diy solar

diy solar

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

??????

New Member
Joined
Sep 25, 2021
Messages
53
This seems to be a very popular setup, but thus far, nobody seems to have cracked the DaVinci code of safely 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. Signature Solar has posted a schematic that appears to have concerns that are not present in this design.

Design Philosophies:​

  • Bake some redundancy into the system. I know the solaredge AT's aren't free, but get a couple of them. Even though you may only *need* one for loads, run two so in case one breaks you are still good to hook. Edit Jan 19, 2021: Do not use multiple ATs in your setup. See Jan 19 2021 update below.
  • Sleep well at night. Transformers are reliable technology, honestly they are.
  • Implement failsafes - When the BOM and schematic is done, we are going to make a video to go through the whole thing in detail. I like the ELI5(explain it like I am 5) method and @automatikdonn has generously offered to do a video at that level in the near future. Here is his implementation of design #159:
  • Complexity can be managed - designs below are as simple as they can be while retaining critical safety features. If you are able to produce schematics for simpler solution that covers all grounding, fault, over-temp, over-current, over/under-voltage, cold-start, and grid-forming timing issues with a non-isolated transformer - please post.

Below are two designs resulting from this thread. The first, #159, is a manually cold-started version that has been deployed in the field and is currently in use by @automatikdonn. The second, #216, featuring temp and current alarms as well as automatic cold-starting, is in development and has not yet been deployed.

Version #159 - Attended Cold-Start

  • 120/240V load center capable of handling up to 30A, with a neutral safety disconnect
  • 240V load center capable of handling 1->8 GW units in parallel (single phase only)
  • Conditional ground/neutral bonding of the 120V load center depending on if the GW is operating on-grid or off-grid
  • No autotransformer temp sensing
  • NOT YET TESTED for GFI while in on-grid mode
Here is a video graciously filmed by @automatikdonn of #159 deployed:

Growatt 5000ES with Neutral disconnect - Separate 120_240 Load Panels & Single Contactor, 120v...png



Version #216 (WIP) Automatic Cold Start - GitHub Link Here (PRs/forks welcomed)​

216.png

Jan 19 2021 Update: Although risk of using two identical SolarEdge ATs (instead of one) is likely minimal (manufactured in the same facility, same processes, and no differential loads between them), there is the potential for manufacturing tolerances to induce latent 'balance current' between multiple non-isolated transformers run in parallel. If you need more balancing capacity, use a single, larger non-isolating midpoint transformer with larger gauge windings and ferrite core.


Useful links:
 
Last edited:
I'm very interested in this. Right now I'm just planning on using this 100% off grid, but I would like to add "grid assist" later on.

What is preventing you from relying on the built in ATS? What doesn't it do?
Does N from the auto transformer need to go to the sub panel and be switched as well?
 
@bgflyguy - great question (y)

You are correct - the GW ATS does not switch neutral. As far as I see it, if we wanted to use the GW ATS to switch L1/L2, we'd have two options to handle the neutral:
  1. Leave the autotransformer neutral permanently connected to our service panel bonded neutral buss.
  2. Externally XOR transfer switch of our main service panel neutral between one of two sources:
    1. Grid neutral
    2. Autotransformer neutral
Let's go into the details of these two options...

Option #1​

Our neutrals from grid and autotransformer are now connected for both on grid and backup states. Although this appears to work in a backup state (grid L1/L2 GW AC IN isolated from GW AC OUT via internal ATS), this does not appear to work in an on grid state for the following reason: When in an on-grid state, L1/L2 AC IN of the GW 'feeds through' to L1/L2 AC OUT of the GW, which - in turn- feeds directly to our autotransformer (AT). Since the neutral of our AT is a) non-isolated, and b) tied directly to grid, this means our AT is in parallel to our grid transformer. If we are the only customer on our transformer at the pole - this is not an issue. But if there are other customers, then our AT will attempt to participate in the neutral-forming for loads outside of our service. This is my theory based on my reading - anybody more qualified than myself, please correct me. This is my key concern for @CCAT Racing 's design proposal which appears to use a permanently connected AT neutral to grid. Attached is the way this could look if wired up:

1633916757241.png

Option #2​

Running an external transfer switch that is controlled by the dry contact of the GW via setting #24 (see pg 12). I don't have a diagram for this, but it'd effectively be a SPDT XOR switch, with the center COM connected to the main service panel's neutral buss, one throw connected to the grid neutral (from the utility pole), and the other throw connected to the AT neutral. Here are some of the challenges of this strategy:
  • We are switching the neutral conductors, which classifies our GW as a separately derived system according to NEC article 100. This means we're held to NEC article 250.35(A) (separately derived system), and therefore need to comply with all of article 250.30(1-8). It's like three pages of design considerations.
  • More importantly, we have a possible latency issue here. We're switching L1/L2 and N with two separate switches. L1/L2 are internal to the GW ATS and are subject to their transfer time. We have another external transfer switch (the SPDT for the neural) which is subject to a different transfer time. So we have exceptions to catch here:
    • What if L1/L2 switch first, shortly thereafter followed by our N switch? Will we have an over-under voltage on our split phase devices?
    • What if we have a failure of one of the two switching devices entirely? How is this error handled?
I do not believe option 2 is viable due to the second consideration of switching across two independent devices. There's a reason why transfer switches house all conductors within the same switch - because the throws are mechanically linked such that it's an all-or-nothing state transfer. The only way that'd be violated is if the throws broke internally and only a fraction of the poles changed state, and physical engineering efforts likely work in a large factor of safety to account for that scenario. I'm sure switching of multiple dependent conductors across two physically unlinked switches is a code violation, but I'm not sure where it is.

A way around this may be to internally rewire the GW to be a 3PDT (L1/L2/N) instead of a DPDT (L1/L2) ATS. Then we'd be switching one ATS instead of two. Assuming we comply with NEC 250.30, I think that'd work and be fully automatic.


If we can get confirmation on my 'Option #1' theory that the AT and grid transformers would parallel in an on-grid state (potentially catastrophically balancing load for other customers), that'd be amazing. If I'm wrong, then Option #1 seems viable. If I'm right, then hopefully this has saved somebody from making a serious error.
 
Last edited:
I was trying to get this very question answered by the people at Signature Solar just this past week. The person I emailed back and forth with was unhelpful and never answered my question about how to wire AC in for the on-grid scenario.

One way I think you could do this is to wire grid power directly into the inverter (AC in) and then wire the inverter (AC out) to the breaker panel (attached to a circuit breaker), where the transformer also would be wired (attached to its own circuit breaker). That is, put the inverter in between the meter and the main breaker panel. The problem with this approach is that the inverter likely is not big enough to run all the circuits on the house.

If you figure this out, I hope you will post it here as I would love to know how to do it with a sub-panel.
 
Thanks @LAS - I've done some more thinking and there's a problem with my original design. When in 'on-grid' mode, the inverter 'feeds through' AC IN to AC OUT. Since the AT is neutral connected to the main panel, this places the AT in parallel with the grid transformer, which places this design in the 'option 1' category from my second post, but only when operating in on-grid mode. A way around this would be to power off the inverter (operate in standby mode) - according to the docs, the inverter will NOT feed AC OUT, but will still charge the battery bank. Then, you'd need to find a way to interlock the turning on of the inverter to the state change of the manual safety switch.

I've thought of a way around all this which uses inverter option #24 to conditionally power the AT using contactors. It's more complicated than I'd like, but it's a fully automatic transfer instead of a manual one, and it also appears to solve David Poz's problem of 'losing the neutral' when operating off-grid. I'll see if I can put a schematic together this week.
 
The "downstream" loads on the inverter cannot be in the same panel or a sub panel that is attached to the grid. This inverter is not a grid tie and can't sync and does not have anti-islanding.

Think of the Growatt as an appliance, plugged in to an outlet connected to the grid.

You would take grid power into a primary circuit breaker panel.
A circuit from that panel would feed the Growatt on the intake side.
Then the Growatt powered circuits would need to be in a COMPLETELY DIFFERENT PANEL isolated from the grid panel.

The Growatt will intake grid power and pass it on to the loads, and switch over to battery or PV power depending on your settings.

So you would need to install a secondary panel (not a sub panel off the primary panel) including it's own grounding rod, and then pull over any loads you want backed up from the primary panel to the secondary/Growatt panel.

The Growatt is an Offgrid system with a Utility/generator input.
You're building your own local subgrid with it.

(Every thing below this had some errors included)
 
Last edited:
Thanks @Desert_AIP - I've taken your feedback and added a couple extras:
  1. Added neutral safety disconnect (will talk more about this in a sec)
    1. Changed auto transformer CB from a single pole to double pole 30A
  2. Added double pole 50A fuse on AC OUT of inverter
  3. Moved inverter G/N bond from the auto transformer enclosure (?) to the inverter's "Derived Main Panel"
The above, in more detail:
  1. The neutral safety disconnect aims to solve the challenge that David Poz appears to have, which is "what happens when my AT CB trips, leaving no neutral forming equipment?" The neutral safety disconnect panel contains four components: two over-voltage (120VAC) devices, and two normally-open contactors (with 120VAC coils). The idea is that either pair of over-voltage/contactors can to interrupt the flow of 240V to the subpanel. In other words, if either leg of the auto transformer falls outside of 120VAC +/- 10VAC (for example), the over voltage device opens, de-energizing the contactor coil, and therefore cutting power to the subpanel. Since the path from the derived main panel to the subpanel travels through both contactors, either contactor opening (losing either leg of the autotransformer) results in the subpanel being disconnected.
    1. Changed to a double-pole CB here because I'm frankly not sure what would happen if you lost one half of a midpoint transformer (single pole trips, one pole does not trip).
  2. The change to a CB on AC OUT of the inverter seems to make sense since there are potentially three energy sources the inverter can draw from (AC IN, PV, BAT) so a CB on AC IN alone doesn't cover the sum of PV & BAT as well.
  3. The move from the subpanel to the derived main panel is not really a functional one, but a packaging one. With a derived main panel, the inverter, autotransformer, neutral safety disconnect, and derived main panel can all be packaged near the service (grid/main) panel. This seems like it'd be a bit easier to install as a retrofit since the subpanel sees this 'packaged' solution as equivalent to what it was receiving before from the service (grid) panel. One thing to note is that we now have a derived system and two G/N bonds. Not sure what article 250 says about this.


Closing thoughts​

  • I'm not a huge fan of relying solely on the ground through the inverter for the derived system. I feel like a superior design would be an additional grounding conductor tying the grounding buss of the derived main panel to the service panel. I'm also making the assumption that there is no ground switching tomfoolery happening within the Growatt device.
  • This design doesn't have no lightning arrester. I believe the best practice is just a two-pole CB, each phase headed into the MOV, which passes potential through to ground. I think the best place for this is within the derived main panel.

Thanks all for the feedback ?
 

Attachments

  • derived_main_panel.png
    derived_main_panel.png
    148.7 KB · Views: 202
I would drive a separate ground rod for the secondary panel, just like you would for a subpanel in an outbuilding.
 
I am using 3x growatt 5000es with the solaredge midpoint transformer (autotransformer is a variable transformer, they should not use that name for this product) and the way I have wired mine up is to feed a 240v-only panel using the growatts output, then feed the midpoint transformer and a 2nd breaker panel from that 240v-only panel using a single 2-pole 30amp breaker. The 2nd panel is for all my 120v loads (though you could use it for 240v loads, its best to keep them on the first panel where more current is available). The L1 and L2 from this breaker connect to the transformer and the 2nd panel, and the neutral from the transformer comes into the 2nd panel as the neutral wire. This provides my 240v loads with all the required power, and the 120v loads are capped at about 5kw that the midpoint transformer can handle. If the breaker trips for the transformer, all 120v power will be disconnected and there will not be a floating neutral. With all the 120v loads in my house connected to it (4 people with computers and TVs, kitchen appliances, lights, etc.), the solaredge transformer doesn't seem to have any trouble.

In the 240v panel I use the neutral bar as the ground, and I bonded ground and neutral in the 120v panel because this is where that neutral originates (this is not a subpanel of the main utility panel at this point, so I do bond ground and neutral to keep them at the same potential). The ground at both panels and across all inverters is still bonded with utility ground in my house, otherwise I would need to add a ground rod.

The main downside I can think of to this set up is if you have a large load that requires L1, L2, neutral, and ground connection, like an outdoor RV connection, or a dryer with a 4-prong connection (240v for the heating element, 120v for small electronics like control panel). You would basically need a separate midpoint transformer for any large loads like this that require a lot of current. I'm trying to think through whether it would be safe to connect multiple midpoint inverters in parallel to the same 2nd breaker panel... I think that it would be, since if one transformer's breaker tripped due to overcurrent then it would just trip the 2nd transformer's breaker as well, and you shouldn't ever be left with power and a floating neutral.
 
Last edited:
A 240V dryer uses 240V for the heating element, 120V for the motor, and 120v to a DC transformer internally for the control board and electronics.

A heating element is around 5800W (24.17 A at 240 VAC) the motor is maybe 1HP (~750W). You'd need two of the 5000W Growatt inverter, but a single transformer.

You can run A LOT of 120V loads on one transformer. Pretty much anything you'd want in a 3000 sqft house.
Electric space heaters would be an issue. But they are in general.
 
Thanks @Coty and @Desert_AIP - great info (y)

@Coty - splitting loads into separate panels by 120V and 240V, and limiting the 120V panel to 30A (current limit of the AT) is a very elegant and simple way to do things - great idea. For your information, I believe it is possible to run multiple midpoint transformers in 'parallel' to increase the neutral-forming capacity of a subpanel. Ben's solar and battery (added the link to my first post) is doing this with his sol-ark 12k.

Your idea to limit the 120V panel to 30A feed is great, but I think it may have an issue if the AT fails internally (30A 'main' breaker does not trip, and neutral no longer forms correctly - i.e. one leg shows 180v and the other leg shows 60v...not sure if this is physically possible). In this failure scenario, it seems we'd have an over/under voltage issue that could cause problems.

I reworked my previous design a bit and I think it is a) able to catch that failure mode, and b) allows 240V and 120V loads to be mixed on the same subpanel, since the 120V legs are monitored separately. Downside of my design is that if you 'lose' your AT, you also lose all your 240V loads at the same time.

Updates over the previous iteration (see attached diagram):
  • Moves the neutral safety logic components to the bottom section of the derived main panel to simplify wiring
  • Adds a one-to-many DIN terminal block for the neutral conductor of the neutral safety logic components (simplify wiring)
  • Adds 2A fast-blow fuses to the feeds of the over voltage protection/monitoring circuits of the two AT legs
  • Removes the 50A double pole CB on the AC OUT connection - this isn't needed since all branch circuits are fused, and removal allows use of a more common/cheaper 6-space loads panel
  • Moves the ground-ground bonding between the service panel and the derived main panel from traveling through the Growatt to the ground/neutral busses. This allows calculation of grounding conductor sizing since grounding conductors through the Growatt do not appear to have their ampacities documented
 

Attachments

  • derived_main_panel_v2.png
    derived_main_panel_v2.png
    156.3 KB · Views: 171
The L1 and L2 from this breaker connect to the transformer and the 2nd panel, and the neutral from the transformer comes into the 2nd panel as the neutral wire. This provides my 240v loads with all the required power, and the 120v loads are capped at about 5kw that the midpoint transformer can handle. If the breaker trips for the transformer, all 120v power will be disconnected and there will not be a floating neutral.
I like your idea of separating the 240V and 120V circuits into two panels and I get the idea of setting up the system to avoid the problems associated with the transformer failing or its circuit tripping,, but I don't understand how you make the connection(s) with the transformer. Normally, as I understand it, the transformer is connected to one circuit breaker in one panel with the L1 and L2 lines, with the ground and neutral going into their respective bus bars in that same panel. But you seem to be saying that the transformer is connected to two panels, that is, both the 240V panel and the 120V panel. Is that right, and if so, how exactly do you establish those connections?
 
The L1 and L2 of the transformer are the same L1 and L2 from the inverter output.
The transformer is really only outputting the neutral. Power passes through it.
It's a 1:1 transformer, equal number of windings on the primary and secondary loops.
240V in and 240V out. With a center tap N to allow 120V operation, which is why you see a pair of windings inside the case.
There's simply no output terminals from the transformer.
So a separate 120V panel would have the bus bars L1 and L2 energized by the Growatt output (through a Circuit breaker from the 240 panel). The transformer would be connected to L1 and L2 of the 120V panel through a two pole breaker in that panel, with its N output attached to the N bus and bonded to ground in that panel only.

A 120v device fed from a single pole breaker is drawing power from the growatt through the transformer via the neutral.
The path flow from L1 or L2, through the transformer to N. That's how it is possible to feed one side while the other has no load
Thanks for the response. One more follow-up question. In this set-up, if the transformer circuit trips, doesn't that turn the 120V panel into a 240V panel (since the 120V panel is being energized from the 240V panel on its own circuit)? Doesn't that put us back to the same problem that David Poz has in the link from the original post in this thread?
 
I don't see a link to David Poz, I do see a mention of a problem he has.

Also, I apologize, I missed the key element "operate in utility first" in the OP.
My neutral bonding advice assumed the utility was for battery charging only and that all loads downstream of the Growatt would be from the inverter.

I'll have to noodle on the bonding question when the grid is passed through the Growatt. The pass-through is 240V and there's still no Neutral brought over, but I need to think about the transformer panel.
 
The neutral safety disconnect circuit is a great idea!
This makes your 120V Loads on the Sub Panel safer than a 120 load connected to the main service Panel, being that loss of Neutral can also happen on the Main Service. Search Google for "loss of neutral wire".
CON: How much power will your contactors draw being energized 24/7?​
 
According to NEC article 100, a separately derived system is:
“An electrical source, other than a service, having no direct connection(s) to circuit conductors of any other electrical source other than those established by grounding and bonding connections.”
What makes this tricky is that the diagram I referenced in post #13 operates in two states:
  1. On-grid (non-separately derived system). GW L1/L2 of AC IN is directly connected to L1/L2 of AC OUT. System is not a SDS in this state, even though we have two neutral forming components (grid/pole transformer, and the SolarEdge midpoint transformer), and two G/N bonds (one in the service panel, and another in our "derived main panel"
  2. Off-grid (separately derived system). GW L1/L2 of AC IN is not directly connected to L1/L2 of AC OUT due to the internal ATS (switch flipped due to loss of grid voltage). In this state, GW is classified as a separately derived system because there are no circuit conductors directly connected to any other electrical source (grid) other than grounding/bonding connections. Note: if we connected the SE transformer neutral to our service panel neutral, we would no longer classify as a SDS.
State #2 above seems copasetic. State #1 is the tricky one since we have two active system and two active G/N bonds. I believe state #1 is still OK due to the fact that the effective ground fault current path through the SE midpoint transformer is through the G/N bond of the derived main panel. See Figure 250-4 of Mike Holt's Article 250 study guide here (screenshot below):

1635176295866.png

Appreciate thoughts on this
 
Last edited:
@LeRoyK - I've looked up the contactor I've spec'd and it's about 600mA @ 120VAC. I'll need to test to be sure. I spec'd 2A for the DIN fast-blow fuses for some headroom because the coil is an inductive load.
 
'Option #1' theory that the AT and grid transformers would parallel in an on-grid state (potentially catastrophically balancing load for other customers),
I think your Diagram "derived_main_panel_v2.png" still puts your AT in parallel with the grid transformer when in On-Grid Mode. Because the AT neutral is connected to the main service panel neutral through the ground wire from Main Service Panel to Derived Main Panel.
 

Attachments

  • derived_main_panel_v2_AT Parallel.jpg
    derived_main_panel_v2_AT Parallel.jpg
    212.3 KB · Views: 75
Thanks @LeRoyK and yes that is a great point. I think there are two solutions to this perceived problem:
  1. Drive a separate GEC for the 'derived main panel'. This allows a permanent G/N bond on the derived main panel and changes the grid neutral return conductor path (the blue line on your diagram) from a low impedance one to a high impedance one (through the dirt).
  2. Dynamically switch neutral conductors of the derived main panel from grid neutral (when on grid) to SE neutral (when off-grid). This is possible (see option #2 of post #3) but has technical design challenges that are (imo) not attractive compared to driving a new GEC.
 
Back
Top