticker if I am not mistaken that says "240V ONLY" . The single pole breakers in that box in the video were just there from the other panel I tore apart to make the new disconnect panel. I am terrible at misplacing things and if they are in the box they can't get lost. Anyone find my 10mm sockets, please send them my way.

3. The contactors I have are 50amp, so I would only be able to switch 100 Amp if I switched the hot legs over two contactors no? How does that work, I am not really picking up what you are putting down with this. Is each pole rated for 50Amp? I get the reasoning for doing it and it makes total sense to me. 99.999% of the time the system will only see a max load of 62.5 amps anyways, so we are really only talking about surge.

I think if I can wire this stuff up and show people it works and is safe, maybe they can trust the schematics and BOM a little more.

I can check on switching from grid / offgrid. I think these units are UPS fast at switching, so I think the timer on that 120v relay would never even notice. It would be nice to know for sure though.


I have also been thinking about the neutral auto bonding circuit and how it plays into portable genset usage. I have several portable gensets and not a one of them is floating neutral. I even made a video on how to hack the inverter gensets I have to make them floating... But with your auto bonding setup, you wouldn't really have to worry about it anymore and just run the generator like it was mains power (for those that have that use case).

 

[automatikdonn]

@Desert_AIP - the reason for potentially wanting an unattended cold start is because the system must be attended for a cold start even if the system is operating normally. It's arguable if the AT disconnect contactor is even needed (creating the cold start problem in the first place), since the AT is protected by the 30A CB on the inverter disconnect service panel. There's also the (untested) use case of the grid-to-offgrid transfer within the GW internal ATS. The manual says this is 10-30ms transfer time. That might be enough to lose the electric field in our contactor coils and force us into a cold start scenario to get our 120V load center back.

Here's my proposed solution to the cold start problem. This design (should):

• Disconnect L1/L2 input to the AT in the event the AT does not produce 120V +/- tolerance on its two legs
• Isolate the 120V load center from the cold start process (solves issue number 2 of #150)

But there's more - this proposed design can have the cold start contactor functionality extended with a few different options:

1. Attended (least risk, pictured in attached diagram): A plain-jane 2PDT HVAC contactor - the operator physically actuates the throws of the contactor (there's a button on the unit for this purpose) - no electricity required - to bootstrap the system from a cold start. Operator must be on-site.
2. Remote-attended (medium risk): A wifi switch that runs on the GW AC OUT 240VAC and actuates the coil of the cold start contactor. This would allow a remote operator to 'push the cold start button' from anywhere in the world via wifi/cellular/etc.
3. Fully autonomous (highest risk): A logic controller that effectively emulates the remote operator of #2 above. It attempts to bootstrap the system n number of times, and throws an error (optional) if bootstrapping fails. No operator required.View attachment 72980

For #3 I was trying to come up with a safe way to use the NC side of the contactor, but I couldn't crack it.

 

[automatikdonn]

I'm thankful for this thread and discussion. Relocating in a few months and want to install a system like this with a ground array.

Well I am running my 3k shop on this setup right now. If I could just get the 3 Phase to run off this too I would be set.

 

[??????]

Okay next iteration - this one is #152 with the following changes:

• Uses 120v coil N safety contactor instead of 240v coil contactor (50A 3-pole DIN-mount instead of 40A 2-pole non-DIN mount)
• Adds electrical momentary cold start switch

This (should) allow electrical actuation of the cold start relay with a momentary pushbutton (option 1 of #152). Note that the cold start relay must be a 240v coil since the only power source available, with the AT off, is the 240v from GW AC OUT. Also note that this momentary pushbutton could be replaced with a wifi switch to allow remote or autonomous cold starting (options 2 & 3 of #152).

One detail is that the cold start button must be at least a 2PDT switch, with one set of N/O and one set of N/C contacts. The N safety contactor coil is routed through the N/C contacts of this switch to avoid closing the N safety contactor during the cold start process (exposing the 120/240v load center to potentially out of tolerance voltage).

The cold start process should look like this:

• Cold start button pressed. N safety coil de-energized. Cold start coil energized. AT receives power and begins forming neutral.
• First over voltage protection relay sees L1/N 120v. 2s delay before N is forwarded to second over voltage protection relay.
• Second over voltage protection relay sees N/L2 120v. 2s delay before L2 is forwarded to the top side of the cold start switch. N safety contactor coil is not energized because its L2 is routed through the cold start switch, which is N/C (so at this point, it's open).
• Cold start switch button is released
  • Coils of cold start contactor change from being powered directly from L2 of the second voltage protection relay.
  • N safety contactor coils are powered from L2 of the second voltage protection relay via the N/C contact of the cold start switch.
  • System is fully functional (if both voltage protection relays pass tests)

So, in short:

• Push the cold start button
  • AT is immediately energized
• Wait 4 seconds
• Release the cold start button
  • Happy path (both 120v leg pass check):
    • N safety contactor should engage immediately afterward
  • Unhappy path (either 120v leg fails check):
    • AT will de-energize immediately after the cold start button is released
    • N safety contactor will never close, leaving 120/240 loads unpowered

 

[automatikdonn]

Yep - exactly right. If we wanted autonomous cold-start (option 3 of #152), the neutral verification is performed by the two over/under voltage protection relays - protecting the 120/240V load panel at all times. The only thing the cold-starting potentially risks is the AT, which is still protected by a 30A DP CB at all times. If anybody has transformer failure experience, that'd be valuable to have here for perspective.

We should keep in mind that the OP intent was to solve the David Poz problem of a trip of the 30A CB for the AT and losing a neutral due to that. I'd say #133 does that already, and #152 is possibly overkill. However, we don't monitor AT temp, which is something SolarEdge does do (#145), so maybe #152 isn't a bad idea, just-to-be-safe.

My personal opinion is that - if @automatikdonn can test the grid/offgrid ATS, and it holds the contactor coils (does not need a new cold-start), I'll probably implement #152 with option 1 (attended cold start). Grid going down for a period that wouldn't be supported by battery - and I wouldn't be present for - is very unlikely in my case, and the added AT safety is nice to have. People's use cases/risk tolerances may vary and they should implement accordingly.

It does NOT hold the contactor coils.. at all. Which gives me faith in the protection circuits, but bums me out because we still have a cold start issue.

 

[??????]

It does NOT hold the contactor coils.. at all. Which gives me faith in the protection circuits, but bums me out because we still have a cold start issue.

I think option 3 of #152 is probably the move here if a fully autonomous system is the goal. This could be executed a couple different ways:

• Blind one-shot (less effort): tie an off-the-shelf timer in parallel to the coil of the N bonding contactor. Program this timer to perform a one-shot ~4-10s hold of the cold start momentary button in #152. This will execute a one-shot, blind cold-start of the system at the same time the neutral bonding contactor is energized (when the ATS goes on->off grid). This is relatively simple to implement and uses off the shelf parts, but has no monitoring, no remote control, and no retry capability.
• Monitored (more effort): use a microcontroller (arduino, esp32, whatever) to handle the cold start process. Big benefit of this is that we'd be able to read voltages, current, and most importantly AT internal temp during bootstrap and running states, and having a connected device with GPIO also opens the door to shipping logs/data, etc.

My opinion is that there's a reason there's a temp sensor in the AT, and right now we don't have any way to monitor it. I'd also put my money on a temp sensor + microcontroller being a safer cold start solution than a human because a) the temp sensor has access to data (internal winding temp) the human doesn't have, and b) the microcontroller can act on sensor inputs many times faster than a human can as well.

 

[automatikdonn]

I think option 3 of #152 is probably the move here if a fully autonomous system is the goal. This could be executed a couple different ways:

• Blind one-shot (less effort): tie an off-the-shelf timer in parallel to the coil of the N bonding contactor. Program this timer to perform a one-shot ~4-10s hold of the cold start momentary button in #152. This will execute a one-shot, blind cold-start of the system at the same time the neutral bonding contactor is energized (when the ATS goes on->off grid). This is relatively simple to implement and uses off the shelf parts, but has no monitoring, no remote control, and no retry capability.
• Monitored (more effort): use a microcontroller (arduino, esp32, whatever) to handle the cold start process. Big benefit of this is that we'd be able to read voltages, current, and most importantly AT internal temp during bootstrap and running states, and having a connected device with GPIO also opens the door to shipping logs/data, etc.

My opinion is that there's a reason there's a temp sensor in the AT, and right now we don't have any way to monitor it. I'd also put my money on a temp sensor + microcontroller being a safer cold start solution than a human because a) the temp sensor has access to data (internal winding temp) the human doesn't have, and b) the microcontroller can act on sensor inputs many times faster than a human can as well.

I have been testing all day on both genset and POCO power and it does not changeover fast enough at all to maintain the contactor connection.

We could also use the NC side of the contactor to charge the circuit, with a timer in the middle so it can't flip on and off too quickly.

 

[??????]

We could also use the NC side of the contactor to charge the circuit, with a timer in the middle so it can't flip on and off too quickly.

This is a good idea, but I think we're just straight up without power for ~30ms during the ATS transition. It's also worth noting that the ATS (according to the docs) takes longer as you add more inverters in parallel. It's possible this isn't an issue with a single inverter ?‍

 

[automatikdonn]

This is a good idea, but I think we're just straight up without power for ~30ms during the ATS transition. It's also worth noting that the ATS (according to the docs) takes longer as you add more inverters in parallel. It's possible this isn't an issue with a single inverter ?‍

Well it would at least get us back to an automated circuit that does cut off AT input because it's upstream of the AT.

I have a few safety concerns with this model however because we would also have to concern ourselves with the loads that would be switched.

I think a PLC or something of that sort (Arduino, Pi) provides the most safety with maintaining degrees for automation. However we are slinging that code because I am not so sure we could use something off the shelf. I only know python, so I would be using a PI.

I don't really want to make this take any longer than necessary, but also don't want to compromise on safety. I don't know how often a system would be switching between inputs, I know mine would likely only switch if there was a problem and I would be present for that anyway. I don't know how big of a deal it is for people to have to be around.

I think we should cut a v1 BOM with the associated caveats to explain the issues with the circuit. We could have two versions, one automated but with safety risks(low, but still a risk) and one that is super safe but needs an adult to operate.

I am just one voice, so what does everyone else think?

 

[LeRoyK]

What if we had an AT temp monitoring circuit that could disconnect AT input Power?
"Lost Neutral fault" would not disconnect AT Input, this would allow it to auto-reset after an ATS transition.
AT would be protected by OCP and Over-Temp protection.

What components do we need to build the AT Over-Temp Protection circuit? We could use a shunt trip circuit breaker as our disconnect instead of a contactor. this would simplify things, allowing the breaker to do both OCP and Over-Temp disconnect.

How do we trip the shunt with AT build-in temp sensor? Does anybody know if the build-in temp sensor is a thermocouple or an RTD? I think we need something like a signal conditioner.

Here is a list of components that I think might work. (Hopefully you guys have more experience with temp monitoring than I do.)

• GMCBU-2B-30 2pole 30A Breaker Tech Specs (Changed breaker to 30A 11/22/21 9:57AM)
• GMCBU-SH110-380VAC Shunt trip for Breaker Tech Specs
• SCU-3100 Signal Conditioner Tech Specs

 

[LeRoyK]

Scaling up with multiple inverters in parallel. I think the GW internal ATS will have the same imbalanced load problem during transfer as the Schneider inverters had #129. So I would recommend if you want to use the grid with multiple GW in parallel as backup, then build a complete external transfer switch with Grid voltage monitoring relay. Do not even connect AC IN to the inverters.

 

[automatikdonn]

What if we had an AT temp monitoring circuit that could disconnect AT input Power?
"Lost Neutral fault" would not disconnect AT Input, this would allow it to auto-reset after an ATS transition.
AT would be protected by OCP and Over-Temp protection.

What components do we need to build the AT Over-Temp Protection circuit? We could use a shunt trip circuit breaker as our disconnect instead of a contactor. this would simplify things, allowing the breaker to do both OCP and Over-Temp disconnect.

How do we trip the shunt with AT build-in temp sensor? Does anybody know if the build-in temp sensor is a thermocouple or an RTD? I think we need something like a signal conditioner.

Here is a list of components that I think might work. (Hopefully you guys have more experience with temp monitoring than I do.)

• GMCBU-2B-50 2pole 50A Breaker Tech Specs
• GMCBU-SH110-380VAC Shunt trip for Breaker Tech Specs
• SCU-3100 Signal Conditioner Tech Specs

Can you explain what a shunt trip breaker is? How does it work?
I have never heard of this before.

 

[automatikdonn]

Scaling up with multiple inverters in parallel. I think the GW internal ATS will have the same imbalanced load problem during transfer as the Schneider inverters had #129. So I would recommend if you want to use the grid with multiple GW in parallel as backup, then build a complete external transfer switch with Grid voltage monitoring relay. Do not even connect AC IN to the inverters.

I thought putting the legs on separate contactors resolves this issue.

 

[??????]

What if we had an AT temp monitoring circuit that could disconnect AT input Power?

I agree - the more I read about transformer failure, the more I see that the industry seems to use ferrite core temps as the single most important health metric for dry transformers like the AT we have here. The AT ships with temp sensing, and I pulled one apart to see that the leads are between the windings and the ferrite core - right where we want them.

Measured resistance of this probe is very low, effectively zero on my ohmmeter. Therefore, I do not believe it is an RTD. It is likely a K or T type thermocouple, and I'm led to believe it's a T type since neither conductor appears to be magnetic.

Re: system design, we're at a decision point:

• Standalone temp sensor (#165) - AT is held by twoconditions (below). This trades the cold-start challenge with the fact that the AT's power source is no longer conditionally monitored by its 120v legs:
  • Temp is below threshold
  • 30A DP OCPD feeding the AT is not tripped
• Integrated temp sensor (#159 with arduino) - AT is held by threeconditions (below). We need to implement either a timer (naieve) or arduino-esque (informed) automated method of cold-starting:
  • Temp is below threshold
  • 30A DP OCPD feeding the AT is not tripped
  • 120v +/- tolerance is held on both legs of the AT

In practice, we're likely to see high temps on our AT windings way before we ever 'lose a leg', so #165 is probably just as good as #159 + automated cold start, and it's zero dev work.

#159 + arduino/esp32/etc has some nice potential 'future-features' though:

• Remote access
• Logging (current, voltage, kWh, etc)
• Multiple transformer/inverter temp sensing through SPI bus
• Orchestration/peak shaving of priority loads on branch circuits during on/off grid states

 

[LeRoyK]

Can you explain what a shunt trip breaker is? How does it work?
I have never heard of this before.

A shunt trip device is an optional accessory in a circuit breaker that mechanically trips the breaker when power is applied to the shunt trip terminals. The power for the shunt trip does not come from within the breaker, so it must be supplied from an external source.

 

[automatikdonn]

A shunt trip device is an optional accessory in a circuit breaker that mechanically trips the breaker when power is applied to the shunt trip terminals. The power for the shunt trip does not come from within the breaker, so it must be supplied from an external source.

Any ideas how we could wire that up, as we are monitoring a loss of power. I am not doubting it will work, I just have zero experience with them.

 

[LeRoyK]

Any ideas how we could wire that up, as we are monitoring a loss of power. I am not doubting it will work, I just have zero experience with them.

monitoring a loss of power.
We are only monitoring the loss of Neutral and Disconnecting the Output (loads) from the AT.
AT Input will only be disconnected by Over-temp condition or OCP.
Loss of power during transfer will not cause Shunt trip Breaker to Trip.

how we could wire that up?

• Wire AT build-in temp sensor to SCU-3100 Signal Conditioner input
• Run 240V thru SCU-3100 Signal Conditioner Relay output to Shunt
  • How does it work? AT Temp is monitored by Signal Conditioner
  • If AT Temp goes above a preset alarm temp, then Signal Conditioner closes relay contact and puts power to Shunt to trip the breaker that feeds the AT.

 

[automatikdonn]

monitoring a loss of power.
We are only monitoring the loss of Neutral and Disconnecting the Output (loads) from the AT.
AT Input will only be disconnected by Over-temp condition or OCP.
Loss of power during transfer will not cause Shunt trip Breaker to Trip.

how we could wire that up?

• Wire AT build-in temp sensor to SCU-3100 Signal Conditioner input
• Run 240V thru SCU-3100 Signal Conditioner Relay output to Shunt
  • How does it work? AT Temp is monitored by Signal Conditioner
  • If AT Temp goes above a preset alarm temp, then Signal Conditioner closes relay contact and puts power to Shunt to trip the breaker that feeds the AT.

I see, so this would be for temp protection. I wasn't picking that up. It would be awesome if we could use a shunt trip breaker to also turn off GW output in the event of a neutral leg loss too.

 

[LeRoyK]

I see, so this would be for temp protection. I wasn't picking that up. It would be awesome if we could use a shunt trip breaker to also turn off GW output in the event of a neutral leg loss too.

Shunt trip breaker results in having to do a cold start.

We want to be able to auto-restart on "loss of neutral" because every time inverter transfers we will have a "loss of neutral".

 

[Jim Lee]

I think we stop at a Full BOM and a working circuit that keeps peoples houses from burning down. The requirements for that are pretty simple IMO. If the neutral leg fails, turn off the things. If the utility power is being passed through, don't bond the neutral leg to the ground in the downstream panel. There has been a lot of great content, and if all someone wants to do is switch off the system when the AT fails, then they are likely to be done. In the spirit of answering questions people ask to the fullest possible extent, I would say we are very close conceptually, but no solid BOM. For me getting in parts to even test our theories has been a whole thing and it's very frustrating because I have a complex of not wanting to wait.. I believe they call it impatience.

The thread may be useful for other things, but I think we are trying to get to the answer for the OP. "This seems to be a very popular setup, but I haven't found anybody who's cracked the DaVinci code of using a Growatt SPF 5000 ES with a SEAUTO-TX-5000 Auto-Transformer and using the AC input of the Growatt in a 'utility first' configuration."

I don't understand the utility first priority??? Don't most people that want an off grid converter that is grid tied because they want PV first then battery then grid then generator. Doesn't that cover all the bases. Power no matter what. I'm in the planning process and ready to purchase 30 - 295 watt panels. 2 x 6s2p and 1 x 3s2p. Each of the 3 string is wire to 1 of the 3 Growatt SPF 5000 ES inverter. 6 Li Fe EG4 Batteries connected to the inverters. The 3 inverters connected to a separate circuit box with a 50A breaker each. That breaker box set up with 1 or 2 (if safer) split phase Mid point transformers connected with 50A breakers. That breaker box is connected to my home main breaker box with a 60A or 100A breaker (not sure yet) connecting both circuit boxes. The Grid and my 19.5Kw Generac Generator pass through the 200A Transfer switch keeping the grid safe during grid outages is already installed. According to Signature Solar this is the system they want to sell me. You guys have me so worried I'm having trouble starting the whole project. Please explain to me why this won't work and what I need to do to make it work. I really don't want to burn my house down trying to improve it. Another question is when the grid goes down I think my generator will auto start even if I have PV or battery power. How could I control that. Any other suggestions would be appreciated, but please keep it simple if possible.