Winston LiFePO4 battery problem?

[Eirik]

Hi,
In my boat I have the following LFP battery bank: 8 Winston 300 Ah in a 2P4S setup, yielding 600 Ah/13V. The bank/setup is further described further below.

During 2019 I have noticed that battery cell set #4 (i.e. 2 x300 Ah cells) has started to behave a bit off: During charging it has a higher voltage than the other cell sets, and during discharge the cell set’s voltage is lower. The difference increases in relation to the rate of charge or the discharge. If the charge current is around 6-10 A (i.e. 0.01-0.02 C), cell set #4 is about 10-20 mV above the other sets. (i.e. 3.165V vs. 3.15 V), but if the charge current is 70 A (i.e. around 0.12 C), the voltage difference increases to 100 mV (3.25V vs 3,15V). The same differences occur when discharging, but then of course in the other direction, and with roughly the same magnitudes.

Furthermore, the voltage on cell set #4 also seems to flutter a bit, i.e. at times (lasting around 10 sec) the difference is reduced, but then increases. All this is measured by the BMS’ volt meters, and possibly follows the refresh rate of the BMS.

Finally, as the battery bank gets fuller when charging, the difference in voltage increases further. This has also resulted in the BMS kicking in and triggered the HVC contactor (to stop charging) because the cell set reached 3.7V (max cell threshold), whilst the others were at around 3.37V (i.e. 13,8V), and the bank hence not being fully charged, with a reported SOC around 80%.

I am thus wondering what the issue could be. Are the cells out of balance (which doesn’t make that much sense since the voltage goes in both directions depending on charging or discharging), or could it be that one or both of the cells in the pack are damaged? Open for suggestions for figuring out the issue and possible remedies.

Information about the bank:
The bank was commissioned in a sailing boat in May 2018, and then top balanced at 3,65V per cell. It has not been balanced since commissioning. I use the REC Active BMS to protect the bank, which also does active balancing. Charging is mainly done via the Mitsubishi alternator (standard on VP engines), which I have reduced to 80 A (Using VRC 100 from Nordkyn), and after deducting from the boat’s overhead, rarely giving more than 70A to the bank. In addition, I have 2X100W solar panels, yielding rarely over 10-12 A (Victron MPPT 75/15 controller), and a Victron Multiplus 1600-70-12 which is rarely used for charging. Max voltage on all charging sources is 14.0 V, and “float”/”hold” is set at 13.3V. Absorption (14.0V) is at max 30 min. In practise, it is the alternator that fills up the bank (solar charging will usually at best cover current consumption, and the Multiplus with shore power is hardly used).

The bank has (apart from the issue described above) worked nicely and also provided plenty of juice for both the usual 12V DC applications as well as 230V AC applications via the Multiplus. The latter is regularly used, and may then draw up to 130A (0.2 C), but usually less. The SOC is rarely brought below 50%, and only occasionally up to 100%. Usage has been for coastal day sails, weekend sails and a few 1-3 weeks sails during summer season. All charging and discharging sources are usually disconnected when the boat is at the dock, and the bank is usually kept at an SOC of 70%-80% (and 50% in winter, when it is not being used).

SOC is calculated by both the BMS and a Victron BMV 712 (peukert 1.05), and they remain very close, usually within 1-2% percentage points. I have not done a capacity test of the bank.


Thanks,

Eirik

[evm1024]

Hi Eirik,

First off it sounds like a nice installation you have there.

It appears to me that cell set #4 has reduced capacity in relation to the other cell sets. One or both of the cells that set is “bad” and should be replaced. Bad in the sense that they have less capacity.

In analogy (for others if you are up to speed on this) each cell is like a bucket. Charging is adding cups of water to each bucket and discharge is taking a cup of water from each bucket. When all the buckets are the same size they fill up to the same level as you add cups. And empty as you take them away. The water levels in the buckets are the same and thus the voltage levels are the same.

When one bucket is smaller in volume than the others it fills up quicker than the others and empties quicker too. And thus, the voltage of the reduced capacity cells rise quicker on charge and drop quicker on discharge.

Of course, it is more complex than this but that is the basics.

Without cell level protection your bad cells could reach nearly 3.9VPC given your 14 volt bulk charging voltage. 14 – (3 * 3.37) = 3.89 That is too close to the Winston max VPC for comfort.

How did the cell go bad? Who knows. A marginal cell to start with? Some physical defect? Some other cause? It is not clear to me what is the root cause. It might even have been the active balancing throwing things off and causing one cell set to over or under voltage leading to reduced capacity.

I have a set of 4 Winston 700 AH cells in my house bank with a weak cell or 2. They do much the same as your cells in terms of one charging to a higher voltage than the others. What I did was to charge the bank fully (as best I could) then did a capacity test. My 700 AH on paper bank has about 500 AH of capacity. I updated the SOC parameters to reflect this and then top balanced the cells to each other. Also, I adjusted the low voltage parameters to stop discharge before the weak cell went too low. Raising the LVC voltage as it was.

In your case all cell groups would be brought to 3.5 VPC in parallel for a top balance and the bank reassembled. This way charging would not be pushing the weak cell too high and with the LVC raised the weak cell would not go too low.

The real answer is to find and replace the weak cell(s). The same for my bank…

One last thought – I use a high quality Fluke multimeter to measure cell voltages. With that you can check the calibration of the BMS and BMV voltage measurements to be sure that they are accurate.

In regards to the fluctuations in cell voltage, I have no “answers” for that. Is it an artifact of the sampling period or BMS issues? Hard to say. A Fluke could help tell if it is real. If it is real I would guess that it is the balancing resistors turning on and off. I have removed the balancing resistors on my cell boards just to avoid any issues they might cause in failure or by design. As others have noted LiFePO4 stays wonderfully in balance for years.

Regards

[rvlandlubber]

Cell voltage fluctuations could be caused by oxidised cell interconnects or bms connections.
I would lightly sand the interconnects and cell terminals, re assemble with a light smear of Aluminox and re torque to spec.
Also check the multi pin connector that plugs into the REC bms.

[john61ct]

I'd take the BMS out of the picture and do a set of thorough testing cycles of the cells without it.

Once four good cells are identified, do thorough testing of the BMS using a known-good if not calibrated ammeter / DMM

Do not put back "into production" until source of the problem(s) identified.

[john61ct]

There will always be some cells with more or less capacity than others, especially as the bank wears.

This is not in itself a problem, so long as going anywhere near "the shoulders" is either avoided manually, or "fenced in" at the per-cell level by the BMS.

Of course overall bank capacity will be limited by the weakest cells.

There are dedicated cell balancing devices that can do a much better job than the circuitry included in a BMS, in which case the latter's should be disabled.

This may then allow you set the charge-finish voltage a bit lower, usually a good move both for longevity and reducing the impact of balancing issues.

[rvlandlubber]

Sorry, typo

Aluminox should read Alminox.

Cheers

[CatNewBee]

Originally Posted by Eirik

Hi,
In my boat I have the following LFP battery bank: 8 Winston 300 Ah in a 2P4S setup, yielding 600 Ah/13V. The bank/setup is further described further below.

During 2019 I have noticed that battery cell set #4 (i.e. 2 x300 Ah cells) has started to behave a bit off: During charging it has a higher voltage than the other cell sets, and during discharge the cell set’s voltage is lower. The difference increases in relation to the rate of charge or the discharge. If the charge current is around 6-10 A (i.e. 0.01-0.02 C), cell set #4 is about 10-20 mV above the other sets. (i.e. 3.165V vs. 3.15 V), but if the charge current is 70 A (i.e. around 0.12 C), the voltage difference increases to 100 mV (3.25V vs 3,15V). The same differences occur when discharging, but then of course in the other direction, and with roughly the same magnitudes.

Furthermore, the voltage on cell set #4 also seems to flutter a bit, i.e. at times (lasting around 10 sec) the difference is reduced, but then increases. All this is measured by the BMS’ volt meters, and possibly follows the refresh rate of the BMS.

Finally, as the battery bank gets fuller when charging, the difference in voltage increases further. This has also resulted in the BMS kicking in and triggered the HVC contactor (to stop charging) because the cell set reached 3.7V (max cell threshold), whilst the others were at around 3.37V (i.e. 13,8V), and the bank hence not being fully charged, with a reported SOC around 80%.

I am thus wondering what the issue could be. Are the cells out of balance (which doesn’t make that much sense since the voltage goes in both directions depending on charging or discharging), or could it be that one or both of the cells in the pack are damaged? Open for suggestions for figuring out the issue and possible remedies.

Information about the bank:
The bank was commissioned in a sailing boat in May 2018, and then top balanced at 3,65V per cell. It has not been balanced since commissioning. I use the REC Active BMS to protect the bank, which also does active balancing. Charging is mainly done via the Mitsubishi alternator (standard on VP engines), which I have reduced to 80 A (Using VRC 100 from Nordkyn), and after deducting from the boat’s overhead, rarely giving more than 70A to the bank. In addition, I have 2X100W solar panels, yielding rarely over 10-12 A (Victron MPPT 75/15 controller), and a Victron Multiplus 1600-70-12 which is rarely used for charging. Max voltage on all charging sources is 14.0 V, and “float”/”hold” is set at 13.3V. Absorption (14.0V) is at max 30 min. In practise, it is the alternator that fills up the bank (solar charging will usually at best cover current consumption, and the Multiplus with shore power is hardly used).

The bank has (apart from the issue described above) worked nicely and also provided plenty of juice for both the usual 12V DC applications as well as 230V AC applications via the Multiplus. The latter is regularly used, and may then draw up to 130A (0.2 C), but usually less. The SOC is rarely brought below 50%, and only occasionally up to 100%. Usage has been for coastal day sails, weekend sails and a few 1-3 weeks sails during summer season. All charging and discharging sources are usually disconnected when the boat is at the dock, and the bank is usually kept at an SOC of 70%-80% (and 50% in winter, when it is not being used).

SOC is calculated by both the BMS and a Victron BMV 712 (peukert 1.05), and they remain very close, usually within 1-2% percentage points. I have not done a capacity test of the bank.


Thanks,

Eirik

Hi Erik,

Sad to hear that. You need to track down and troubleshoot your setup first and then find a solution.

The voltage difference while charging and discharging, resulting imbalances etc. always results of differences in resistances. And this means not only the obvious cell inner resistance, but also all contacts, bridges, screws, BMS contacts on the cells etc. The smaller cells like the one you use have only a single contact per battery pole and are very vulnerable to contact errors.

The other thing is, the LiFeYPO4 cells are non linear regarding cell resistance on the shoulders, when a cell reaches this SOC, the resistance rises almost exponentially, small changes result in big jumps.

Cells between 20 and 90% SOC have usually some mOhm, while the contacts have some 10 to 100 mOhm because of surface contact and pressure, material, size of the bridges, a little more or less tension, unclean or uneven surface on a single connection can lead to less charge on a paralleled cell, what will cause in capacity imbalance. Just draw a circuit diagramm of yout battery and add on the battery cell and each connection a symbol for a resistor, you will see a meshed resistor matrix, small changes lead to different voltages and currents.

So empirically you can start by firing up your inverter under heavy load and touching each screw and bridge on the battery by hand, try to feel temperature differences. If there is a warmer connection, tighten the bolt a little more or clean the contact. If all cells are equally warm, you likely have a problem either with a cell or the BMS.

To sort out the bms you can use a clamp amperemeter and check there is a balancing current when charging above the start balance voltage on one cell. If it is, the BMS is fine.

Next would be to check each individual cell, you need to disconnect the pairs and do a test. If you have a notorios low cell, that always is charged and one that always is discharged, you can swap the partner cell, to make both equal capacity.

I have similar issues when the battery is near full, my inner cells are always a little different then the outer cells of the pack, on high charge current about 200A (0.2C), and full battery it can be 90..100mV, but this is normal, very little SOC differences lead to dramatic resistance changes, with smaller currents the voltage difference is smaller. Your BMS balances with 2A, it is one percent of this current, you see the voltage jump of few millivolt in this SOC area on the BMS when the balancer is active or turns off to measure. It is normal there, but it should not lead to disconnects. If your BMS disconnects, there is either a misconfiguration or a big imbalance. You have to allow some margin for balancing, but 3.7 vs 3.37 is definitely out of range.

[MikeFergie]

Personally I have not used a REC BMS but was going to buy one for my system so I can’t comment on it but I prefer series linked balance units outside of the BMS. I’m not sure if “every cell” is linked to your BMS or only the groups of parallels ?

It could be that you need a proper top balance again and I would do it by disconnecting all the cells and charging them up with a single cell charger to 3.65v each and reconnect them. I have separate cell balancers and when I connect all my cells back to make a battery pack (3P4S) and charge or discharge if there is any differences the particular cell light comes on the cell balancer for that cell and it starts equalizing. I have seen one cell in a parallel grouping light showing red but the other 2 are green still showing me it’s just that cell balancing.

I would then do a deep discharge and recharge back up to 95% watching to see if any particular cell light comes on.[ATTACH]204954

[CatNewBee]

Originally Posted by MikeFergie

Personally I have not used a REC BMS but was going to buy one for my system so I can’t comment on it but I prefer series linked balance units outside of the BMS. I’m not sure if “every cell” is linked to your BMS or only the groups of parallels ?

It could be that you need a proper top balance again and I would do it by disconnecting all the cells and charging them up with a single cell charger to 3.65v each and reconnect them. I have separate cell balancers and when I connect all my cells back to make a battery pack (3P4S) and charge or discharge if there is any differences the particular cell light comes on the cell balancer for that cell and it starts equalizing. I have seen one cell in a parallel grouping light showing red but the other 2 are green still showing me it’s just that cell balancing.

I would then do a deep discharge and recharge back up to 95% watching to see if any particular cell light comes on.[ATTACH]204954

If cells are parallel, any balancer will balance both cells together, cell modules on each cell in parallel configuration make no sense.

There is no advantage of having separate cell modules on the cells, makes wiring and shortcut protection more complicated and also error prone, you have to daisy - chain the OVP / LVP signals for the charge and discharge bus, also reaction on currents, soc and temperature are more difficult to implement, you likely need a data bus between the modules or again to a central unit.

[MikeFergie]

CNB I get your point of individual balancers being an overkill as the buzz bars between the cells should balance them however I just see it as redundancy and I don’t see how it complicates the OVP or LVP as the BMS is separate and has wires going to each parallel string to measure the voltage for itself. The temp is the same. I have a shunt fitted to the neg terminal as it leaves the BMS and this feeds info correctly as far as SOC and amp hours in and out goes

[john61ct]

Originally Posted by MikeFergie

I don’t see how it complicates the OVP or LVP

With fragile electronics less is better, the BMS circuitry failing is more likely to murder cells than the cells themselves failing.

Once cells are paralleled, they functionally become one cell, there is nothing different between them a BMS can measure.

Better off using bigger cells, again, fewer points of failure.

[MikeFergie]

CNB I get your point of individual balancers being an overkill as the buzz bars between the cells should balance them however I just see it as redundancy and I don’t see how it complicates the OVP or LVP as the BMS is separate and has wires going to each parallel string to measure the voltage for itself. The temp is the same. I have a shunt fitted to the neg terminal as it leaves the BMS and this feeds info correctly as far as SOC and amp hours in and out goes

[CatNewBee]

thats what i am saying, cell modules plus extra bms, complicated and more connections waiting to fail. I prefer well built systems with less connections and safely enclosed.

[nebster]

I think it's pretty likely you have a failing cell. Not just a normal degradation or a slip in balance, but one that has deteriorated quite a bit.

The good news is that I bet it'll be easy to find because of how bad it's gotten and, assuming you can source a replacement wherever you are, things will likely go back to normal when you swap it out.

[peter57]

Winstons for long life should not be charged over 3.5 Volts as there is no advantage to charge over 3.5 Volts for storage.