diy solar

diy solar

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

Okay - so I updated the wiring diagram a bit.

Changelog:
  • Adds additional neutral safety circuit - both legs are monitored high/low for redundancy
  • Adds terminal blocks for all connections
  • Removes 15A single pole CB within 'Inverter Disconnect Service Panel'
    • Frees up two spaces on a 6-space load center (yay) allowing a 'failsafe' 240VAC load to be run in the event of neutral failure
    • Changes all contactors from 120VAC coil to 240VAC coil (simplifies wiring, better part availability)
  • Changes dry contact to switch 240VAC instead of 120VAC
    • Routes dry contact wire through conduits #4 and #2 instead of #6 (removed in this version), simplifying conduit runs
As you can see, things get a little hectic when the terminal blocks are fully laid out. We need six terminal blocks in our 'Switching Logic Enclosure':
  • Neutral (6 conductors)
  • Ground (4 conductors)
  • L1 IN (6 conductors)
  • L2 IN (6 conductors)
  • L1 OUT (3 conductors)
  • L2 OUT (3 conductors)
View attachment 72518

Please review the above work and let me know if there are any questions/problems. We've got a lot of conductors in conduits #4 and #2. I'm thinking a trough/gutter may be the best move here for physical layout. To answer your question @automatikdonn - this design can carry 80A of ground fault current, which (to my knowledge) only needs to be 63A by code ( >= 1.25 * our 50A DP OCPD feed within 'Inverter Disconnect Service Panel').

Next steps:
  • Review above schematic for issues
    • Neutral safety circuit
    • Ground fault current calculations
  • Spec conductors & conduit
  • Verify conductor fill limits for conduit spec'd
  • Physical layout and BOM
The only thing I don't understand is why each leg is going through it's own contactor? I have two contactors in my circuit because there are two AT's and two inputs. I will get the cover off that other panel today so everyone can see and give me some feedback. I don't want to do dumb things and more eyes are always helpful to prevent that. I am not a sensitive person, so please anyone who has constructive criticism of the design and wiring.. let er rip.
 
Thanks @automatikdonn and good question - the reason for that is simply because we're more likely to lose a contactor coil (both throws of a single contactor) than we are to lose half of a contactor (one throw only). In this failure case, we lose half of our split phase on the 'Inverter Load Service Panel', and - assuming we are running < 32A single-pole breakers, our 120V loads will either a) be disconnected completely, or b) remain operational, but still maintain fault capacity to trip their OCPD.

I've made an additional update that simplifies this a bit by chaining the contactors together through the voltage protection devices. This makes the system all-or-nothing (ideally), such that both contactors are either fully engaged or fully disengaged (effectively acting like a 4PDT, with each pole carrying 40A).

I also added in where I'd recommend putting the scope leads when simulating a neutral failure (and on-grid to off-grid) transition. Ultimately the goal is to measure what the voltage spike/dip would be for 120V devices in the 'Inverter Load Service Panel'.ES5K_4A.png
 
Last edited:
Thanks @automatikdonn and good question - the reason for that is simply because we're more likely to lose a contactor coil (both throws of a single contactor) than we are to lose half of a contactor (one throw only). In this failure case, we lose half of our split phase on the 'Inverter Load Service Panel', and - assuming we are running < 32A single-pole breakers, our 120V loads will either a) be disconnected completely, or b) remain operational, but still maintain fault capacity to trip their OCPD.

I've made an additional update that simplifies this a bit by chaining the contactors together through the voltage protection devices. This makes the system all-or-nothing (ideally), such that both contactors are either fully engaged or fully disengaged (effectively acting like a 4PDT, with each pole carrying 40A).

I also added in where I'd recommend putting the scope leads when simulating a neutral failure (and on-grid to off-grid) transition. Ultimately the goal is to measure what the voltage spike/dip would be for 120V devices in the 'Inverter Load Service Panel'.View attachment 72588
I also wanted to share this contactor I found - https://www.amazon.com/gp/product/B07NHJN59Y/ref=ppx_yo_dt_b_asin_title_o08_s00?ie=UTF8&psc=1

It's a bit too big to fit in my little 18din boxes, but its rated for 50amp and UL listed. My concern with it is the coil voltage shows 200-220... not sure if it would be fine at 240.
 
Well - if we wanted to limit the output of the GW to 5kW (20.8A), we could just replace our 50A breaker with a 30A one in the 'Inverter Service Disconnect Panel'. That'd allow us to use the single 40A contactor I have in the schematic, which is > 37.5A fault current we'd need for the 30A upstream CB, and is also UL listed.

Downside of that is that I know you want to parallel your GWs, so you'd need a way to scale up that contactor. If you wanted to split your load panels (240 and 120), then you could simplify this a bit and limit the 120V devices to 30A, and omit a contactor entirely for the 240V loads. This may be the way to go since it'd scale past 1 GW unit. This actually works out perfectly because we have two poles free in our 'Inverter Service Disconnect Panel' (N/C breakers), and we could stick 50A breakers there to power a 240V panel and not be limited by our contactor fault current.
 
Last edited:
Well - if we wanted to limit the output of the GW to 5kW (20.8A), we could just replace our 50A breaker with a 30A one in the 'Inverter Service Disconnect Panel'. That'd allow us to use the single 40A contactor I have in the schematic, which is > 37.5A fault current we'd need for the 30A upstream CB, and is also UL listed.

Downside of that is that I know you want to parallel your GWs, so you'd need a way to scale up that contactor. If you wanted to split your load panels (240 and 120), then you could simplify this a bit and limit the 120V devices to 30A, and omit a contactor entirely for the 240V loads. This may be the way to go since it'd scale past 1 GW unit. This actually works out perfectly because we have two poles free in our 'Inverter Service Disconnect Panel' (N/C breakers), and we could stick 50A breakers there to power a 240V panel and not be limited by our contactor fault current.
That downside is exactly why my setup is built the way it is, so I have been noodling ways around it that make sense. I don't want to depend on the output of a single GW to power the transformer, but the way I am doing it now also is incorrect. It's kinda like I want to have my cake and eat it too. It almost seems easier to do the monitoring circuit further downstream and then have it kill the output from the GW upstream. Like put the contactor inline directly after the output breaker on the GW, but the monitoring part of the circuit lives at the point where its all brought together in parallel.
 
I've made an additional update that simplifies this a bit by chaining the contactors together through the voltage protection devices. This makes the system all-or-nothing (ideally), such that both contactors are either fully engaged or fully disengaged (effectively acting like a 4PDT, with each pole carrying 40A).
I think you need to use 120V coils on your "Lost N Contactors" if you plan on wiring them in series as shown on your last schematic.
 
I think you need to use 120V coils on your "Lost N Contactors" if you plan on wiring them in series as shown on your last schematic.
We are pulling from the GW outputs to drive the contactor. The neutral is only in circuit for monitoring purposes. You have to look closely to see though. That neutral never comes through and only lives on the input side.

outputs.png
 
@LeRoyK - you're totally right, I was noodling that one. Updated in the attached.
@automatikdonn - I've separated the 120/240 and 240 load centers in the attached (but kept the parallel contactors, which could continue to be paralleled and scaled).Growatt 5000ES with Neutral disconnect - Separate 120_240 Load Panels.png
 
Why did you change your "Lost N contactor" to switch both L1 and L2 on each contactor instead of one contactor switching L1 and the other switching L2 as you had it previously?

I think there is a chance that the two contactors will not disconnect simultaneously, so the last contactor to disconnect will have to break the full load.
 
This reminds me of some Schneider inverters that I worked on a couple of years ago. we had 3 inverters in parallel each rated at 60A pass-thru current. We burned up the transfer relay in one of the units.

After doing some research I found this document.
 

Attachments

  • Schneider 6848 XW+ Internal Transfer Relays.pdf
    300.2 KB · Views: 24
Thanks @LeRoyK and great point - although in this specific case I think we're still OK because our upstream OCPD in the 'Inverter Disconnect Service Panel' is 30A (downrated from 50A) so even if one contactor switched all the load, we'd still be within the 37.5A fault current operating range with even a single contactor (rated at 40A per pole).

If our upstream OCPD is > 30A though, then yes I'd agree that the design has an issue. Either way, I think it's a good idea to group phases by contactor as you said - this makes it so there are no design philosophy changes as the upstream OCPD goes above 30A.

Updated the schematic to reflect this change (y) Growatt 5000ES with Neutral disconnect - Separate 120_240 Load Panels.png
 
Thanks @LeRoyK and great point - although in this specific case I think we're still OK because our upstream OCPD in the 'Inverter Disconnect Service Panel' is 30A (downrated from 50A) so even if one contactor switched all the load, we'd still be within the 37.5A fault current operating range with even a single contactor (rated at 40A per pole).

If our upstream OCPD is > 30A though, then yes I'd agree that the design has an issue. Either way, I think it's a good idea to group phases by contactor as you said - this makes it so there are no design philosophy changes as the upstream OCPD goes above 30A.

Updated the schematic to reflect this change (y) View attachment 72615
If you reduce the upstream OCPD to 30A then you only need one "Lost N Contactor".
 
Agree - simplified schematic with single contactor, 30A max 120/240 load panel:
  • L1/L2 input to AT does not disconnect from 'Inverter Disconnect Service Panel' in the event of a neutral leg failure
    • No cold start process needed since AT input never disconnects
  • L1/L2 output from 'Lost N Safety Contactor' to '120v/240v Load Panel' is disconnected in the event of a neutral leg failure
  • 30A limit to 120/240v Load Panel
Growatt 5000ES with Neutral disconnect - Separate 120_240 Load Panels & Single Contactor.png

Would like to create a similar schematic for the GW x2 and x3 and xn editions, as time permits...
 
Last edited:
Here is a photo of the panel you wanted a peek into.

Starting from left to right I have
Inverter 1 50amp output breaker
Inverter 2 50amp output breaker
Inverter 3 50amp output breaker
AT 1 30amp Input breaker
AT 1 30amp Output breaker
AT 2 30amp Input breaker
AT 2 30amp Output breaker
Dinkle Connector Set 1 from Inverter 1 50 Amp output to 30 Amp AT1 Input (This part of the circuit is where I have issues and needs to be corrected)
Dinkle Connector Set 2 from Inverter 2 50 Amp output to 30 Amp AT2 Input (This part of the circuit is where I have issues and needs to be corrected)
240V POCO Mains monitor (This is for future use)

IMG_20211117_180842.jpg


I did want to also answer the ferrel question - https://twcontrols.com/lessons/wire...lated-vs-non-insulated-and-ul-508a-guidelines

UL 508A 29.3.4 of the Wiring Methods section covers the use of ferrules in a UL control panel.

It states:

  • 29.3.4 A connection to a terminal of a component shall be made by:
    • a) Wire inserted directly into a pressure wire terminal of the component;
    • b) Quick-connect terminal of the component, where the mating part is provided with a dimple, depression, or spring-type connection such that a mechanical snap-action connection is made that does not rely solely upon friction between the two parts;
    • c) Crimped-on pressure terminal connector or closed-loop eyelet;
    • d) Solder terminal specified in 29.3.2;
    • e) Wire-binding screw specified in 29.3.3;
    • f) Open-type eyelet specified in 29.3.5; or
    • g) Wiring ferrule specified in 29.3.6.
G of UL 508A 29.3.4 refers us to 29.3.6 for the specifications of how ferrules are to be used in an industrial control panel stating:

  • 29.3.6 A wiring ferrule shall be:
    • a) Used with stranded copper wire(s) only;
    • b) Terminated in a connector rated for copper wire and rated for the number and size of wire(s) crimped to the ferrule;
    • c) Crimped with an appropriate tool as recommended by the ferrule manufacturer before terminating in a terminal of a component;
    • d) Sized in diameter appropriate for the number of wires and wire size(s) as recommended by the ferrule manufacturer; and
    • e) Crimped to the wires such that the length of the uninsulated portion of the wires does not result in the reduction of electrical spacings when the ferrule is installed.
 
Last edited:
You could simplify things by putting the 240V loads on the "Inverter disconnect Service Panel" instead of having another separate Load center just for 240V.
 
You could simplify things by putting the 240V loads on the "Inverter disconnect Service Panel" instead of having another separate Load center just for 240V.
In my head it makes sense to turn the whole system off if there is a malfunction. Specifically on the input side of the AT, if its broken then I don't want to keep providing it power. At least that was my line of thinking.
 
In my head it makes sense to turn the whole system off if there is a malfunction. Specifically on the input side of the AT, if its broken then I don't want to keep providing it power. At least that was my line of thinking.
In the current schematics, you guys are not disconnecting the input side of the AT, only disconnecting the loads from the AT.
 
When you do this remember my "burned up the transfer relay" experience.
I am guessing as you scale it up with multiple GW you will end up using a bigger contactor rather than paralleling multiple contactors.
I agree, and that is what I was afraid of. In my current state of having 3 in parallel I would need a 100 amp minimum contactor. It's probably the best architecture, but $$$$

For instance these two contactors are pretty close

120V Coil - https://www.mcmaster.com/6564K795/
240V Coil - https://www.mcmaster.com/6564K796/
 
Last edited:
In the current schematics, you guys are not disconnecting the input side of the AT, only disconnecting the loads from the AT.
You are not wrong. We need to shut down the outputs of the Inverter sooner. This is why I have a prototyping setup so we can work through these issues and get a resolution out the door.
 
Back
Top