SOC mismatch for two parallel batteries model SWA LiFePO4 51.2V-100Ah.

[Eng.Mohamed]

I did installed two battery lithium LiFePO4 in an off-grid system, After individually charging both units to 100% SOC and connecting them in parallel, we monitored their discharge over two consecutive weeks. A consistent SOC mismatch was observed, reaching up more to 13% difference between Battery 1 and Battery 2.
Values recorded during discharge for Battery 1 Battery 2:
Pack Voltage: 52.96V, 52.93V
Current Im: -3.82A , -3.3A
SOC : 72.60% , 59.47%
The inverter screen showed an average SOC of 66% for two batteries.
Steps taken:

• Charged each battery separately to 100% SOC.
• Allowed 24-hour rest period for voltage stabilization.
• Both batteries connected in parallel and via RS485 BMS communication.
• Configured inverter parameters precisely to lithium battery mode LI.
• Monitored charge/discharge cycles for 2+ week.

Key technical observation:
There are slight differences in voltages between the two batteries, resulting in unequal currents flowing as recorded for swarm currents during the

Although the voltage is water balanced, battery 2 discharges much faster. This indicates an effective capacity mismatch, possibly a software issue in the calibration of the BMS SOC algorithm, or a manufacturing defect in the BMS that leads to cell imbalance, not achieving a balanced system between the two batteries

• BMS VER : P16S100A-41325-1.00 ,Serial number is : 413251044800102P,
• Battery Model: SWA-WM-16S100
• Rating: 51.2V - 100Ah
• Quantity: 2 units
• Configuration: Parallel
• Growatt PV OFF-GRID Inverter ,Model :SPF 6000ES PLUS
• Communication: inverter and batteries BMS detected successfully via RS485 connected, Batteries parallel Communication via RS485B to RS485A.

I would like engineers and interested parties to help us diagnose this issue and provide us with appropriate solutions in addition to the following solutions, if appropriate

• Provide the BMS calibration file or firmware update to resolve the SOC calculation inconsistency.
• Advise us on precise diagnostic steps to evaluate BMS, cell health and internal resistance.
• Clarify the warranty and replacement process based on serial numbers and diagnostic logs.
• Provide me with new firmware update for Growatt PV OFF GRID Inverter Model: SPF 6000ES PLUS.

Photos, screenshots, and videos for screen batteries and PbmsTools HS 1.0.6 report files been included .

Thank you very much for your support and cooperation.​

 

[WrenchLight]

This is common. I wouldn't worry about it.

 

[Eng.Mohamed]

This is common. I wouldn't worry about it.

I would like to say that the Batteries Analysis was checked using the PbmsTools program, and it was read:
* BMS error and alarms log (memory info file).
* Mismatch difference between the two batteries SOC
* Voltage difference between cells.
* Temperature.
* Unbalancing Status.
All these steps and investigations prove that there is a mismatch in the SOC between the two batteries, and the results showed that the second battery discharges faster than the first and indicates a lack of capacity
Attached is a screenshot of the memory log errors and alarms since the battery's manufacture date

 

[Eng.Mohamed]

Although both batteries are of the same model and nominal specification (100Ah), there is a difference in effective capacity maybe caused by

• A manufacturing defect in the cells of one of the batteries.​
• Disparity in BMS or SOC Algorithm calibration.​
• Weak Cells in the second battery, which affects its discharge speed and readings.​

Evidence of this?
During an equal voltage and current discharge, the SOC can only drop faster if the actual capacity is lower.

 

[Eng.Mohamed]

I think Some BMS systems rely on estimation algorithms to calculate SOC (such as Coulomb Counting + Voltage Correlation).

If the algorithm is not accurate or is incorrectly calibrated (Calibration Error), the SOC reading will be unrealistic.

This is often the case with the second battery, as it appears to be discharging "faster", when it may simply be showing a lower percentage than what you actually it has or vice versa.

 

[Eng.Mohamed]

what about a new BMS update firmware ,is it a good step to fixes SOC mismatch and
other synchronizes batteries parallel?

 

[Eng.Mohamed]

This is common. I wouldn't worry about it.

Mr. WrenchLight
I may agree with you if you can technically prove that this is a common issue, or you can provide us with real-life examples, manufacturer's reports, battery management systems, or scientific research to address this issue.
I would like to say that we have checked and audited more than one storage system for similar systems and batteries with the same details, and we have never found such an issue.

 

[marionw]

Can you provide screenshots of the "Realtime Monitoring" tab for both batteries during the charging and when the PbmsTool software is showing the cells being balanced (BL will appear next to the cells that are being balanced.

As the cells approach 3.45 vdc you may want to lower the charge voltage and/or charge current to limit the charge current to less than 5 amps per battery. This may allow the BMS to balance the higher voltage cells and allow the lower ones to fully charge.

It appears that one or more cells are hitting Cell Over Voltage Protection before the BMS is able to fully charge the batteries. If each battery experiences a Cell OVP at different times during the charge they will by default not be charged to the same capacity because the BMS stops the charge when a cell hits Cell OVP.

 

[jgh]

I agree: common. Minor actual differences in cells (eg: electrolyte chemistry) result in voltage differences that, while *tiny* result in quite large different current (both during change and discharge) flows due to the very flat voltage curve for the major portion of the SOC range of LiFePO4. This in turn means different actual SOC even if the current measurement required for SOC-estimation is perfect.

Which it isn't, which adds another factor resulting in different reported SOCs.

The first factor doesn't matter because, as soon as either endpoint of the SOC range is approached, the voltage curve becomes steep - which results in a restoring current imbalance. Said another way: the battery that was lagging takes over the lead.

The second factor does not matter for any decent BMS because is resets it's estimation of SOC, triggered by a voltage measurement (which is far more simple to do sufficiently accurately than a current measurement) at one end of the operating range. Commonly the top end. So long as your charge supply reaches whatever voltage the BMS wants, for however long it wants it, rather than deciding to autonomously cut off - the BMS resets its SOC estimation to 100%.

Obviously there is potential for poor configuration resulting in such a reset not getting done.

 

[rogerheflin]

Mr. WrenchLight
I may agree with you if you can technically prove that this is a common issue, or you can provide us with real-life examples, manufacturer's reports, battery management systems, or scientific research to address this issue.
I would like to say that we have checked and audited more than one storage system for similar systems and batteries with the same details, and we have never found such an issue.

There are posts on multiple different manufacturers BMSes doing exactly this. Typically I have a variation of 3-5% in a day, and the more days I don't hit 100% the worse SOC compared between the batteries and compared to voltage gets. And some of the BMSes ONLY use charge counting and do not wait to report 100% until it sees the proper voltage spike (the JK bms used to report 100% when charge counting said to and would stop charging a few % lower each cycle--a firmware update fixed the JK, but other some other BMSes are still 100% relying on charge counting).

The base issue is inaccurate current measurement and inconsistent current measurement/calibration between units (being used for the charge counting). There is also a inaccurate current measurement across the range (ie the error seems to be % of full range so at 100A the error is say 1% but at 10A the error is 10% of value, and at 1A the error is significant--and most of us tend to charge in a much higher range than we discharge causing the error to be different in charge vs discharge). The defect seems to be large range (guessing since most BMSes can go -200A to +200A that the range is probably +=350A, and that they use an A/D converter that does not have enough bits-12bits limits you to .25A). I would wonder if one could have designed it more like a multi-meter that has parallel hardware for at least 2 ranges say one for full range and one for 10% of the range and measure both but use the best range for a given reading, but that clearly would have taken a bit more hardware and more software to implement.

It is also possible that if you have unequal resistance paths to the 2 batteries for one battery to be being used more. How are your batteries wired?

 

[HarryN]

Two questions:
- Can you post a photo and wire diagram of exactly how they are wired to the charging source and load? This can make a big difference.

- Do you happen to have a clamp meter to verify the results ?

This is how I wire two batteries in parallel to achieve "fairly similar" results.

Some of what you might be observing is the difference between using matched, high quality cells vs grade B cells.

 

[Eng.Mohamed]

Two questions:
- Can you post a photo and wire diagram of exactly how they are wired to the charging source and load? This can make a big difference.

- Do you happen to have a clamp meter to verify the results ?

This is how I wire two batteries in parallel to achieve "fairly similar" results.

Some of what you might be observing is the difference between using matched, high quality cells vs grade B cells.

Attached is an actual photo, as well as an explanatory diagram of the two batteries we mentioned in the report. From the actual photo of the two batteries, we can see that they are uniformly and equally coordinated in terms of the cables connecting them in parallel, as well as the connection to and from the charging source and load, which is the inverter.

The two batteries are connected in parallel and connected with DC cables of equal size and length for the positive and negative poles. Similarly, the cables connecting the batteries to the inverter are of the same size and length for the negative and positive poles.

Conclusion:
1) We would like to clarify that the two batteries are the same model, from the same manufacturer, and have the same technical specifications for the cells and BMS, and are not different in that regard.

2) We would like to say that we have installed many solar energy systems and storage systems, including batteries, inverters, and solar panels, for many manufacturers, and we have installed systems with the same components as those in this report, i.e., two batteries connected in parallel and connected to a 6 kW hybrid inverter, as well as the same sizes and lengths of DC cable connectors, but we have not encountered such a problem of SOC mismatch except in the two batteries mentioned in the report.

Mr. HarryN
Please review the attached photos as requested, note any technical observations you may have, and provide us with appropriate and practical solutions that will resolve and address the issue.

Thank you for your attention.

 

[Eng.Mohamed]

Can you provide screenshots of the "Realtime Monitoring" tab for both batteries during the charging and when the PbmsTool software is showing the cells being balanced (BL will appear next to the cells that are being balanced.

As the cells approach 3.45 vdc you may want to lower the charge voltage and/or charge current to limit the charge current to less than 5 amps per battery. This may allow the BMS to balance the higher voltage cells and allow the lower ones to fully charge.

It appears that one or more cells are hitting Cell Over Voltage Protection before the BMS is able to fully charge the batteries. If each battery experiences a Cell OVP at different times during the charge they will by default not be charged to the same capacity because the BMS stops the charge when a cell hits Cell OVP.

This was monitored using the BMS TOOLS program for the two batteries, and no BL appeared next to the cells.

As for the cut-off voltage for the cells ("Cell OV, Peak OV, Cell OVP, Alarm, Protect"), they are set to the same values for both batteries, as you can see in the Parameter Setting window in the BMS program. The question here is, how does the BMS program activate the protections in only one battery without activating them in the second battery and the connections in parallel?

The problem, as we can see in the error and alert log in the Memory Info window, a screenshot of which is attached, is that “Cell OV, Peak OV, Cell OVP” appear randomly and at different readings of the battery voltage and cell voltage.

Please review the settings window screen, the error and alert log window screen, and the files attached to the full report of the problem that we sent you.
Please review the attached report files and images, especially the data recorded by the BMS program for the two batteries. We would like you to do the following:
1) Please write down your technical observations in diagnosing the problem by reading the error and alert logs in the Memory Info file.
2) Please provide us with possible solutions that you deem appropriate to address the problem, whether in terms of adjusting the settings for the two batteries, such as voltage, charging and discharging protection, and other settings and adjustments.

Thank you for your attention.

 

[Eng.Mohamed]

I agree: common. Minor actual differences in cells (eg: electrolyte chemistry) result in voltage differences that, while *tiny* result in quite large different current (both during change and discharge) flows due to the very flat voltage curve for the major portion of the SOC range of LiFePO4. This in turn means different actual SOC even if the current measurement required for SOC-estimation is perfect.

Which it isn't, which adds another factor resulting in different reported SOCs.

The first factor doesn't matter because, as soon as either endpoint of the SOC range is approached, the voltage curve becomes steep - which results in a restoring current imbalance. Said another way: the battery that was lagging takes over the lead.

The second factor does not matter for any decent BMS because is resets it's estimation of SOC, triggered by a voltage measurement (which is far more simple to do sufficiently accurately than a current measurement) at one end of the operating range. Commonly the top end. So long as your charge supply reaches whatever voltage the BMS wants, for however long it wants it, rather than deciding to autonomously cut off - the BMS resets its SOC estimation to 100%.

Obviously there is potential for poor configuration resulting in such a reset not getting done.

Sir: jgh

In your reply, you mentioned that “it is clear that there is a possibility of malformation that prevents this reset process from taking place.”​

Please, how can we address this issue you mentioned? I hope you can provide us with details and possible steps for solutions that you deem appropriate.

Thank you for your cooperation.

 

[marionw]

This was monitored using the BMS TOOLS program for the two batteries, and no BL appeared next to the cells.

As for the cut-off voltage for the cells ("Cell OV, Peak OV, Cell OVP, Alarm, Protect"), they are set to the same values for both batteries, as you can see in the Parameter Setting window in the BMS program. The question here is, how does the BMS program activate the protections in only one battery without activating them in the second battery and the connections in parallel?

The problem, as we can see in the error and alert log in the Memory Info window, a screenshot of which is attached, is that “Cell OV, Peak OV, Cell OVP” appear randomly and at different readings of the battery voltage and cell voltage.

Please review the settings window screen, the error and alert log window screen, and the files attached to the full report of the problem that we sent you.
Please review the attached report files and images, especially the data recorded by the BMS program for the two batteries. We would like you to do the following:
1) Please write down your technical observations in diagnosing the problem by reading the error and alert logs in the Memory Info file.
2) Please provide us with possible solutions that you deem appropriate to address the problem, whether in terms of adjusting the settings for the two batteries, such as voltage, charging and discharging protection, and other settings and adjustments.

Thank you for your attention.

It's hard to tell since you do not show the "RealTime Monitoring" tab for each battery so we can see the voltage of each cell.
In order for the to activate a cell must exceed the "Balance Threshold(V) and the difference in voltage between the lowest cell and the highest cell must exceed the "Balance Delta Vcell(mv)" setting which defaults to 30mv. If there is less than 30mv between the highest and lowest cells the balancer will not be activated, thus you will not get the "BL" displayed next to each cell as displayed on the "RealTime Monitoring" tab

 

[frankrudis65]

Hi Eng.Mohamed,greetings to Yemen....I am not a fan of ready build battery banks ,cramed in a (metal)box ,connected to a bms of which condition is mostly a guess and a proper air circulation is hardly given.....I have build 17s DIY banks open on a wooden shelf,covered with plexyglass sheets.In that case you are actually able to examine,to test -for ex. temperature- and provide adequate air circulation.
In your case I'm afraid you would have to dismantle the whole lot,to be able to control even every single connection....and do a capacity test of every single cell.....The question is,if it's worth all the effortBest regards Frank

 

[Mark-]

Mr. WrenchLight
I may agree with you if you can technically prove that this is a common issue...

Here is a 6 battery rack, all the same manufacture. Here is the current state

When all batteries go into float voltage from inverter charging, I can put a battery charger on for 6-7 minutes and they all magically "pop" to 99.66% SOC.

 

[Eng.Mohamed]

It's hard to tell since you do not show the "RealTime Monitoring" tab for each battery so we can see the voltage of each cell.
In order for the to activate a cell must exceed the "Balance Threshold(V) and the difference in voltage between the lowest cell and the highest cell must exceed the "Balance Delta Vcell(mv)" setting which defaults to 30mv. If there is less than 30mv between the highest and lowest cells the balancer will not be activated, thus you will not get the "BL" displayed next to each cell as displayed on the "RealTime Monitoring" tab

Mr.marionw
This is screenshots of the "Realtime Monitoring" tab for battery 1 (Pack1) with 76% SOC , and battery 2 (Pack2) with 62% SOC as shown in the attached images during the discharging and when the PbmsTool is showing the Pack information and cells voltage (mV).
You can view the full details of the Realtime Monitoring for each battery, as shown on the BMS screen, where the difference in SOC between the two batteries was approximately 14%. The first battery had a capacity of 76% and the second had a capacity of 62%. As for the difference between the cell voltages of each battery between the Max-volt and Min-volt, as shown in the first battery was 1 mV, while in the second battery was 5 mV.

We are always happy to receive your feedback and comments.

 

[marionw]

Mr.marionw
This is screenshots of the "Realtime Monitoring" tab for battery 1 (Pack1) with 76% SOC , and battery 2 (Pack2) with 62% SOC as shown in the attached images during the discharging and when the PbmsTool is showing the Pack information and cells voltage (mV).
You can view the full details of the Realtime Monitoring for each battery, as shown on the BMS screen, where the difference in SOC between the two batteries was approximately 14%. The first battery had a capacity of 76% and the second had a capacity of 62%. As for the difference between the cell voltages of each battery between the Max-volt and Min-volt, as shown in the first battery was 1 mV, while in the second battery was 5 mV.

We are always happy to receive your feedback and comments.

Until you get both batteries charged to where cell voltages are at or above the Balance Threshold of 3.45 and the "VoltDiff" is greater than whatever is set (20mv is default as displayed on the "Parameter Setting tab "Balance Delta Vcell(mv)") then the BMS will not reset SOC to 100%.

You need to charge the batteries to where the BMS starts balancing the cells ("BL" will be displayed), which it will not happen if the "VoltDiff" displayed on the "Realtime Monitoring" tab does not exceed the "Balance Delta Vcell(mv)" as displayed on the "Parameter Setting" tab.

You may have to push the charge current to something around 10 amps or more. Generally I see the "VoltDiff" become greater the higher the charger current as the cells reach the upper knee of the charge curve. With a higher charge current you can drive the "VoltDiff" to exceed 20mv.

-Or-

Get the battery charged to "Balance Threshold(V)" (as displayed on the "Parameter Setting" tab) and hold the charge until the BMS resets SOC to 100%.

It may be that the BMS needs to see the battery voltage reach "Balance Threshold(V)" (as displayed on the "Parameter Setting" tab) and then see a Cell Over Voltage Protection fault.

I do not know for certain what the criteria is required for the BMS to reset SOC to 100%, but whatever it is you need to get both batteries charged to the same level and get the BMS to reset SOC to 100%

 

[Eng.Mohamed]

There are posts on multiple different manufacturers BMSes doing exactly this. Typically I have a variation of 3-5% in a day, and the more days I don't hit 100% the worse SOC compared between the batteries and compared to voltage gets. And some of the BMSes ONLY use charge counting and do not wait to report 100% until it sees the proper voltage spike (the JK bms used to report 100% when charge counting said to and would stop charging a few % lower each cycle--a firmware update fixed the JK, but other some other BMSes are still 100% relying on charge counting).

The base issue is inaccurate current measurement and inconsistent current measurement/calibration between units (being used for the charge counting). There is also a inaccurate current measurement across the range (ie the error seems to be % of full range so at 100A the error is say 1% but at 10A the error is 10% of value, and at 1A the error is significant--and most of us tend to charge in a much higher range than we discharge causing the error to be different in charge vs discharge). The defect seems to be large range (guessing since most BMSes can go -200A to +200A that the range is probably +=350A, and that they use an A/D converter that does not have enough bits-12bits limits you to .25A). I would wonder if one could have designed it more like a multi-meter that has parallel hardware for at least 2 ranges say one for full range and one for 10% of the range and measure both but use the best range for a given reading, but that clearly would have taken a bit more hardware and more software to implement.

It is also possible that if you have unequal resistance paths to the 2 batteries for one battery to be being used more. How are your batteries wired?

As I explained earlier, before starting the discharge, I charged each battery separately until it reached 100% SOC and allowed each battery to rest for 24 hours to stabilize the voltage.
Then,I connected them in parallel and charged both batteries to the same full level, after which I monitored their discharge over a whole month and noticed a continuous discrepancy in SOC.

For information, the components of the solar system are considered ideal according to the engineering design, and the batteries charge well and reach 100% early in the morning , like at 11 A.m, the batteries are fully charged and begin to discharge at sunset, as you can see in the attached image of the inverter monitoring program screen. The loads are also well adjusted and evenly distributed between the two batteries, as we noticed on the screen of each battery separately, but the problem of soc mismatch remains.

Finally, I ask my engineer friends to share their engineering expertise in addressing the problem and provide us with the appropriate solution.
We are always happy to receive your feedback

 

[rogerheflin]

As I explained earlier, before starting the discharge, I charged each battery separately until it reached 100% SOC and allowed each battery to rest for 24 hours to stabilize the voltage.
Then,I connected them in parallel and charged both batteries to the same full level, after which I monitored their discharge over a whole month and noticed a continuous discrepancy in SOC.

For information, the components of the solar system are considered ideal according to the engineering design, and the batteries charge well and reach 100% early in the morning , like at 11 A.m, the batteries are fully charged and begin to discharge at sunset, as you can see in the attached image of the inverter monitoring program screen. The loads are also well adjusted and evenly distributed between the two batteries, as we noticed on the screen of each battery separately, but the problem of soc mismatch remains.

Finally, I ask my engineer friends to share their engineering expertise in addressing the problem and provide us with the appropriate solution.
We are always happy to receive your feedback

NOTHING in the above changes what I said. You can do the entire process as RIGHT as possible and it still won't work because of BMS design issues.

The BMS was designed ideally by "EXPERT" "engineers" based on theory, they ignored the underlying reality of the real components that they were using. They failed to understand that they could not measure current accurately and the inaccuracy was different in different ranges and did not understand this would cause SOC calcs to be inaccurate if charging and discharging were in different current ranges. And I bet money they "TESTED" it by using the same charge and discharge currents masking the design defect. And no one also appears to have calibrated the BMSes and so even if the underlying component is +%1 of full range, full range is -200A to +200A so 1% off is 4A. And at 50A operating(charge for me) that is a 8% error, and at 10A(typical discharge for me) that is a 40% error. And with solar about 2-3x more time is spend in the discharging area.

You don't seem to actually be listening. And you seem to be saying the whole system was designed correctly and because of it being designed correctly then there cannot be an SOC issue (but there is, so maybe your system is not designed as ideally as you think it is, at some level).