Issue with EG4 18Kpv/Powerwall batteries dumping to 0%

[Dennis C]

Looking to see if anyone else has experienced this, or has any suggestions. EG4 Support has looked at it a few times now but it appears they're stumped as well.

The system is an 18Kpv with three Powerwall batteries, all purchased through Signature Solar. The system is grid-tied as a feed-in only (zero export), with a total of 20 panels in 3 banks of 10, 6, and 4. Normally the system operates just as I want - uses solar during the day to run things & charge the batteries, then runs off batteries overnight, only drawing on the utility once the batteries get down to 20%.

The problem is that about once a month or so, the app shows the battery SOC drops from wherever it is (25%, 90%, anything in between) to zero in an instant. The controller then kicks in & recharges the batteries from the line up to the minimum recommended 20%. No other warnings, no problems reported, everything seems to be communicating properly. The EG4 recognizes all 3 batteries and appears to accurately report voltages and SOC. Overall configuration is almost out-of-the-box standard other than setting it to zero export.

With three batteries, that's a LOT of stored energy just being dumped... somewhere? It feels like a fire risk at a minimum, but aside from that, it ultimately makes the entire system unreliable.

Thoughts, suggestions anyone? Has anyone else seen this with the same or a similar setup? Thank you for any assistance.

 

[Quattrohead]

Firmware versions of the inverter and batteries ?
Go through the data logs, it will show you something.
Or, setup Solar Assistant and see if it catches the condition.
The inverter behavior is correct though, it will recharge the batteries from grid when it see's they are very low.

 

[EG4TechSolutionsTeam]

What protocol and lithium brand are set? Also, would you be able to export the data history when this happened and your settings and post in thread?

With the Wallmount batteries, I would suggest having CAN EG4/LUX and lithium brand 1 set.

 

[rogerheflin]

Looking to see if anyone else has experienced this, or has any suggestions. EG4 Support has looked at it a few times now but it appears they're stumped as well.

The system is an 18Kpv with three Powerwall batteries, all purchased through Signature Solar. The system is grid-tied as a feed-in only (zero export), with a total of 20 panels in 3 banks of 10, 6, and 4. Normally the system operates just as I want - uses solar during the day to run things & charge the batteries, then runs off batteries overnight, only drawing on the utility once the batteries get down to 20%.

The problem is that about once a month or so, the app shows the battery SOC drops from wherever it is (25%, 90%, anything in between) to zero in an instant. The controller then kicks in & recharges the batteries from the line up to the minimum recommended 20%. No other warnings, no problems reported, everything seems to be communicating properly. The EG4 recognizes all 3 batteries and appears to accurately report voltages and SOC. Overall configuration is almost out-of-the-box standard other than setting it to zero export.

With three batteries, that's a LOT of stored energy just being dumped... somewhere? It feels like a fire risk at a minimum, but aside from that, it ultimately makes the entire system unreliable.

Thoughts, suggestions anyone? Has anyone else seen this with the same or a similar setup? Thank you for any assistance.

You likely need to set the discharge to use an equivalent battery voltage to 20%. My batteries (using JK BMS) just did exactly that for a 2nd time this morning. And looking at it it was showing a SOC of 31% but the battery voltage of 43V (<5% real SOC). I am setting voltage to be 51V so around 20% and assuming that SOC (if you have not fully charged in several days) is not really reliable on any BMS.

 

[Dennis C]

Firmware versions of the inverter and batteries ?
Go through the data logs, it will show you something.
Or, setup Solar Assistant and see if it catches the condition.
The inverter behavior is correct though, it will recharge the batteries from grid when it see's they are very low.

Firmware was up to date on all units as of a month or two ago, but I'll check again. Understood that the auto-recharge is correct behavior.

What protocol and lithium brand are set? Also, would you be able to export the data history when this happened and your settings and post in thread?

With the Wallmount batteries, I would suggest having CAN EG4/LUX and lithium brand 1 set.

Yeah, that was one of the early changes SS suggested. They're all LUX protocol now, as per their request, but we've tried both that and EG4. Behavior is the same either way. Let me see what data I can get off the 18Kpv when I'm home next - the App doesn't give too much visibility - but you guys already have remote access as well.

You likely need to set the discharge to use an equivalent battery voltage to 20%. My batteries (using JK BMS) just did exactly that for a 2nd time this morning. And looking at it it was showing a SOC of 31% but the battery voltage of 43V (<5% real SOC). I am setting voltage to be 51V so around 20% and assuming that SOC (if you have not fully charged in several days) is not really reliable on any BMS.

Interesting, thank you for the lead! Batteries 20% SOC and 51.6v which seems fine but at least that's something I haven't tried yet. I'll take some readings & talk to SS about trying to use voltage vs. SOC for the configuration.

Appreciate the help guys!

 

[marionw]

You mention an "App". Is this for the Inverter or for the batteries. I assume you are operating in "closed loop". The "Master" BMS reports the batteries as "One Big Battery" to the Inverter. The SOC reported to the inverter is the average of the SOC of each battery. Depending on what the Master BMS does with the battery data it is pulling from each battery (which is undocumented) it is possible for example that one of the batteries experiences a Cell Under Voltage Protection where that batteries BMS would then open the Discharge MOSFETS thus removing one battery as a source of power. I would expect that the SOC and total battery bank capacity being reported to the Inverter would change significantly. In addition the Master BMS reports battery alarm and fault conditions and again, what the Inverter does with this information is also undocumented.

If you do not periodically charge the batteries to the point where the BMS starts to balance cells and resets the SOC to a more accurate 100% then the SOC being reported to the inverter is most likely inaccurate.

You can download software (including the latest firmware) from the EG4 website. This software will allow you to monitor each battery including individual cell voltages during the charging process. You need to ensure that cells are in fact getting charge to whatever the balance voltage setpoint is (the software should show you this) and that the BMS is given the opportunity to perform cell balancing. You could for example have one or more "runner" cells in a battery, that is a cell that charges much faster than the others and reaches Cell Over Voltage Protection at which time the BMS will turn off the Charging MOSFETS stopping further charging. This situation can become worse over time if you do not periodically manage the battery charge to ensure cell balancing happens.

Just need to verify the batteries are in fact getting fully charged to eliminate under charged batteries as the fault.

 

[Dennis C]

You mention an "App". Is this for the Inverter or for the batteries. I assume you are operating in "closed loop". The "Master" BMS reports the batteries as "One Big Battery" to the Inverter. The SOC reported to the inverter is the average of the SOC of each battery. Depending on what the Master BMS does with the battery data it is pulling from each battery (which is undocumented) it is possible for example that one of the batteries experiences a Cell Under Voltage Protection where that batteries BMS would then open the Discharge MOSFETS thus removing one battery as a source of power. I would expect that the SOC and total battery bank capacity being reported to the Inverter would change significantly. In addition the Master BMS reports battery alarm and fault conditions and again, what the Inverter does with this information is also undocumented.

If you do not periodically charge the batteries to the point where the BMS starts to balance cells and resets the SOC to a more accurate 100% then the SOC being reported to the inverter is most likely inaccurate.

You can download software (including the latest firmware) from the EG4 website. This software will allow you to monitor each battery including individual cell voltages during the charging process. You need to ensure that cells are in fact getting charge to whatever the balance voltage setpoint is (the software should show you this) and that the BMS is given the opportunity to perform cell balancing. You could for example have one or more "runner" cells in a battery, that is a cell that charges much faster than the others and reaches Cell Over Voltage Protection at which time the BMS will turn off the Charging MOSFETS stopping further charging. This situation can become worse over time if you do not periodically manage the battery charge to ensure cell balancing happens.

Just need to verify the batteries are in fact getting fully charged to eliminate under charged batteries as the fault.

Just referring to the "EG4 Monitor" app. The Inverter does recognize 3 batteries (one primary, multiple secondary) but does combine the displayed value to one SOC. I'll get the software downloaded this weekend & see what I can figure out. Between this & the info above about SOC being less accurate in general than voltage, I'm guessing that's the right path.

The problem manifested back after I first installed, 1 battery & summer sun, so it was cycling 100% - 20% daily. Now in winter with 2 more batteries, so it's 70-85% at best, or just holding when it's cloudy. If there's a bad cell, it's in that 1st battery.

Again, thanks for the help.

 

[rogerheflin]

Just referring to the "EG4 Monitor" app. The Inverter does recognize 3 batteries (one primary, multiple secondary) but does combine the displayed value to one SOC. I'll get the software downloaded this weekend & see what I can figure out. Between this & the info above about SOC being less accurate in general than voltage, I'm guessing that's the right path.

The problem manifested back after I first installed, 1 battery & summer sun, so it was cycling 100% - 20% daily. Now in winter with 2 more batteries, so it's 70-85% at best, or just holding when it's cloudy. If there's a bad cell, it's in that 1st battery.

Again, thanks for the help.

voltage (for battery remaining) only works ok-ish from 90%-100% and 0-20%. In the middle say 20% to 90% the voltage changes very little with with significant increase/decreases of charge making it difficult to tell where in the 20-90% range you really are.

 

[Dennis C]

Got it. Now that I'm paying attention to the battery voltage, it's dropped from 51.6v to 47.5v in a week of sitting at about 20% SOC. We're not getting any sun right now, so it's neither charging nor discharging, but it is connected.

That feels like an excessive drop over a 7-day period for a LiPo just based on what (little) I can find online. Event records on the inverter show a "low battery voltage" at the same time as the regular, sudden drops to 0% SOC, which I thought corresponded to an auto-discharge of some type, but now I'm thinking the voltage just drops to the point that the inverter triggers a reset of the SOC.

Maybe one of the 3 batteries - the original one, I suppose, since this has been happening since day 1 - is bad?

 

[400bird]

The inverter consumes energy just being on, before powering any loads. The voltage difference you are seeing, isn't likely internal discharge, but energy used to power the inverter.

 

[rogerheflin]

Got it. Now that I'm paying attention to the battery voltage, it's dropped from 51.6v to 47.5v in a week of sitting at about 20% SOC. We're not getting any sun right now, so it's neither charging nor discharging, but it is connected.

That feels like an excessive drop over a 7-day period for a LiPo just based on what (little) I can find online. Event records on the inverter show a "low battery voltage" at the same time as the regular, sudden drops to 0% SOC, which I thought corresponded to an auto-discharge of some type, but now I'm thinking the voltage just drops to the point that the inverter triggers a reset of the SOC.

Maybe one of the 3 batteries - the original one, I suppose, since this has been happening since day 1 - is bad?

The battery will eventually hit a low enough voltage that the BMS figures out the SOC is wrong (ie 20% but voltage says 1% really) and it will reset to what the voltage says. My JK BMS has done it a couple of times, and multiple other brands of BMSes exhibit this same behavior. The inverter will use some amount of wattage just sitting there. You may want to use a DC current meter on the battery when the inverter has zero load and see how much it is consuming. Note that 10w load comes to 1.68kwh in 7 days, and the inverter is likely using more than just 10w.

 

[Dennis C]

Voltage still creeping down, now 44.8 (still "21%") and the inverter now has a Low Battery Voltage warning, so it sounds like your explanation is playing out. I really appreciate the understanding.

Much more helpful than EG4/Signature Solar's support. I don't know if that's common to the industry but at this point I'd recommend trying a different brand if you're starting a new system.

 

[rogerheflin]

Voltage still creeping down, now 44.8 (still "21%") and the inverter now has a Low Battery Voltage warning, so it sounds like your explanation is playing out. I really appreciate the understanding.

Much more helpful than EG4/Signature Solar's support. I don't know if that's common to the industry but at this point I'd recommend trying a different brand if you're starting a new system.

Everyone's tech support for everything (not just all solar) is sub-standard. They can handle you did not plug in the power cord, you did not turn it on and other similar obvious stuff and/or common stuff, but anything more complicated and they don't understand how the hardware works (other that it is magic). If someone at any company becomes good enough to understand that much they get promoted to someplace else (engineering/development and such) and you (or I) will rarely ever be able to talk to that person as they are too valuable to be let out of the dark room they are in.

I had a Linux filesystem/backup issue that we had bounced around with our Linux support provider for months. They asked for all sorts of data and test and could not find anything to explain it (man-weeks of collecting/discussing). A VP at the company I worked for called in some favors with the support provider and got me 5 minutes with one of those real experts. The entire conservation with the real expert did not take the entire 5 minutes for him to make a guess and give me a test to confirm or deny that guess.

And the sad reality is just about everyone's tech support is simply bad, I typically have ugly issues(typically because I solve/understand the easy issue before I call) and I typically have low expectations on the ugly issues being solved and I find often enough that my expectations were not low enough...

 

[GamERT]

I have had the same issue, still performing my analysis..

1. I have an 18kpv
2. 2 EG4 Wall mount batteries
3. my On-Grid Cut-Off SOC(%) set to 35%

I have been running the same setup, with no issues since last month.. a sudden dump, loss or weird power wasn't there, to begin with ?

Interesting..

Maybe just move to voltage versus %.. Does anyone have a good translation table on SOC % to Voltage?

 

[rogerheflin]

What you see happening is typically what happens when the BMS thinks SOC is higher than it really is.

The SOC jumps to 0% when the BMS sees some too low of voltage. Generally it looks like just about all of the BMSes exhibit this behavior and/or design defect. Everyone's best guess is that the BMS does not measure lower currents right (say like under 10a). I know my meter said .5A and my BMS was still reporting 0, so there is definitely an inaccuracy around 0. The longer you run the bms without charging or discharging it fully the worse the SOC seems to get off.

Here is a chart someone had before: (one has to love a "code" formatting that defaults to a non-fixed font...)

Percentage (SOC)1 Cell    12V        24V        48V
100% Charging         3.65    14.6    29.2    58.4
100% Rest             3.4     13.6    27.2    54.4
90.00%                3.35    13.4    26.8    53.6
80.00%                3.32    13.3    26.6    53.1
70.00%                3.3     13.2    26.4    52.8
60.00%                3.27    13.1    26.1    52.3
50.00%                3.26    13      26.1    52.2
40.00%                3.25    13      26      52
30.00%                3.22    12.9    25.8    51.5
20.00%                3.2     12.8    25.6    51.2
10.00%                3       12      24      48
0.00%                 2.5     10      20      40

 

[solarkid4486]

This has happened to me twice now as well. 18kPV + 2x EG4 Indoor Wallmount 280AH batteries. I am on a grid-tied (zero-export) with a back-feed breaker configuration.

The first time on 2024-10-25. It dropped from 55% to 0% instantly.

I called Signature Solar the next day, and they reviewed some data but ultimately provided no solution/conclusion except for guidance to keep an eye on it and measure the battery voltage across the leads if it happens again.

It just happened a second time today, dropping from 35% to 0% instantly.

The EG4 app/monitor website showed the batteries at 44v. After power-cycling the batteries/BMS they came on and were showing 48.86v and 48.7v. I only measured the voltage across one battery with a multimeter, and it read 48.7v. As you can see on the graph, it started to charge back up all the way from 0%. I had original concerns as the OP -- did 55%/35% (first occurrence/second occurrence) of the power in the batteries actually discharge that quickly or did was the BMS SOC% that far off? In both of my cases, the batteries were up to 100% several hours before which is supposedly when the BMS gets the most accurate SOC/Voltage calibration for these LFP batteries?

Here's another chart that includes the battery voltage.


Unfortunately, I don't have any feedback to improve the situation; mostly just a "me too" post...

 

[400bird]

The EG4 app/monitor website showed the batteries at 44v.

I'll agree that should prob be about 0% SOC

After power-cycling the batteries/BMS they came on and were showing 48.86v and 48.7v. I only measured the voltage across one battery with a multimeter, and it read 48.7v. As you can see on the graph, it started to charge back up all the way from 0%. I had original concerns as the OP -- did 55%/35% (first occurrence/second occurrence) of the power in the batteries actually discharge that quickly or did was the BMS SOC% that far off? In both of my cases, the batteries were up to 100% several hours before which is supposedly when the BMS gets the most accurate SOC/Voltage calibration for these LFP batteries?

Here's another chart that includes the battery voltage.
View attachment 269507

Looks like battery voltage was about 52.6 on the left of the graph. You said the SOC was at 100%, but unless I'm reading the graph wrong, it's in the 50's

 

[blueforce]

Similar issue here as well. I have 3x powerpro batteries with an 18kpv and get free night time charging so during the day I have my batteries set to discharge down to my On-Grid cutoff at 10% SoC. When everything was set up it was using SoC to charge and discharge so when I charged at night the SoC would go to 100% but the voltage never got above 54v.

First day of running the batteries in June it dropped from 11% to 0% and I pondered the expensive mistake I just made LOL. Then for the rest of summer it would drain down to 10% and switch to grid without issues and the voltage would always end around 50v. Looking back at the data though I can see I wasn't getting the full 90%-~5% loss of battery energy but didn't have a cutoff after that initial one. Fast forward through fall and never drained to 10% because of less power use and better weather and only charged based on SoC. Then one day we have a cold morning in December and bang! SoC went from 18-0. Same story though, didn't have a cutoff after that day. My guess is the BMS SoC got out of whack in the fall.

Ended up working with SigSolar on a firmware issue where we had the batteries in lead-acid for a bit and afterwards I maintained using voltage for charging and SoC for discharge. My settings are shown below for "AC Charge Mode" and since changing I can see that I'm getting the full 90% of battery power (100%-10%) minus 4-5% for loss (36.6kwh from 42.9kwh of powerpros) and the battery voltage around 10% gets down around 50v.


Disclaimer 1: I'm using these settings based on the battery spec sheet.
Disclaimer 2: With free night time charging I charge my batteries to 100% every night.

Also, it would be real nice to have a conditional safety where say the batteries are set to stop discharging at 10% but if the BMS goes wonky again it would also rely on voltage (46, 47, 48, whatever you're comfortable with) to cut off discharging.

Here's December 7th when it finally got cold:

 

[marionw]

Something I always recommend when folks feel their batteries are not performing up to snuff. Download the manufacturer provided software which will allow you to get a very detailed view of the battery. With the software you can monitor the battery during charging to ensure there is no fault that is preventing a full charge. Some BMS will reset SOC to 100% when a Cell Over Voltage Protection fault happens even though the battery is nowhere near being fully charged. If during the charge the BMS is not allowed to active it's balancing circuitry and most of the cells do not get charged to 3.5 volts then you are not getting a full charge. The software is free so give it a try.

Download the BMS_Tools software form the EG4 website:

EG4® WallMount All Weather - EG4 Electronics

Discover the reliable and durable EG4 WallMount All Weather Battery, featuring IP65 outdoor rating, 10kW power delivery, and 8000 cycle life.

eg4electronics.com

 

[solarkid4486]

I'll agree that should prob be about 0% SOC

Here's the tabular form of the data. I don't know why it read 45.5%


When I eventually power-cycled the battery breakers + BMS, they were showing as follows:





So, I don't know why they ever reported ~45V?

Looks like battery voltage was about 52.6 on the left of the graph. You said the SOC was at 100%, but unless I'm reading the graph wrong, it's in the 50's

Unfortunately, on that graph, the percentage axis isn't visible, so you'll have to corroborate it with the graph immediately above it. The left axis is Voltage and only ranges from 45-55.5...
Here is another screen capture hovering over the lines showing the SOC% at 98% on the left edge:

 

[400bird]

So, the SOC becomes inaccurate over time and needs to be recalibrated. This should happen when the cells hit a certain voltage. Do you have a way to graph cell voltage? Specifically when the SOC last hit 100%?

Or the pack could be out of balance. But this is unlikely because your cells are pretty even at 3 volts.

On the BMS that I frequently work with, the SOC recalibrates when the cell voltage hits 3.5-3.65
I don't know EG4 's relabeled BMS's spec, but likely something similar.

So, likely you haven't really hit 100% with cell voltages high enough to recalibrate the SOC in a while.

 

[jeff.lankford]

Here's the tabular form of the data. I don't know why it read 45.5%

From the data, it looks like the resting battery voltage after the shutdown was 47.0V, which is less than 10%. Before the SOC drop, the battery was discharging at about 130 amps, which was pulling the voltage down. It appears that 45.5V is the magic number at which the BMS will readjust its SOC to 0% and that triggered a shutdown. These data points are 4 minutes apart, and the voltage was dropping rapidly, so the 0% trigger point may be lower than 45.5V. Perhaps it's 45.0V or even less. After the shutdown, the resting voltage of the battery recovered to 47.0V.