SolarEdge cold start

[jc95]

I have a SE 7600H with SE Backup Interface & SE battery. We had a recent grid outage. When the outage occurred, the BI switched to inverter and PV sustained the load, then battery after sunset. The battery lasted until the middle of the night when it was depleted and the load lost power. All as expected. At sunrise, however, when the PV could have been producing, the system remained cold & dark. No charging of the battery or powering the load. Nothing until the grid came back up in the afternoon, when the system started working normally again, powering the load, charging the battery, and exporting to the grid.

I expected the system to cold start and at least charge the battery and eventually produce AC to power the load. But it didn't. I have three possible explanations:

1. it is working as intended. No cold-start capability. Grid must be present to start.

2. Installation or configuration error.

3. Faulty firmware or component failure.

SE has an application note regarding generators. I don't have a generator, but in the app note, the generator is behind its own transfer switch and not integrated with the BI or inverter. And the app note is careful to point out that the generator is never connected to the BI. It says when the PV starts producing it will charge the battery while the gen powers the load. Then, the generator is switched off and the BI is connected to the load. No grid power thru all this. That seems to say that #1 is not correct, but it introduces a new possibility:

4. The app note is wrong.

Does anyone have any idea what I'm seeing and what the cause might be?

 

[pv_n00b]

This is due to the BESS being completely discharged. Without the BESS providing an AC sync signal to the PV inverters they will not turn on in the morning. When the grid comes back the AC sync is restored, the PV inverters can turn on, and the BESS can recharge.
The way to prevent this is to not completely discharge the BESS overnight so that in the morning if the grid is still down the BESS can provide the AC sync signal to the building and the PV inverters can start, assuming an AC coupled BESS. Another way is to have a generator that can provide the AC sync signal and black start the system.

 

[wwhitney]

This is due to the BESS being completely discharged.

An intelligently designed BESS would shut down at, say, 1% SOC, rather than going all the way to 0% SOC, precisely to cover the usage case in the OP, so that it could black start when PV is back. Are you saying that SolarEdge doesn't do this?

Cheers, Wayne

 

[jaggedben]

This is due to the BESS being completely discharged. Without the BESS providing an AC sync signal to the PV inverters they will not turn on in the morning. ...

No, the latter part is wrong regarding Solaredge, because it is DC coupled. The inverter should be able to source power from the BESS and charge from the solar without outputting AC power. But, the part about the batteries being completely discharged is correct, as with Solaredge I'm pretty sure the PV can't output without the inverter being awake.

 

[jaggedben]

An intelligently designed BESS would shut down at, say, 1% SOC, rather than going all the way to 0% SOC, precisely to cover the usage case in the OP, so that it could black start when PV is back. Are you saying that SolarEdge doesn't do this?

Cheers, Wayne

It sounds from the OP like Solaredge either doesn't do this or in his case the system wasn't properly configured. Solaredge ought to do this more smoothly since it ought to be able to do direct control of battery charging and recharge before outputting power to loads. But it wouldn't surprise me if SE has once again dropped the ball on good programming.

To your comment, an intelligently designed system will initially shut down at much higher SOC than 1%. I'd recommend from 10-25% 'usable' SOC, or about 15-30% real SOC. For AC coupled systems other than Enphase IQ8 I'd recommend on the higher end of that. A system should stop trying to wake up again around 1-2% usable SOC or higher, i.e still above 5% or so real SOC.

Also make sure homeowners are educated on how to reduce load to ease such 'wakeups', how to tell whether their power coming back on is from the grid or their system waking up, and not to assume it's the grid and use too much power so that the system can't recharge. Don't ask me how I know.

 

[pv_n00b]

No, the latter part is wrong regarding Solaredge, because it is DC coupled. The inverter should be able to source power from the BESS and charge from the solar without outputting AC power. But, the part about the batteries being completely discharged is correct, as with Solaredge I'm pretty sure the PV can't output without the inverter being awake.

I can't find much on the SE website on the operation of the system in backup mode. My guess is that it requires some minimum battery SOC to black start and can't black start from the PV array only. Maybe even though it was recharging from the PV array in the morning that SOC was not reached before the grid came back up. Lots of options, this stuff is all magic.

 

[Steve16]

It does require some power from the battery to "wake" the inverter. I've always kept the SoC around 5% to hopefully avoid nuisance issues like this. You would think the manufacturer would have something built in to avoid this

 

[jc95]

Resurrecting this after some further investigation & testing...

According to my installation contractor & SolarEdge, my system should support dark start. My question was how? If I understand how it is supposed to work, the battery discharges down from the programmed reserve (9% in my case) to some non-configurable safety reserve (5%?). It never goes to 0 to protect the battery. It then shuts off and the load & inverter go dark. When the sun rises and the panels start to produce, the optimizers need a signal from the inverter to wake up and start sending power above their safety shutoff 1v. There is an ordinary 9v battery in the inverter that powers just enough electronics to send the wake up signal to the optimizers and they'll start sending whatever they can based on the limited sunlight available. Not enough to power much of a load at first, but as the sunlight increases, more and more power is available. What the system does with this is less clear.

I did a few tests.

1. Turned the grid off in the evening and ran on battery until it depleted ~ 2AM when everything went dark, as expected. A bit after the sun came up, the system stayed dark but after some minutes, the load turned on for < 1 sec and then off. Stayed off until I turned the grid back on ~1 hr later. Then everything came back up.

At this point, the contractor swapped the inverter. He said SE saw some error codes, but didn't know what. But new inverter...

2. Repeated the test. Same result. Lights blinked on, then off.

3. Next test. The same, except after depletion but before sunrise, I turned off the load (house) so the inverter, battery, backup interface, PV panels were all isolated. No load except whatever the internal inverter and backup interface had. This time, the inverter appeared to power up normally. Inverter status LEDs were normal. About an hour after sunrise, I turned on a small load (<100W). Worked OK. Another 100W. Still OK. No more easy loads so I turned the rest of the house on (~1000W, I think). This appeared to work OK, but only for ~5 minutes. Then everything tripped off again and the inverter went dark. Stayed off until I turned the grid back on, then everything normal.

My suspicion is that when the PV started, it may have been trying to charge the battery with DC from the PV, or maybe not and it was just powering the inverter (with no/small load) from PV and the battery remained discharged. When the load was fully connected, maybe a motor started somewhere (refrigerator?) and this spiked the current and pulled the voltage down and the inverter tripped off with an under-volt condition. Not sure where to go with this. SE tech support has been useless and just responds to the case with an unrelated answer and then closes the case.

Another data point: my neighbor has a similar system. Same inverter (though he has an AC-coupled battery). His inverter is the same model number (SE7600H-US) but a slightly different part number and the description is the same as mine but with the added "with stand alone mode". Not sure what this means, but it sounds suspiciously related. He is not in a position to replicate my tests, however, and hasn't been in a situation where a dark start would be necessary.

So, any suggestions? Thanks for reading thru my story.

 

[wwhitney]

Missing from your recent test result report are the behavior of the battery level over time, as well as how your attempted house demand compared to the PV generation, instantaneously. Do you have access to current reporting of battery level, PV power generation, and house power demand? If not, it will be hard for you to suss out what is happening.

If dark start is working correctly, and your house loads are off, then you should see the PV generating the expected amount, and the batteries charging accordingly. If you then try to turn on your house loads, as long as the house demand stays always below the PV output, the house loads should run while the batteries continue to (slowly) charge from the surplus.

In other words, at an hour after sunrise, with the batteries depleted, if the PV is making less than 1 kW, there's no way for the system to support 1 kW of house load.

Cheers, Wayne

 

[jc95]

I can't see battery level without turning on the network which is about 300W of load. After the grid is restored and everything back up, I can see that the battery was < the programmed reserve (9%) when the grid was restored.

I understand that the PV (or PV + battery) can't supply more than it's producing.

I think the key might be that the inverter can never know what the load is until it tries. It's a 7.6kW inverter. My PV array, while rated for 10, never produces more than 7kW (orientation and multiple planes), and even that is the peak on a good day. The battery can supplement the PV if there's charge, but it's possible for the load to exceed that (oven + A/C + ...). Point is that if the inverter tries and the load is > whatever power the inverter can source, I think the inverter will do a hard trip, throw a fault code, shutdown and never retry until the grid comes back.

If the load is tiny when the sun comes up, maybe it can power the tiny load and divert whatever is left to the battery. Stay in that config until the sun is much higher, eventually producing enough to power more load and I can slowly, manually, turn on more load. Just keep the load always < the available PV. I think it would not be good for the inverter to try to power the load, trip, try again later, trip again, try again, .... Once it hard trips, there doesn't seem to be any way out of that state until grid returns.

 

[solarken]

Resurrecting this after some further investigation & testing...

According to my installation contractor & SolarEdge, my system should support dark start. My question was how? If I understand how it is supposed to work, the battery discharges down from the programmed reserve (9% in my case) to some non-configurable safety reserve (5%?). It never goes to 0 to protect the battery. It then shuts off and the load & inverter go dark. When the sun rises and the panels start to produce, the optimizers need a signal from the inverter to wake up and start sending power above their safety shutoff 1v. There is an ordinary 9v battery in the inverter that powers just enough electronics to send the wake up signal to the optimizers and they'll start sending whatever they can based on the limited sunlight available. Not enough to power much of a load at first, but as the sunlight increases, more and more power is available. What the system does with this is less clear.

I did a few tests.

1. Turned the grid off in the evening and ran on battery until it depleted ~ 2AM when everything went dark, as expected. A bit after the sun came up, the system stayed dark but after some minutes, the load turned on for < 1 sec and then off. Stayed off until I turned the grid back on ~1 hr later. Then everything came back up.

At this point, the contractor swapped the inverter. He said SE saw some error codes, but didn't know what. But new inverter...

2. Repeated the test. Same result. Lights blinked on, then off.

3. Next test. The same, except after depletion but before sunrise, I turned off the load (house) so the inverter, battery, backup interface, PV panels were all isolated. No load except whatever the internal inverter and backup interface had. This time, the inverter appeared to power up normally. Inverter status LEDs were normal. About an hour after sunrise, I turned on a small load (<100W). Worked OK. Another 100W. Still OK. No more easy loads so I turned the rest of the house on (~1000W, I think). This appeared to work OK, but only for ~5 minutes. Then everything tripped off again and the inverter went dark. Stayed off until I turned the grid back on, then everything normal.

My suspicion is that when the PV started, it may have been trying to charge the battery with DC from the PV, or maybe not and it was just powering the inverter (with no/small load) from PV and the battery remained discharged. When the load was fully connected, maybe a motor started somewhere (refrigerator?) and this spiked the current and pulled the voltage down and the inverter tripped off with an under-volt condition. Not sure where to go with this. SE tech support has been useless and just responds to the case with an unrelated answer and then closes the case.

Another data point: my neighbor has a similar system. Same inverter (though he has an AC-coupled battery). His inverter is the same model number (SE7600H-US) but a slightly different part number and the description is the same as mine but with the added "with stand alone mode". Not sure what this means, but it sounds suspiciously related. He is not in a position to replicate my tests, however, and hasn't been in a situation where a dark start would be necessary.

So, any suggestions? Thanks for reading thru my story.

What exact model is this inverter? 7600H could mean the old setapp inverter model or it could mean the Energy Hub. SE7600H-USS3BBC14, SE7600H-USSNBBL14, are Energy Hubs. SE7600H-US000BNI4 is an old setapp inverter that does not support battery storage without an extra add-on kit. Just trying to understand more anout the basics. Also, what model ESS is connected to the inverter? When you connect to it with the SetApp configuration tool, does it show the status of the ESS and the inverter, the optimizers, backup interface, all good? Any communication errors, like RS-485 issue with the backup interface?

 

[jc95]

SE7600H-USS3BBC14. (BTW, Neighbor's is SE7600H-USMNBBL14. any idea what that is?)
ESS is SE Energy Bank (9.7kWh).
Status is all good. No errors.

 

[solarken]

SE7600H-USS3BBC14. (BTW, Neighbor's is SE7600H-USMNBBL14. any idea what that is?)
ESS is SE Energy Bank (9.7kWh).
Status is all good. No errors.

The difference is in the AC capacity during grid forming (grid down, battery operation) scenario. The USS3BBC14 can access I believe 6kW max from the battery, while the USMNBBL14 can access 10kW. Since the energy bank can only output 5kW, it would be the limiting factor.

It is aggravating, but the best way to get support for solaredge would be to call support while on the way to the site, choose the Energy Hub support team in the menu, and ask for a callback when the recording tells you there is a long wait, and get there before the call back. hopefully you get a good tech and you can check things live while they are on the phone as they look from their end.

As far as testing how depleted the battery can get without cold start / initial gridforming issues, maybe the minimum battery level can be temporarily set to a high value, like 5% below what it is currently at in order to simulate a low battery condition to cold start from.

 

[jaggedben]

...

3. Next test. The same, except after depletion but before sunrise, I turned off the load (house) so the inverter, battery, backup interface, PV panels were all isolated. No load except whatever the internal inverter and backup interface had. This time, the inverter appeared to power up normally. Inverter status LEDs were normal. About an hour after sunrise, I turned on a small load (<100W). Worked OK. Another 100W. Still OK. No more easy loads so I turned the rest of the house on (~1000W, I think). This appeared to work OK, but only for ~5 minutes. Then everything tripped off again and the inverter went dark. Stayed off until I turned the grid back on, then everything normal.

...

My best guess is that you simply used too much energy, too quickly, and simply dropped the SoC back down to its cutoff point. The hour after sunrise is not a high production time. It's reasonable to estimate you get a couple hundred watt hours out of the array at most, not much more unless the array is east facing and unobstructed and the sky is clear. Some of the production will be lost to running the system electronics and some to roundtrip efficiency. Then you use the remaining 100W that actually charged in 5 minutes, and get back down to cutoff SoC. This scenario isn't at all unrealistic, especially if your estimate of the amount of load or timing is off by a bit.

You did the first part right, which is to turn all the load off. I tell my customers that if they are in this situation to only leave on the internet router until they can see that the production and battery charge is high enough. Which will probably be, like, 10am at the earliest on a good day. (Maybe earlier in June.) I suggest you reconfigure your 300W network so that you can run just one device that allows the Solaredge to connect while you're checking it. Or maybe Solaredge has another way to communicate directly.

The problem is not power, it's energy.

 

[pv_n00b]

Great work jc95 on the testing and I agree with Jaggedben. If the battery is at or below the cutoff SOC but above the black start SOC and the sun comes up the system will start charging the battery, but it will not have enough excess energy to supply much of a load. I would not be surprised if it had to run most of the day just charging the battery before it was ready to support its rated backup loads without quickly shutting down.
This is one of the joys of being your own utility. It's a lot more hands-on than the typical grid-tied solar customer is interested in getting involved in. Now if you have a completely off-grid system most of those customers totally geek out over this stuff.

 

[jc95]

The difference is in the AC capacity during grid forming (grid down, battery operation) scenario. The USS3BBC14 can access I believe 6kW max from the battery, while the USMNBBL14 can access 10kW. Since the energy bank can only output 5kW, it would be the limiting factor.

It is aggravating, but the best way to get support for solaredge would be to call support while on the way to the site, choose the Energy Hub support team in the menu, and ask for a callback when the recording tells you there is a long wait, and get there before the call back. hopefully you get a good tech and you can check things live while they are on the phone as they look from their end.

As far as testing how depleted the battery can get without cold start / initial gridforming issues, maybe the minimum battery level can be temporarily set to a high value, like 5% below what it is currently at in order to simulate a low battery condition to cold start from.

I read the inverter p/n incorrectly...
The original inverter was SE7600H-USS3BBC14. (installed August 2022)
This was replaced by SE7600H-USSNBBL14. (installed April 2023)
Neighbor's new inverter is SE7600H-USMNBBL14. (installed April 2023)
I don't know how to decode the SE p/n's.

 

[jc95]

Great work jc95 on the testing and I agree with Jaggedben. If the battery is at or below the cutoff SOC but above the black start SOC and the sun comes up the system will start charging the battery, but it will not have enough excess energy to supply much of a load. I would not be surprised if it had to run most of the day just charging the battery before it was ready to support its rated backup loads without quickly shutting down.
This is one of the joys of being your own utility. It's a lot more hands-on than the typical grid-tied solar customer is interested in getting involved in. Now if you have a completely off-grid system most of those customers totally geek out over this stuff.

On a "normal" California sunny day, my baseline load is ~1400W. With grid on, PV in the morning first goes to offset some of the grid until PV > load. In mid-May, this was ~8:00AM. After this, PV powered the load and excess went to battery. Battery is fully charged ~11:00AM. After this, excess is sold back to PG&E. (Battery profile is MSC.) This continues until the evening when battery starts supplementing the diminishing PV and eventually carries the full load until depletion ~ 1AM. Grid powers the overnight load until sunrise. Times vary with the load, season and weather, of course, but this is the typical pattern.

Grid really only comes into play overnight into the early AM By 8AM on a clear day, I'm usually all PV/battery until after midnight. OR if there's a grid outage that isn't restored before the battery depletes. THAT'S the case I'm struggling with. Fortunately, that's a rare situation. Unfortunately (or fortunately?) that situation occurred 1 day after the battery was commissioned. Otherwise I might never have seen it.

I think I'm convincing myself that it's working as designed/configured and I need to aggressively shed load in the event of a grid outage so the battery will last overnight. That would have 2 effects: delay or eliminate battery depletion, and allow black start to occur without tripping the inverter. Still, it would be nice to have a way to restart a tripped inverter without the grid.

There are 2 configurable settings I think might affect this: the backup SOC limit (currently 9%, the minimum allowed) and the battery profile (currently maximize self-consumption). Thoughts? The principal purpose of adding the battery was to shift imported energy away from, and exported energy into, the peak rate period. Maybe switching to a different battery profile after 12AM? Backup was a secondary consideration.

 

[Joethemechanic]

I don't know how often you have power failures, but probably your cheapest and most reliable option is a small propane powered backup generator. You really don't need that big of one. During hurricane Sandy I powered my whole house with a little 4 cylinder 1800RPM 10 kilowatt Kohler I got for free. I even let my neighbors on each side plug cords into my outdoor receptacles so they could have light and keep their fridges running. We went for over a week like that and only went through a little over 200 pounds of propane.

 

[wwhitney]

By 8AM on a clear day, I'm usually all PV/battery until after midnight

So given that, repeat Test 3 on a clear day this month, but wait until 9 AM to turn on any house loads.

Although first, have you done the test of flipping the service disconnect at noon on a clear day (when battery is fully charged), to be sure you have enough inverter power to run all your house loads (say for several hours)? Maybe there is an intermittent high power house load you haven't accounted for.

Cheers, Wayne

 

[mddorogi]

Can't help much, but during February ice storms here in Michigan we had around 18 customers on battery backup for up to 7 days. I think there were three instances of batteries being depleted overnight. They all started up automatically the next day. I don't recall any settings related to this. The only detail that I clearly recall, when one of the SolarEdge engineers visited us, was that the inverter will not start back up without that 9V battery that comes with the energy hub inverter. I never realized what the purpose of that battery was until he said that. I think the 9V battery allows the inverter to continue checking for DC voltage from the solar panels and to start back up when there's enough.