diy solar

diy solar

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

Feel free to reply over here:

 
I'm pretty sure as long as you don't go over max OC voltage, you can overpanel.
The system only draws so much current.
If the panel array is "capable" of providing a higher level of current, they won't "push" it down the line.
So the wattage limits are the limit of the inverter/charger, not what you attach to it.
 
I'm pretty sure as long as you don't go over max OC voltage, you can overpanel.
The system only draws so much current.
If the panel array is "capable" of providing a higher level of current, they won't "push" it down the line.
So the wattage limits are the limit of the inverter/charger, not what you attach to it.
 
They miscalculated the max string voltage in that reply.
They used the MPPT voltage of 34VDC to arrive at 102VDC for a string of 3 panels.
Rather than the OC voltage of 40.7VDC to arrive at a string voltage of 122.1VDC.

I don't agree with that opinion.
Current doesn't work that way.
What happens when the batteries are full and there are no loads?
The system stops drawing power from the panels.
Plenty of discussion around here regarding over paneling.

Why doesn't my 20A wall outlet circuit destroy the 120W appliance I plug into it?
The circuit is "capable" of supplying 20A, but the appliance only draws 1A.

I use a victron 70/15 charger to charge my camping fridge battery from the car 12V system.
I have to run it through a 24V boost converter to get a high enough voltage differential to get it to start charging.
The booster is rated at 30A on the 12V side. The charger takes the 24V input and converts it to 12V nominal through the MPPT and limits its current draw to 15A. The 12V system of the vehicle has a lot higher capacity to supply more than 15A, but the appliance, the charger, is limiting its draw.
 
Last edited:
They miscalculated the max string voltage in that reply.
They used the MPPT voltage of 34VDC to arrive at 102VDC for a string of 3 panels.
Rather than the OC voltage of 40.7VDC to arrive at a string voltage of 122.1VDC.

I don't agree with that opinion.
Current doesn't work that way.
What happens when the batteries are full and there are no loads?
The system stops drawing power from the panels.
Plenty of discussion around here regarding over paneling.

Why doesn't my 20A wall outlet circuit destroy the 120W appliance I plug into it?
The circuit is "capable" of supplying 20A, but the appliance only draws 1A.

I use a victron 70/15 charger to charge my camping fridge battery from the car 12V system.
I have to run it through a 24V boost converter to get a high enough voltage differential to get it to start charging.
The booster is rated at 30A on the 12V side. The charger takes the 24V input and converts it to 12V nominal through the MPPT and limits its current draw to 15A. The 12V system of the vehicle has a lot higher capacity to supply more than 15A, but the appliance, the charger, is limiting its draw.
The problem was the GW was rated at 7Kw input. What occurred was due a large load, both supplying the inverter and charging the battery banks, more than 7Kw was coming into the unit as the loads were high enough. The supply of watts was high, over 7Kw, and the unit kept drawing it.

You're confusing a limited number of watts being drawn (an appliance) with drawing as many watts as possible due to a large load capable of drawing a huge number of watts thru a device rated at a maximum capacity. It's no different than putting a 30a device on a 14 gauage, 15a rated circuit and breaker, it will draw the 30a, overloading the wire and tripping the breaker if the breaker is correct for the wire size. But if that breaker is replaced with a 30a breaker, the wire will overheat with a 30a load and burn up.
 
On the neutral issue:

Wouldn't using a isolation transformer to create the neutral get rid of the problems associated with the autotransformer?

You would just use the output of the isolation transformer to feed your loads and ground the neutral. If there was a fault on the input, the output would still have a neutral.
 
Thanks @Trip.Diamond for the AT schematic, I looked everywhere for this - dunno where you found it but I'm glad you did. Lessons learned:
  • AT never participates in neutral forming with the grid transformer (despite neutral connection between the AT and grid) due to the XOR created by the 'Automatic Switchover' of L1&L2 in the schematic.
  • SolarEdge, in their schematic, uses the grid transformer to make the neutral when operating on-grid, and the AT to form the neutral when operating off-grid. Our designs, #133 and #159, use the AT to form the neutral for both on and off-grid.
  • The SolarEdge schematic shows a neutral running back to the 'black box' inverter object. I cannot find a reason why this would be needed other than to monitor the 120v legs with respect to L1/L2. This leads me to believe that the SE StorEdge solution monitors AT neutral forming voltages and is able to 'fail off' by cutting production of L1/L2 via the inverter output.
  • The schematic you posted has the added benefit of an AC 'Manual Bypass' switch. This could be added to our designs if needed, but would really only be helpful in the event that the GW internal ATS failed (might be prudent in systems that put many cycles on the internal ATS)


I did some qualifying work on the internal temp sensor for the AT by performing a temp sweep and measuring the resistance change of the sensor. I had two PT100 temp sensors in the transformer - one at ambient temp and one firmly bolted to the iron core. When the ambient temp was within a degree or two of the iron core temp, I considered the system at temp equilibrium. It took hours to temp saturate the iron core - it has a very high thermal mass.

I'm led to believe the OEM temp sensor is a switching PTC thermistor for the following reasons:
  • Low resistance ~2.4Ohm at room temp that stays low from 16C up to about 95C (a PT100 RTD would start with a higher resistance, and also vary across that range)
  • There appear to be copper leads from the sensor, as well as SE document stating copper leads can be used from DIN block back to StorEdge terminal block. This rules out thermocouples since they'd need thermocouple extension wire specified to avoid creating an additional metallic junction.
A switching (not silistor) PTC thermistor should have a curie temp ('switch temperature') at which resistance sharply increases. Based on datasheets of several similar thermistors, this appears to generally be in the 140-200C range. I only went up to 95C in my qualification test.

Screen Shot 2021-12-13 at 3.10.27 PM.png
During the test, the resistance of the thermistor consistently read 2.4Ohms:
Screen Shot 2021-12-13 at 3.00.04 PM.png
I decided to call it at about 95C (200F) as things were getting about as toasty as I'd ever want them to get in service. Unfortunately, I'd have to unwind the transformer to isolate the temp sensor since it's between the windings and the iron core. My suspicion is that SolarEdge has a temp shutoff of about 150C (300F) for this unit, near the transition temp of the thermistor.

Since I didn't want to disassemble the AT to test this (unless somebody wants to buy me another AT), I elected to:
  • Get the closest proxy measurement I could, and
  • Change the sensor response from a boolean to a continuous one
A good choice for this would be a PT100 RTD or K-type thermocouple. I had a thermocouple on hand, so I drilled and tapped the case of the AT to allow me to directly monitor the iron core temp. This isn't a perfect sensor location (the OEM one just about is, albeit it only monitors one winding instead of both, which is less than ideal), but I'm able to detect an overtemp condition much earlier than the factory sensor, so a lagging response is more than made up for with an earlier threshold trigger. The iron core is thermally massive so temp fluctuations are very slow.

Screen Shot 2021-12-13 at 3.01.09 PM.png
This is then pressed into the core of the transformer via the threaded connection, in conjunction with thermal paste to aid in conductivity. The whole assembly is grounded and out of the way of other transformer conductors to mitigate faults.
Screen Shot 2021-12-13 at 3.22.38 PM.png
This should satisfy temp monitoring used by SolarEdge and provide a conservative leading metric to transformer failure. The plan is to mate this with a neutral current transducer as a 'soft breaker' that operates redundantly to the panel OCPD for the AT, allowing alerting to the user to reduce 120V phase imbalance prior to meeting either an over-temp or over-current fault condition.
 
Last edited:
The problem was the GW was rated at 7Kw input. What occurred was due a large load, both supplying the inverter and charging the battery banks, more than 7Kw was coming into the unit as the loads were high enough. The supply of watts was high, over 7Kw, and the unit kept drawing it.

You're confusing a limited number of watts being drawn (an appliance) with drawing as many watts as possible due to a large load capable of drawing a huge number of watts thru a device rated at a maximum capacity. It's no different than putting a 30a device on a 14 gauage, 15a rated circuit and breaker, it will draw the 30a, overloading the wire and tripping the breaker if the breaker is correct for the wire size. But if that breaker is replaced with a 30a breaker, the wire will overheat with a 30a load and burn up.
I didn't consider the load passthrough as an additional load to the charger.

Still surprised they designed it to draw a load higher than it can handle.
Whether the solar input goes to inverter loads or charging, the thing should not be able to draw beyond its design rating.
 
Thanks @Trip.Diamond for the AT schematic, I looked everywhere for this - dunno where you found it but I'm glad you did. Lessons learned:
  • AT never participates in neutral forming with the grid transformer (despite neutral connection between the AT and grid) due to the XOR created by the 'Automatic Switchover' of L1&L2 in the schematic.
  • SolarEdge, in their schematic, uses the grid transformer to make the neutral when operating on-grid, and the AT to form the neutral when operating off-grid. Our designs, #133 and #159, use the AT to form the neutral for both on and off-grid.
  • The SolarEdge schematic shows a neutral running back to the 'black box' inverter object. I cannot find a reason why this would be needed other than to monitor the 120v legs with respect to L1/L2. This leads me to believe that the SE StorEdge solution monitors AT neutral forming voltages and is able to 'fail off' by cutting production of L1/L2 via the inverter output.
  • The schematic you posted has the added benefit of an AC 'Manual Bypass' switch. This could be added to our designs if needed, but would really only be helpful in the event that the GW internal ATS failed (might be prudent in systems that put many cycles on the internal ATS)

Out of curiosity, what are the concerns about the AT remaining connected to the grid neutral when the GW switches to on-grid?
 
Out of curiosity, what are the concerns about the AT remaining connected to the grid neutral when the GW switches to on-grid?
Just a quick response to this one.
If you are absolutely sure every house in you neighborhood is perfectly grounded and neutral bonded then nothing should happen...
Can you really be sure of that?

Otherwise; you are putting your AT in parellel with the Grid transformer and your AT can then be "called on" to balance half the imbalance load of the Grid transformer (a typical pole mounted Grid Transformer is designed to handle about 50-100 times your little AT). This would hopefully just trip for 30A breaker saving your AT, but also leaving you without a neutral for when your system switches back to Solar/Battery operation. Worst case is melts down your breaker, AT... and everything around it...
Either way not a risk I would want to take for long term operation.

Note: I am currently working on and testing a design to work with Grid Neutral connected that will automatically switch between Sol/Bat and Grid operation. I am having good success and will post a larger write up and wiring diagram when I'm sure it really works as planned.

USUSUS - Thanks for your hard work I hope you can crack the temp sensor problem. It will solve the safety feature missing to us GW users.
 
Just a quick response to this one.
If you are absolutely sure every house in you neighborhood is perfectly grounded and neutral bonded then nothing should happen...
Can you really be sure of that?

Otherwise; you are putting your AT in parellel with the Grid transformer and your AT can then be "called on" to balance half the imbalance load of the Grid transformer (a typical pole mounted Grid Transformer is designed to handle about 50-100 times your little AT). This would hopefully just trip for 30A breaker saving your AT, but also leaving you without a neutral for when your system switches back to Solar/Battery operation. Worst case is melts down your breaker, AT... and everything around it...
Either way not a risk I would want to take for long term operation.

Note: I am currently working on and testing a design to work with Grid Neutral connected that will automatically switch between Sol/Bat and Grid operation. I am having good success and will post a larger write up and wiring diagram when I'm sure it really works as planned.

USUSUS - Thanks for your hard work I hope you can crack the temp sensor problem. It will solve the safety feature missing to us GW users.
Is this theoretical, or is there a test one can do to see if the AT is truly being used to balance your neighbor's loads? Perhaps putting a clamp meter on the Auto-trans. N-line? As I understand, every house has there own LI/L2/N lines going to their own panel; and I would think that the electricity being used by loads in those panels go back to their original source (i.e. transformer on the pole), so how can loads connected to their own neutral, used by your neighbor, want to be balanced by an AT inside someone else's house? Perhaps someone knows someone that this happened to, if so please share.

I look forward to see your progress, I'm learning a lot on this forum.
 
Instead of using the 100A contactor what about using a breaker with a trip coil.


if you lost neutral it means something bad has happened and you should be looking at it any way. My system is wired with interlocking breakers on the load panel so that all I need to do is open my feed breaker from the inverters and close the breaker from the grid and I’m back in operation again till whatever happened can be resolved.
 
Current thru center tap Neutral from transformer


Can this also happen from an autotransformer?
 
Current thru center tap Neutral from transformer


Can this also happen from an autotransformer?
What most likely is happening is farther up the utility line, there is either resistance to ground at a transformer or someone is backfeeding the neutral.

It happens often in rural areas, you find stray voltage and current from other points on the line. Lines run from a substation and carry power to several locations many miles apart. But you will find buildings wired back in the 40's and 50's and poor grounding of transformers as things age. Many a dairy farmer has had stray voltage affect their herd and even rewiring the whole farm, they still have the problem. The thing is, it's not their service that is causing the problem, it is someone else farther up or down the line.
 
Okay I overhauled the wiring schematic a bit. In retrospect, I am not confident that designs #133 and #159 would clear ground faults while in on-grid mode. This should resolve that issue, as well as incorporate additional functionalty.

Changelog:
  • Adds temp monitoring to AT
    • Through K-type thermocouple (continuous measurement) - pictured
    • Through OEM switching PTC thermistor (?) - not pictured, optional via NodeMCU ADC
  • Ties 120/240V Load Panel neutral permanently to Main Service Panel neutral, eliminating need for G/N bonding contactor (StorEdge wiring method)
  • Removes physical cold start push button
  • Adds AT neutral current, L1/N, and power factor sensor
  • Adds conductor/conduit labels for wiring schematic/BOM
  • Changes dry contact signal from 240VAC sensing to 3vDC sensing (logic level)

Growatt 5000ES with NodeMCU Control.png

The most important thing this changes is it gives the ability to log sensor data, detect trends, and throw alarms at more conservative values. Since we are now in control of all contactors within the system via software, we can do things like:
  • Attended, remotely attended (via wifi/internet), or unattended cold start
  • Controlling the switchover time of the cold start (how long we want to wait after we stabilize the AT before we allow 120V to our loads panel via the lost N safety contactor)
  • Disconnecting power to the AT, the loads panel, or both - contingent on any combination of software-configurable measured conditions, such as:
    • Over neutral current
    • Power factor
    • Over/under voltage
    • Over/under temp (either via built-in AT thermistor (?), K-type thermocouple, or both)
Things about this design I'd like to improve:
  • Low voltage (3/5/12VDC Arduino/NodeMCU) components are in the same enclosure as high voltage (120/240VAC) components and may be at risk of EMI
  • Thermocouple sensor lines 3.5 and 3.6 are run in the same conduit as L1/L2/N for the AT. Although the sensor cable is shielded, I am not sure if this will affect readings.
  • PZEM-004T current sensor has a 1s response time. It's suitable for slower moving trends, but not fast enough to catch over/under voltage conditions (the SVP-912 is 0.1s, in comparison). If we had a sensor with a faster response time, we could get rid of the 2qty SVP-912 devices entirely and control everything with the NodeMCU
 
Last edited:
Okay I overhauled the wiring schematic a bit. In retrospect, I am not confident that designs #133 and #159 would clear ground faults while in on-grid mode. This should resolve that issue, as well as incorporate additional functionalty.

Changelog:
  • Adds temp monitoring to AT
    • Through K-type thermocouple (continuous measurement) - pictured
    • Through OEM switching PTC thermistor (?) - not pictured, optional via NodeMCU ADC
  • Ties 120/240V Load Panel neutral permanently to Main Service Panel neutral, eliminating need for G/N bonding contactor (StorEdge wiring method)
  • Removes physical cold start push button
  • Adds AT neutral current, L1/N, and power factor sensor
  • Adds conductor/conduit labels for wiring schematic/BOM
  • Changes dry contact signal from 240VAC sensing to 3vDC sensing (logic level)

View attachment 76400

The most important thing this changes is it gives the ability to log sensor data, detect trends, and throw alarms at more conservative values. Since we are now in control of all contactors within the system via software, we can do things like:
  • Attended, remotely attended (via wifi/internet), or unattended cold start
  • Controlling the switchover time of the cold start (how long we want to wait after we stabilize the AT before we allow 120V to our loads panel via the lost N safety contactor)
  • Disconnecting power to the AT, the loads panel, or both - contingent on any combination of software-configurable measured conditions, such as:
    • Over neutral current
    • Power factor
    • Over/under voltage
    • Over/under temp (either via built-in AT thermistor (?), K-type thermocouple, or both)
Things about this design I'd like to improve:
  • Low voltage (3/5/12VDC Arduino/NodeMCU) components are in the same enclosure as high voltage (120/240VAC) components and may be at risk of EMI
  • Thermocouple sensor lines 3.5 and 3.6 are run in the same conduit as L1/L2/N for the AT. Although the sensor cable is shielded, I am not sure if this will affect readings.
  • PZEM-004T current sensor has a 1s response time. It's suitable for slower moving trends, but not fast enough to catch over/under voltage conditions (the SVP-912 is 0.1s, in comparison). If we had a sensor with a faster response time, we could get rid of the 2qty SVP-912 devices entirely and control everything with the NodeMCU
EPIC!!!

I can wait to build a real world prototype of this design and test it out.
 
Ok, after waiting for weeks for parts and taking the time to test everything I thought I would post what I came up with to solve the problem.

This is what I came up with for myself and may or may NOT work for you. If don’t understand the issues involved the please consult a qualified person to help you. Electricity is dangerous and can injure or even kill if handled improperly.

After studying the SolarEdge wiring diagrams I felt it was best to try and duplicate the same functions they do when using the Auto Transformer. The solution I came up with uses what I’m calling an Inverter Combiner Box. One big advantage to this design is I can just use the Main Breaker in the Sub Panel and don’t need to duplicate breakers being used in the Inverter Combiner Box.

Pic #1 shows what I feel is the very least you must do be safe running the Growatt with the AT. (does not protect against loss of neutral for reasons other than a breaker trip)

Pic #2 shows what I have been running now over a week without any issues. (so far…) I does still have a potential problem, as mentioned in an earlier post, that it could cycle the Main Contactor if for some reason the sensor was seeing 120v only part of the time. (not sure what could cause that to happen though...)

Let me know what you think
Sorry I don't have same fancy design software being use by USUSUS

For those who are just reading this post and not the whole thread:
Growatt issues:
  • Output is 240v only.
  • Should be shut down if loss of neutral occurs.
  • Dry contact does not work like we need it too.
  • Has no direct way tell which source is being output.
Auto transformer:
  • Needs 25A breaker (per SolarEdge)
  • Needs breaker trip protection
  • Should be shut off when running on Grid power
Several key metrics should be met.
  • Neutral Bonding – Neutral should be bonded to the grounding conductor and in one place only and must be a permanent bond. We should not be switching it on and off using a contactor. (per NEC requirements)
  • We need to avoid placing the Auto Transformer in Parallel with the Grid.
  • Want to be able to have the Growatt switch to Grid/Utility as a backup to Solar/Battery system automatically.
  • Want the system to cut all power to the Load if:
    • The neutral is lost for any reason. (loss of 120v)
    • One the breakers running to the Auto Transformer(s) trips.
Given the above constraints and following the same method that SolarEdge uses I ran a Neutral line from the Main Panel to my Sub Panel. My Neutral Bond is in my main panel so that will insure all Neutrals are tied to Ground regardless of which source is powering the Loads.

I tested the Growatt and AutoTransformer running off PV and Battery in this configuration to insure it didn’t affect anything. It did not. Everything functioned exactly the same with or without the Neutral wire coming from the Main Panel and no current could be detected on that wire. This is what should happen as the L1 & L2 coming from Main Panel are isolated from the Sub panel in PV/battery mode.

I then tested the Dry-Contact on the Growatt which I determined does not work as we need it too. It activates when the battery is low and then switches back once it transfers to Grid Power. The solution I came up with is to use a Current Sensor on the lines coming from the Main Panel to the Growatts. (I also discovered you can use 230V sensor placed between one leg of the output from the Main Panel and the other leg of the output from the Growatts.)

This solution uses shunt trip breakers, Auxilliary Contact Breakers, two (or more) Contactors, and Voltage and Current Sensors.

Shunt trip Breakers on the output from the Growatts – which trip if the AT Breakers trip.

Auxiliary contacts on the Breaker for the Auto Transformers – Which will trip the Breakers coming from the Growatts in the event of an overload or short circuit of the AT.

A Normally Closed Contactor for the Auto transformer which insures it stays on unless the system is running in ‘On Grid’ mode.

Current Sensor on the output Wires from the Main Panel which will activate the N.C Contactor on the AT, cutting the flow, when power is flowing from the Main Panel.

I then put a large Normally Open Contactor off the output of the entire system tied to a Voltage Sensor which will ensure Power only flows when a proper 120v is being detected.
 

Attachments

  • Pic #1.png
    Pic #1.png
    316.6 KB · Views: 96
  • Pic #2.png
    Pic #2.png
    366.7 KB · Views: 101
I have to ask, how much does this cost? I haven't seen many shunt trip breakers that were cheap unless the breaker was lower amps.
 
I have to ask, how much does this cost? I haven't seen many shunt trip breakers that were cheap unless the breaker was lower amps.
The ones I purchased are EATON FAZ-C type - The Breakers were $46 and the Shunt trip adapters were $39 for the 50A and $33.50 for the 25A Breaker with an additional $21.50 for each Aux Contact. But since I don't need breakers for my Sub Panel I saved $42 and $27 for each of them. That means I paid $43 more for each of the 50A and $28 more for the 25A. so I paid $114 extra for the breakers. Found the terminal block on ebay for $16. The ABB N.C Contactor was $56 on Amazon and the Allen-Bradley C85 100A Contactor was $158 off ebay. and found the current sensor on ebay for $48 but they can be purchased new for $112. So total cost was about $400 for me but could cost more depending on where you find stuff.

Was well worth it for me.
 
Back
Top