New EVE 12-cell Marine House Bank

[Dave McCampbell]

We've just commissioned our new 3P4S 12v EVE LF280K prismatic cell house bank with useable capacity around 900ahrs. Max charge will be about 100amps, max load about 150amps. We are using our Electrodacus SBMS0 for BMS. I have wired it as per the attached photo. A couple of questions:

1. Should I double up on the 3 connecting bars between the 4S sets of cells?

2. Are the main battery cables placed properly at the final cells on opposite ends of the U shape configuration?

3. What is the proper terminal nut torque for these welded studs cells?

4. Any other suggestions for improvement?

[sailingharry]

My suggestions:

*  Where possible, you should connect to the middle cell of each 3P.  For the battery cables, move in one cell.  For the U-Turn, move to the middle cell.  For the first and third, it would be awesome if you could use a cable from middle-to-middle, but that may be a bridge too far.  At the very least, the jumpers (the ones between each 3P) should be doubled up.

*  Measure the actual copper in those jumpers, and compare that to this site:  https://www.copper.org/applications/electrical/busbar/bus_table1.html  You will find that the busbar vendors massively overrate their bars.  They are woefully inadequate.  Look at your fuse size (or at the very least your realistic ongoing current), and see the cross section required.  If you are fused at under 200A, you are probably OK.

*  You have two problems and one solution.  Your batteries are not held down, and you have a risk of shorting terminals across the cells.  If you lay a 2x2 board, held down at each end, over each row of cells, you will minimize risk of dropping a wrench across the terminals, and also hold them down.  I make a groove over each vent spot so I don't block them.

*  The solution above doesn't protect between bank 1/2 and 3/4.  Consider that.

*  You also have a risk between 1 and 4.  You have a barrier plate running down the length, but I can't tell how high it is.  Mine protrudes up, so it again minimizes shorting risk.

*  In those hold-down boards I also ran a groove length wise, and inserted a string of thermal fuses (about 150F, I forget the actual number) set over each cell, so that if any cell overheats, it trips the contactor.

[Dacian Todea (electrodacus)]

1. Not sure what the characteristics of those bars are but at 150A max load is likely fine.

2. It is not optimum but for your application it will work fine.

3. 6Nm

4. Optimum will have been to have the groups of 3p cells rotated by 90 degrees.  That way you can have 3 connections. But I do not think you have the space in that box to rotate those cells as 3x72mm = 216mm so more than width of a cell 207mm and box seems to be exact right size as it is.

Just apply max possible load for 15 minutes then check if all connections are around ambient temperature or if you have a thermal camera even better check battery but also all your other connections up to your load. 

[Dave McCampbell]

Thanks to Harry and Dacian for the comments.
1 Regarding cell bussbar carrying capacity, I found on BatteryDesign.net that the capacity for good quality Cu bars was 5.0-5.9A/sq mm. All bussbars that I've seen provided with cells are 2mmx20mm.  I was able to find quality 3mmx20mm connecting bars, so 60mm squared times 5 equals a minimum of 300A carrying capacity. Plenty I think for my needs, and the extra 3mm threaded hole in these bussbars is perfect for connecting voltage sense leads, etc. without having to use the cell stud connection.  The flexible braided Cu bussbars, like Dacians, have 5mm thick terminal ends, so would not work with 10mm long EVE studs.
2 I cleaned all the cell bussbar connections with vinegar/water mix and isopropyl alcohol and torqued them to 5 Nm so there was no danger of exceeding EVE torque spec of 6 Nm max.  The small electronic torque guage available on the internet for about $30 is perfect for doing this.
3 Since the EVE cell terminal stud length is just 10mm there is not enough room to do more than the required in some places double the 3mm thick bussbars and add the nice 4mm thick provided nut with the bussbars.  That should be plenty for my use. 
 
Dacian, I should have checked first here before I built my box so I could have gotten the optimum arrangement.  Next time.  I will check to see if any connections develop heat.
Harry, thanks for the several suggestions.  As mentioned above, given the constraints with the short cell studs, I can't do much more than I have already done with additional connections, ie two 30mm bussbars stacked on top of each other.  Connecting anything more to the middle cells with already two bussbars connected would be impossible.  I do have a hold down system, not shown, to keep the cells in place if the boat inverts.  Re protection for the cell sides all have plastic between cells and between cells and box sides.  Also, there is a significant amount of cushioning under the box and between the box and bottom of the cells.

[Dacian Todea (electrodacus)]

the 60mm^2 bars about 8cm long will have a resistance of around 0.025mOhm so at 300A you get about 2.25W of heat and with the cells terminals helping with cooling it should not be a problem.  The contact resistance needs to be low but if you cleaned the surface and applied 5Nm then it should be OK.

[sailingharry]

The important part of my comment regarding the bus bars is to ensure you design for it.  You've done that, so you're set!  Ali is the most diverse source for bus bars, but some vendors are a little....loose?... with their stated specs.

[Dave McCampbell]

Thanks Dacian and Harry.  I have one more issue:

For some time now both our temporary 4 cell 4S battery and our new 12 cell 4S3P battery have had a problem with the first and last cell voltages going high early during charging.  See the screen shot on our Node Red display below.  This divergence starts at about 3.42vpc with cells 1 and 4 shooting high and cells 3 and 4 remaining around 3.42vpc until the end of charging.  At HVD 3.55vpc the spread between the two high and 2 low cells is well over 100mv.   The drop in voltage between the first and second spikes was because we had to turn off charging for a while, but the end result near the top of the charge is the same. 

For the past year, while hauled out under a shed with no solar, we have been using our shore charger charging at about 45 amps.  This rate was less than .2C for both batteries.  Although the temporary battery may have had balance problems to begin with, we carefully top balanced the entire new battery cells, but it still exhibits what appears to be the same cell divergence problem.

Our supply/charge positive and negative cables are connected at the first and last cells as shown in the image above.  I am wondering if the positioning of those cables would have anything to do with the uneven divergence of the charge voltages.  As an option I could make the cable connections to the battery at cell terminals 4 negative and 9 positive.

[Dacian Todea (electrodacus)]

Dave,

There is zero problems with any of the cells. This is typical behavior for LiFePO4.

The cells start to diverge for the last 10 minutes and average charge current over those last 10 minutes seems to be 30A (starting from around 40A and dropping to 20A).

30A * 10/60 = 5Ah 

5Ah/280Ah = 1.8% delta between cells.  While this seems a bit more than typical 0.5% to 1% that I observed on my batteries is not significantly high.

Not sure if you have cell balancing enabled or not as that may explain the higher closer to 2% deviation vs less than 1% expected. Or maybe this is not a 280Ah battery as I assumed.

If this is the 4S3P battery in the first post then that will be a 840Ah battery and then 5Ah / 840Ah = 0.6% deviation and will be as expected and as I also measured on my cells.

Where the cells sense are located makes little if any difference at this low charge currents so that has nothing to do with your readings at the end of charge. And the place you chose to connect the sense wires are perfectly reasonable for that setup.

[Dave McCampbell]

Thanks for that Dacian.  The balancing was not on at that time of the screen shot and it is for the larger 4S3P 840ahr battery.  My concern is that cells 2 and 3 seem to stop charging completely while the others, 1 and 4, continue to receive all the charge, although the connections were all cleaned and torqued evenly.  So you don't think my connecting of the main battery cable at cells 1 and 4 have anything to do with those cells continuing to charge while the other two do not?

[Dacian Todea (electrodacus)]

Dave,

Not sure I understand what you mean by "cell 1 and 4 continue to receive charge" cells 2 and 3 where also charging. It is impossible for just some of the cells to charge. Is either all or none.

I asumme that a few seconds after you took that screenshot the charging stopped as cell 1 exceeded 3.55V.

[Dave McCampbell]

Dacian. 

By my statement above that  'cells 2 and 3 seem to stop charging completely while cells 1 and 4 continue to receive charge', I mean that cells 1 and 4 voltages continue to rise while cells 2 and 3 voltages stop rising and just remain the same for the final few minutes of charging before HVD.  This happens  starting around 3.420vpc which seems rather early. 

I would have expected that all voltages would continue to rise, but those for cells 1 and 4 would just rise faster than those of cells 2 and 3.  This exact same thing happened with cells in the same positions while using the temporary 4S battery earlier, but using entirely different cells.  What would could cause voltages for cells 2 and 3 to completely stop at that point while voltages for cells 1 and 4 continue to rise?  Maybe something to do with my wiring or the connection point for my main battery cables on cells 1 and 4?

[Dacian Todea (electrodacus)]

Same current flows trough all cells especially since you disabled the cell balancing.

Voltage very slightly decreases on cells 2 and 3 because the charge current decreases from a bit over 40A down to around 20A over that 10 minute period.

Cells 1 and 4 where already full so they had no more space to store electrons thus the voltage rapidly increased.

From around 12:30 to 12:45 you can better see all cells voltage continue to increase just cells 1 and 4 at a faster rate. Not quite sure what stopped the charging then since no cell got to 3.55V and some similar thing happened around 12:01 when charging stopped for no apparent reason for a few minutes.

Also a bit before 11:00 when charging first started you see that initial change in voltage over maybe around 10 minutes after charging started but then all cell voltage remained constant for the next hour or so.

If you where able to continue charging it will have taken maybe no more than 10 to 15 minutes before cell 2 and cell 3 will have also reached 3.55V.

Cells are not equal they have different capacity and internal impedance. So every time you charge and discharge them some of the energy will be lost as heat and that amount of energy lost as heat is not equal for all cells.

The cell balancing will keep the pack from exceeding 1% imbalance but can not keep them at 0% imbalance (not for LiFePO4). For other type of lithium cells where charge discharge curve is less flat cell balancing can do a better job and keep the pack at maybe as low as 0.1% imbalance.

If you disable cell balancing you could see the imbalance of the pack increase by about 2% per month (based on my experiment a few years ago). But if you keep the cell balancing enabled then pack imbalance will be maintained below 1% typical around 0.5%

The point is that there is nothing abnormal about your pack and you can not expect to have better pack balance unless you are willing to waste a lot of time and energy at the end of each end of charge to just decrease the imbalance from typical 0.5% to maybe 0.1%.  There is nothing to be gained by doing so.

You can see the same thing on my pack.

At around 9:07 I had a full charge for the day and I was charging at that time with about 1000W / 26V = 38A so about the same current as in your case is just that I have a 8s2p

It just happens that cell 1 was also the first to get to 3.55V and charging stopped.

It is a bit hard to see due to cell balancing but cell 6 and 7 are at around 3.47V and if you look at the graph cell 1 was at that voltage with 2 minutes earlier meaning that cell 6 and 7 are just 38A / (60/2) = 1.26Ah behind cell 1 so 1.26Ah/560Ah = 0.225% 

Cell 3 looks like the lowest after charge ended but during charging cell 5 was lowest at around 3.43V and cell 1 was around there about 4 minutes earlier thus around 0.45% delta from cell 5

At this point my battery is a few years old and this cell delta deviation remains around this typical 0.5% delta. 

[Dave McCampbell]

Many thanks for that detailed explanation Dacian.  It is interesting that your first and last cells, 1 and 8, are exhibiting the same performance as my first and last cells, 1 and 4.  We spent quite a bit of time capacity testing and top balancing these 12 well matched EVE cells, so I think their balancing is  pretty close.  Are your main battery cables also connected to your first and last cells? 

I turned off the SBMS0 balancing temporarily to better see what was happening to the cell voltages during charging.  Now I will turn it back on.  Also, I manually turned off the charging for a bit in order to reset the shore charger as it only charges for an hour at a time.  It is a pretty lame Sterling charger which is advertised to have a LFP charging setting but doesn't really.  So we have to use a custom setting that only goes for an hour.  I'll be glad when we can get back to using solar for charging through our two Victron MPPTs.  Dave

[Dacian Todea (electrodacus)]

This sort of cells have fairly low internal impedance so the busbars and connections can add some significant increase to the internal impedance.

Here is a photo of my battery cell 1 is top right and cell 8 is bottom right. So yes cell 1 voltage measurement does not include a bussbar  but cell 8 does.

If cells will have been perfectly equal in characteristics (capacity and internal impedance) and the cell connections and bussbars where also perfectly equal then cells will stay in perfect balance with no need for any cell balancing.

But the characteristics are not equal so cells will get out of balance typical 0.5% per month so could be 6% per year but if cell balancing is used the cell imbalance will be forever limited to typical 0.5% max 1% for LiFePO4

[Dave McCampbell]

Update:

All seems well now with charging and balance for my new EVE 4S3P house bank.  After a couple of weeks and a few balancing sessions with my big 10a NEEY active balancer things seem to have settled out.  Cells are charging relatively equally and no indication of stalled cells.  Currently my MPPT charge termination is set 14.0v and SBMS0 HVD at 3.55vpc.  Thus the MPPTs are doing and terminating the charging and the SBMS0 will do a HVD if any cell goes high.  

With no balancing all the way to test charge termination at 14.0v using the Victron 100/50 MPPTs, the SBMSO is reporting 15mv cell voltage difference.   My very accurate UNI-T multimeter and the NEEY both report about 8mv difference.  After balancing using the NEEY, it and the UNI-T agree at about 2mv difference at charge termination.  The SBMS0 measures about 8mv at the same time.  I assume that this difference in readings is due to some imbalance in resistance in the wiring or connections between the cells and SBMS0.  Those readings have been consistent over the past week.

Next week I will reconfigure parameters to use the SBMS0 200ma passive balancer to see if it can do a proper balancing job long term.  I suspect that with much better matched cells and more careful assembly it can.  Overall this is a much better situation than what I was experiencing earlier with up to 100mv difference in cell voltages at charge termination.

[sailingharry]

A couple of questions, if I may?

*  I believe the SMBS0 has a single voltage (Type 1) that both terminates the charge and resets the Ah counter.  If the MPPT terminates successfully on its own accord (rather than the Allow-To-Charge, or ATC), do you ever reset the Ah timer?  There is a school of thought (s/v Jedi, for example) who firmly believe that the BMS should only be a backup, not primary (which is what you seem to be doing), but I am fine with the BMS being primary, and the MPPT being backup (you still have two independent controls working in safe voltages).  For me, I set the MPPT at a very small delta (.1V?) over the ATC. HVD is higher still -- no one ever wants that to happen. (On a re-read, I might be misunderstanding -- do you use the Type 1 to control the MPPT at all?  On my boat, all charge sources shut down on their own at a set voltage, but the BMS can also shut them down -- the Alternator and shore charger shut off before ATC, the MPPT is just above ATC so the solar brings the battery up to the Type 1 reset voltage).

*  Is 15mv delta REALLY an issue? Up in the knee, what's of key importance is that they are all close, but if it's 15mV vs 2mV, I'm not sure it matters.  It would be interesting to measure the Ah to create that last 15mV rise before charge termination -- delta Ah is the real measure of imbalance, but since we can't easily measure that, we use Volts.  If the delta is under 1Ah, I'd lose interest in solving it!  It is necessary, regardless of errors in measurement, to "balance as close as we can" -- so if the BMS forces a technically perfectly balanced system to some less-than-perfect balance, that's fine -- if we didn't try at all, it would eventually be WAY out of balance.

*  I'm not sure I agree that resistance matters (in fact, I am not sure I agree with Dacian on his concern about fuses in the sense lines), as the current flow should be approximately zero.  At the currents that exist in the sense wires (sub 1mA, I suspect?), it would take a 15 ohm "bad connection" to make a15mA voltage drop -- and I'm not sure I could intentionally make a connection over 1 Ohm!  I used 7A pico fuses in my sense lines, with a rated resistance of 0.01 Ohms -- at 1mA current, that's a 0.01mV delta (ie, .00001V, or 1/100th of what the BMS0 can display).

Of course, this is all just musing.  It's these mind games that build my understanding (and hopefully, my sharing them in open discussion helps others).  Your system is already highly successful -- I'm not casting stones!  

[Dacian Todea (electrodacus)]

Dave,

My large battery is an 8s2p and after a few years now it is still within less than 0.5% delta SOC between cells. So if cells are OK SBMS cell balancing will keep them in balance well below 1% SOC delta between cells.

At the end of charge above 100mV delta is normal and basically impossible to be less. A few minutes after charging has ended depending on load the cells will return to much lower delta. This is the nature of how this LiFePO4 cells work and very different from NMC or other high energy density cells where voltage and SOC are fairly correlated.

The MPPT even set at 14V will not be the one to stop the charging. Typically 3 of your 4 cells will be very close to fully charged 99.5% SOC while the 4th cell will be at 3.55V 100% charged and that should terminate the charging thus 3.4*3 + 3.55V = 13.75V (this is just an average). But if you ignore SBMS0 and force charge above that then that 4th cell can get to 3.8V well above what will be normal for LiFePO4

If you manually cell balance the cells it will only take a few days or few charge cycles for the cells to drift to around 0.5% SOC witch is the natural delta between new cells and the only one that can correctly terminate the charging is the BMS.

The Victron MPPT 100/50 can be controlled ON/OFF by the SBMS0 and that is the only way to protect the cells from overcharge or undercharge. There is no fixed voltage you can set the MPPT that will work correctly.

14V is to high and the highest cell will be guaranteed to get to around 3.7 to 3.8V  and this will trigger secondary level high voltage lock that will turn off your loads also and overtime degrade your battery. Setting a lower voltage like 13.75 or below will result in your battery never getting to full charge and thus you will no longer know what the battery SOC is.

"*  I'm not sure I agree that resistance matters (in fact, I am not sure I agree with Dacian on his concern about fuses in the sense lines), as the current flow should be approximately zero. "

During cell balancing the current trough the sense lines will be somewhere around 120 to 150mA and depending on fuse used that can result in significant voltage drop (enough to affect the functionality).

If your fuses are 0.01Ohm then at 120mA there will be just a 1.2mV drop on the fuse and not affect the functionality in any way. Even if it is 0.1Ohm it will still be fine as the included ribbon cable has about 0.4Ohms

My comments where about 500mA or 1A fuses with very high resistance over 1Ohm in some cases and those will create problems. Quality low resistance 5 to 10A fuses will be fine and actually needed if you extend the cell monitoring wires not just use the included ribbon cable. 

[Dave McCampbell]

Harry, thanks for your comments.

 

      *I do the charge termination of the MPPT and the SBMS0 HVD opposite from you.  See my earlier Update post.  MPPTs control charge and turn it off at 14.0v charge termination. If needed SBMS0 can safely disconnect solar panels using Type 1 on DSSR 50s at 3.55vpc or .2v above MPPT charge termination. (We use the MPPT VE connections for our Node Red monitoring.)  This makes more sense to me than the opposite and still lets SBMS0 shut down charging if HVD needed.  Victron MPPT and other LFP chargers are purpose built to control charging so why not let them do their job and use the SBMS0 do its job when required?  It helps to have well balanced cells so that one cell does not reach HVD before the MPPT can reach charge termination.  Earlier with over 100mv delta in cell voltages near charge termination that was the case.  Our other two chargers, alternators and shore charger work the same with chargers doing their job controlling charging and SBMS0 using SSRs controlling HVD if needed.

2      *"balance as close as we can" with the tools we have is our goal and especially if it negatively affects MPPT charge termination.  With the NEEY balancer we have the tool to do it so why not?

3      *I don’t know enough about how resistance affects monitoring small MV voltages to comment on this.  However I can see that there is a noticeable MV difference between what the SBMS0 and NEEY/multimeter read.  So it is an interesting challenge for us to find out why as we have time.  I’ll defer to your and Dacian’s comments on this as we investigate.

 

Dacian, thanks for your comments.

 

You said: “At the end of charge above 100mV delta is normal and basically impossible to be less.”  I don’t understand this comment as my 4S3P bank is now showing less than 15mv at charge termination and without any balancing.  With the NEEY balancing I can bring that down to less than 2mv.  Maybe it is not even necessary to balance for quite a while with this well matched EVE LF280K bank.  And yes after charging has ended and the voltage gets lower the cell delta hovers around 8mv for the rest of the day with only the load on the bank.  I will add your SBMS0 balancer to the mix soon.

 

You said: “if you ignore SBMS0 and force charge above that then that 4th cell can get to 3.8V well above what will be normal for LiFePO4.”  I am not ignoring the SBMS0, but rather just using it to execute a HVD only if necessary, not as a daily charge termination control.  It is there to do its job as needed.  At no time do we force the charge above 3.55vpc as this is our HVD.  See details above in notes to Harry.

 

You said: “There is no fixed voltage you can set the MPPT that will work correctly.

14V is too high and the highest cell will be guaranteed to get to around 3.7 to 3.8V…”

14.0v is the most recommended voltage for LFP charge termination by others with, like you, years of LFP experience.  It is working well for us and many others now, gives us very close to full use of the bank’s full useable capacity, and still allows for a HVD at .2v/.05vpc above that if required.  And above that we have set a full charge buss disconnect using the SBMS0 at your default of 3.75vpc.  With this setup I see no reason not to do it this way rather using the SBMS0 terminating the charge at 3.55vpc and the MPPT at something above that and closer to the HVD Lock voltage. 

 

I have not extended the cell monitoring ribbon cable and there are no fuses in this cable wiring. However, the SBMS0 ribbon cable and NEEY/UNI-T multimeter are wired to different but similar cell bar points and using different sense wires.  How would you explain the roughly 8mv difference between what the SBMS0 and NEEY/UNI-T multimeter are reading?  Maybe it is just different wiring and connection resistance?

 

Thanks for the interesting discussion.  Dave

[Dacian Todea (electrodacus)]

Dave,

That Victron MPPT and any other charge controller on the market is designed for Lead Acid batteries. Those limits you set are for Lead Acid batteries.

Any type of Lithium battery requires a BMS and the BMS is the only one that can protect the battery thus it is responsible for stopping charging or discharging based on individual cell voltage with total battery voltage having no role or meaning.

For the past few days (just before new year) I started testing two of my old batteries (almost 14 years old) the GBS100Ah and A123Systems 20Ah. That GBS battery was my main house battery for about 4 years 2013 to 2017 and since then it was in storage with a short period of a few weeks when it was used to develop the DSSR20 in 2019

The GBS battery will go to recycling as is basically dead it had about 91Ah when new in 2012 then was tested in 2016 it had 74Ah and now in 2026 after many years of storage it tested around 42Ah. But this was always the worst LiFePO4.

In contrast the A123 Systems of same age was mostly in storage since new with occasional use maybe less than 50 cycles over time had 18.3Ah when new in 2012 and now in 2026 it has 17.4Ah so just a 5% degradation in about 14 years.

Here is the second charge cycle for the A123 system battery done at constant 3A (close to 0.2C charge rate).  This should give you an idea of what the cell voltage are for a quality LiFePO4 and this will be typical and close to best case scenario.

1)  Voltages 3:48:10 about half an hour after full discharge down to 2.8V (all standard SBMS0 settings) voltages recovered to 2.889V lowest cell and 3.060V highest cell  171mV delta.  (0% SOC)

 

2) Less than half an hour in to the charging so around 7% SOC  the cell voltages are much closer 3.236V low cell and 3.250V highest cell 14mV delta. 

3) About two and a half hours in to the charging SOC around 40% SOC cell voltages are super close 3.328V minim and 3.336V max thus just 8mV delta.

4) After 5 and a half hours is close to full charge around 92% SOC minimum cell voltage 3.375V max 3.383V so delta of 8mV 

5) Just moments before charging ended (100% SOC) highest cell 3.551V and lowest cell 3.481V  70mV delta  (keep in mind this is better than typical where you can easily see delta around double this 150mV just at the end of charge).

This battery is very well top balanced that is why there is just 70mV at the top and 171mV at fully discharged.

From this data I can know that delta in capacity between this 8 cells is around 2.5% and that was the case also 14 years ago and this are fairly good matched cells.

So cells are not equal in any aspect including capacity but also internal impedance and this internal impedance is why cell balancing is needed so that it can keep the cells from drifting over time.

But cell balancing will not be able to fix the delta between cells at the full charge and fully discharge ends as seen from this graph.

You can perfectly charge each cell to exactly same top voltage as I did for the very first capacity test but after a full discharge and then charge again there will be a drift that can not be fixed unless you want to reduce the current at the end of the charge and spend a lot more time at that 95 to 100% SOC slowly correcting the drift and you need to do this all the time wasting energy and degrading the cells.

This is particular for LiFePO4 as this has a super flat charge curve from 5% to 95% and it is not as much of an issue for NMC or other high energy density Lithium cells that that have a much more linear change in voltage with SOC.

C3 is the highest capacity cell and C8 is the lowest capacity cell in this particular battery.

[Dave McCampbell]

Hi Dacian,

Maybe my earlier wording “when required or when needed” is a bit confusing regarding the SBMS0 being able to execute a High Voltage Disconnect as a backup for a rogue cell going too high early or MPPT problems.  As described in detail in earlier posts on this thread, BOTH the Victron MPPTs and the SBMS0 are permanently installed and able to individually terminate charging.

We have our excellent Victron MPPTs set to execute a charging termination at 14.0v on a daily basis.  This is a commonly recommended setting for MPPTs.  Victron MPPTs are built for that purpose for ALL battery chemistries, including LFP, according to their literature.  This includes a Lithium setting (which we don’t use) and an Expert mode (which we do use) for fully adjustable custom parameter settings.  I agree that other MPPTs may not be suitable, especially if they have long nonadjustable absorb times for LA batteries.

In our setup if for some reason the MPPT does not stop the charge, or if a cell goes high early, the SBMS0 will execute a HVD at 3.55vpc thus acting as a backup to stop charging.  This HVD, and the High Voltage Lock disconnect at 3.75vpc, disconnects the charge source and gives us two backups to the MPPT failing.  Having this small voltage separation between these three charge termination set points, means that the balance must be fairly close.  This arrangement is common practice across much of the LFP professional installation industry.  It is especially important to have these backups in the marine world where the environment can cause more frequent equipment failures.

Regarding cell voltage balance, it is not difficult using today’s modern technology, to get the cell balance much closer than your 70mv at the top of the charge.  I agree, that if primarily using the BMS to terminate the charge, that is not necessary.  But if using the BMS as a backup to a MPPT, as in a marine installation is desired, the cell balance needs to be closer.  Over the course of less than a week’s charging our new installation, our 10a NEEY active balancer was able to bring our cell balance from over 100vpc where we started, to less than 5vpc.  And it has stayed there for the past week without further balancing.   So it is possible.

I fail to see how this use of both a MPPT and backup external relay BMS is not a valid setup, especially if using your excellent SBMS0 with its default settings controlling external relays or charging equipment directly at HVD.  In my view this is far superior to installing an internal Mosfet BMS, like Daly or JK that disconnects the battery if there is a cell problem.   This arrangement necessitates using multiple batteries and multiple other protections for installed equipment.  Multiple batteries means battery balance problems unless all resistances across the batteries, connections, and wiring are equal.  And then there is the issue of quality control and customer support with Asian equipment. 

Again thanks for the discussion.  Dave

[Dacian Todea (electrodacus)]

Any MPPT or PWM charge controller will work with Lithium batteries as long as they can be controlled ON/OFF by a BMS.

The default 14.4V is as good of a setting as any since you will likely never see the MPPT be able to terminate the charging even if set at 14V if the SBMS has ON/OFF control over the MPPT.

The real world example I just provided barely got to 14V when SBMS0 terminated the charging but if I where do do a few more cycles it will be significantly below 14V.

 That delta in capacity between those 8 cells of around 2.5% is close to best case scenario for matched cells and even if you start with perfectly balanced cells they will settle in time at around 0.5% to 1% balance at the top and remain there.

So the SBMS0 will almost always terminate the charging with very little chance that MPPT will be able to terminate the charging if MPPT is set at 14V and voltage drop on the MPPT cable is manageable.

If you set the voltage maybe lover say 13.9V and MPPT stops the charging that will not be good since the SBMS0 will not be able to correct for SOC errors and set the SOC to 100% at end of charge.

Yes the second level 3.75V will be able to terminate the charging by tripping OFF a breaker controlled by EXT IO6 set as type 5 but this will only happen if something fails like damaged MPPT.

The SBMS0 will not be able to keep LiFePO4 lower than 70mV at the end of charge in fact it will likely be above that. If it was NMC or LiCoO2 cells then yes it could keep those way below 20mV but those are very different chemistries.

To keep LiFePO4 at below 70mV when battery is close to fully charged will require that charge current is reduced way to much around 300mA so cell balancing is still active and low enough that cell balancing can keep them below 70mV. But this will require an extra charger that provide that 300mA or the ability of the charger to reduce the current at 300mA. But this ads extra complexity and there is absolutely nothing to gain and most likely detrimental to LiFePO4 (increased degradation).

A 10A cell balancer is just a bad idea and will not be able to do anything other than waste a lot of energy.  The cells are already below 10mV delta from at least 10% SOC to over 90% SOC

Those last 1 or 2% SOC when cells go from around 3.4V to 3.55V takes like 5 minutes if you look in my example and that is less than 0.2C so very typical for solar charge rates.

So it takes over 4 hours to get from 3.3V to 3.4V and then about 5 minutes to get from 3.4V to 3.55V

The 10A cell balancer will interfere with the SBMS0 balancer (unless you disable) and the 10A cell balancer will waste a ton of energy for nothing as I expect it has not enough information to know when to balance and when not to do so. Also as you likely will be charging at 50 ot 60A it will not be able to keep the cells below 70mV at the end of charge in that 98 to 100% SOC range that takes about 5 minutes.

So regarding your 10A cell balancer. If that does not know if battery is charged or discharged it will actually create imbalance then will try to correct the imbalance that it created.

Say you have just a 10% delta between cells in regards to internal resistance.

3 cells with 1mOhm and one cell is 1.1mOhm and you have a load of 100A (just a round number). Without any load cells may be perfectly equal say 5mV delta.

Then while the 100A load is connected the 3 equal cells voltage drop will be 1mOhm * 100A = 100mV and the other cell 1.1mOhm * 100A = 110mV and now you will have up to 15mV delta between cells. While SBMS0 knows this is a load not charging it will not enable the balancer but your external balancer not knowing this info will start to balance when what it is actually doing is creating imbalance (the opposite of what is needed).

Now when there will be a charge current it will need to recover what it did during discharge.

The 10A balancer is likely an active one maybe with as good as 70 or 80% efficiency so move charge from one cell to the other but there will still be significant loss due to balancing then balancing the battery.

That small amount of energy wasted on the SBMS0 internal balancer is less than 1kWh per year and all of that comes directly from solar panels (assuming solar is the charge source) and nothing from battery since balancing is only done during charging so the balanced cells just see a lower charge current as energy from solar panel is dissipated on the balancer.

[Dave McCampbell]

Hi Dacian,

I agree that if using the default HVD of 3.55vpc “the SBMS0 will almost always terminate the charging with very little chance that MPPT will be able to if MPPT is set at 14.0v”…  I would add and if the cell delta is relatively high at 70+mv.  Your real world example, and our earlier experience, of the MPPT set at 14.0v (equals 3.5vpc) and the SBMS0 default set at 3.55vpc is a perfect example of what will happen if the cell delta is relatively large.  One cell will go high early and reach HVD before the MPPT has a chance to finish its charge algorithm.  So the bank never gets fully charged to reach full usable capacity and the SOC suffers inaccuracy.

Relatively large deltas may be the norm for LFP batteries with only the SBMS0 balancer on, but not with quality EVE cells initially top balanced and actively balanced when needed with our NEEY.  For the past two weeks now our cell delta has been less than 15mv according to the SBMS0, mostly without any balancing.  The NEEY is not like earlier dumb Heltec and similar active balancers with no user control, adjustable presets and a wide cell delta required before it can be activated.  Also the NEEY does not need to be on all the time.  In the future I will only use it if there is a growing cell delta, and then only during the last few minutes of the charge.  I have adjusted its presets to turn on at 3.42vpc and off at 3.1vpc and if the cell delta is greater than 5mv according to the NEEY, not the SBMS0. 

You said: “The SBMS0 will not be able to keep LiFePO4 lower than 70mV at the end of charge; in fact it will likely be above that.”  And: “To keep LiFePO4 at below 70mV when battery is close to fully charged will require that charge current is reduced way too much, around 300ma…”  Now I understand why I was having so much trouble keeping my earlier RJ cells balanced with just the SBMS0 balancer.  That is precisely why I have the NEEY now and can use both the MPPT and SBMS0 to control my charging.  A couple of additional benefits to using Victron MPPTs are that I can adjust the amount of charge amps going to the bank, minimize shading effects by splitting up the array with multiple MPPTs, and monitor their performance on our Node Red display.

Based on the above information and because we want to have the Victron MPPTs control our charging with the SBMS0 as a backup, I will take your advice not to activate the SBMS0 balancer also.  Unlike other dumb balancers, because the user can adjust the NEEY presets, it knows where it is in the charge cycle and the user can make it react accordingly.  We are not concerned with the few additional amps that might be used up with NEEY for a few minutes at the top of the charge in order get the benefits of using the MPPTs as they are intended.  For anyone interested, you can simply Google ‘NEEY Active Balancer’ or’ Off-Grid Garage’ to learn more.  Modern technology strikes again.

Marine LFP and other equipment installations are all about backups, especially in remote locations with no Home Depots around.  We view having the SBMS0 back up the Victron MPPT as an ideal situation.  We also like that we can carry a relatively small SBMS0 BMS, various DSSR relays and other LFP system components as spares.  And since we have designed and built our own LFP system theoretically we should have the skills and tools to make repairs.  This is a much better situation than having to have some shore based electrician do our system for us with ‘drop in’ batteries, multiple Chinese internal Mosfet BMSs, and unknown fusing and connections, all without a schematic.

[Dacian Todea (electrodacus)]

Dave,

"One cell will go high early and reach HVD before the MPPT has a chance to finish its charge algorithm.  So the bank never gets fully charged to reach full usable capacity and the SOC suffers inaccuracy."

The battery bank actually gets fully charged and and the SOC will be accurate as it is properly reset at 100% real capacity.

SBMS0 is the one that needs to terminate charging and then it will know the battery is fully charged and set the SOC to 100%.

If the MPPT is the one that terminates the charge the SBMS0 will not know that charge was terminated and will not know to set the SOC at 100% thus SOC will drift over time and become inaccurate.

"Also the NEEY does not need to be on all the time.  In the future I will only use it if there is a growing cell delta, and then only during the last few minutes of the charge.  I have adjusted its presets to turn on at 3.42vpc and off at 3.1vpc and if the cell delta is greater than 5mv according to the NEEY, not the SBMS0."  

In order to be able to top balance the balancer will need to work only during battery charging and never during battery discharging. Since your external balancer does not have this info it will balance any time cell delta is above the threshold (say 5mV) and thus it will create imbalance as it will try to balance the cells while battery is discharging.

 

" A couple of additional benefits to using Victron MPPTs are that I can adjust the amount of charge amps going to the bank, minimize shading effects by splitting up the array with multiple MPPTs, and monitor their performance on our Node Red display."

Not sure why or when you will want to reduce the charge current as normally you will want to use all the available power from Solar. 

Not sure relative to what will you minimize shading ? As DSSR50 is more shade tolerant than Victron MPPT.    

Here I made a test with 6 types of shade comparing the performance of DSSR50 to Victron 100/50 MPPT  https://groups.google.com/g/electrodacus/c/cmLykQadmUk/m/dLPBRmcDBgAJ

"Marine LFP and other equipment installations are all about backups, especially in remote locations with no Home Depots around.  We view having the SBMS0 back up the Victron MPPT as an ideal situation.  We also like that we can carry a relatively small SBMS0 BMS, various DSSR relays and other LFP system components as spares.  And since we have designed and built our own LFP system theoretically we should have the skills and tools to make repairs.  This is a much better situation than having to have some shore based electrician do our system for us with ‘drop in’ batteries, multiple Chinese internal Mosfet BMSs, and unknown fusing and connections, all without a schematic."

The SBMS0 already has backup functionality trough the EXT IO6 set as type 5 witch then can trip a circuit breaker to isolate the battery or disable contractors to do the same thing.

If you understand what I was trying to say above it is that SBMS0 can not be the backup to Victron MPPT but the other way around.

If the MPPT terminates charging then SOC will be inaccurate and battery may not be fully charged. 

What you want will work with NMC cells but will not work with LFP. Below there is a rough graph showing the difference in voltage between two cells and comparing LFP to NMC.

The point of the graph is to show that LFP has a very flat voltage curve and so it is impossible to see the imbalance between cells other than for the last 5% or so at the end of charge where with NMC the difference will always be visible since there is an almost linear relation between voltage and SOC. That is the reason you can not have less than 70mV delta at the end of charge with LFP but you can have almost no delta with NMC.

Also 70mV delta for LFP at end of charge may be equivalent of just 0.5% SOC delta, where 70mV delta for NMC may be 15 to 20% SOC delta. The graph is made by Gemini so not that accurate is just to show the difference between NMC and LFP 

[Dave McCampbell]

Hi Dacian,

These will have to be my last comments on this as we still have much work to do on the boat before leaving for the Med soon.  The bottom line here for us is that our solar panel voltage is not the same as our system voltage, so use of a MPPT for us is required.  And because of that a better balance than your SBMS0 balancer can provide is necessary. Your comments in blue, mine beneath in black.

The battery bank actually gets fully charged and the SOC will be accurate as it is properly reset at 100% real capacity.
Yes real capacity, but not as full as it could be if better balanced with no cells going high way too early.  Without good balancing, we will some lose capacity.  And many users say that poor balancing does have some effect on the cells.  However, this is all a mute point as on our boat we have a 12v nominal battery system and a 24v nominal solar array.  So this means we need MPPTs to convert from 24v to 12v. 

In order to be able to top balance the balancer will need to work only during battery charging and never during battery discharging.

I agree with this thinking, although based on what I see on our cell graphs, the individual cell voltage positions remain the same at the top of the charge, both up and down, over 3.42vpc.  The new NEEY Gen 4 balancer has an even better level of parameter adjustment control than the SBMS0 balancer with adjustable start and stop voltages, voltage delta down to 1ma required to start, balancing amperage, and Bluetooth comms for monitoring and quick adjustments.  Additionally it uses up to 10a active balancing.  It does not turn on and off based on delta alone like the earlier common Heltec and similar cheap balancers.  It is relatively expensive and comes in several balance amperage models.

Not sure why or when you will want to reduce the charge current…
Cruisers often prefer to reduce solar output current when we are away from the boat and not using all the normally used equipment like refrigeration.  Again this issue of DSSR50 vs MPPT is mute for us because of the need for voltage conversion.  Shading may not be an issue for shore based installations but it sure is on a boat.  Also, unless I misunderstood your testing graphs, it looks to me like the MPPT is the clear but slight winner for raw output in your tests.

The SBMS0 already has backup functionality through the EXT IO6 set as type 5…  If the MPPT terminates charging then SOC will be inaccurate and battery may not be fully charged.
Based on what you said earlier, I think I do now fully understand your reasoning for using the SBMS0 for charge termination using its HVD feature rather than a MPPT.  As I understand it that is necessary because the SBMS0 200 ma passive balancer is not able to balance a large LFP bank closer than 70-100mv.  In that case only the SBMS0 will know when a cell goes high early risking overshooting safe cell voltages.

But for the many reasons I mentioned above, we and many other experienced users prefer to have the MPPT terminate its charge first with the two levels of SBMS0 HVD mentioned in the charging system as backup.  If closely balanced, the MPPT, using its adaptive charging feature, will allow the battery to reach full capacity at 14.0v which is only a very few ahrs short of SBMS0 14.2v/3.55vpc full capacity.  In order to reset the SOC, it is an easy job to once in a while let the SBMS0 perform a HVD.  And again, we need to have the MPPT active in the system, as there is a mismatch between our system and panel voltage which helps our MPPT perform its function.

The point of the graph is to show that LFP has a very flat voltage curve and so it is impossible to see the imbalance between cells other than for the last 5%.
I agree that the LFP voltage curve is very flat until the last 5% of charge.  But with our Node Red display, see below, we can easily see the voltage imbalance between our cells, and it is very consistent until the last 5% of the charge.  The last 5% is the only place we are balancing with the NEEY.  In fact it is only above 3.42vpc, because that is where we can see that the cell voltages get to their final end of charge positions.  Starting balancing before that voltage point would be way too early and detrimental for our system.  Others experienced users I know with similar systems do the same.

So enough for me on this now.  We will continue to gain experience with our current system as is and see how it goes.  Hopefully with the much better matched and compressed LF280K EVE cells, new better cell connector bars, Victron SmartSolar MPPTs, the NEEY balancer, now fully complying with ABYC standards things will go well.  More later if it does not.  Dave SV Soggy Paws

[Dacian Todea (electrodacus)]

Your comments in Blue.

Yes real capacity, but not as full as it could be if better balanced with no cells going high way too early.  Without good balancing, we will some lose capacity.  And many users say that poor balancing does have some effect on the cells.  However, this is all a mute point as on our boat we have a 12v nominal battery system and a 24v nominal solar array.  So this means we need MPPTs to convert from 24v to 12v. 

As mentioned 0.5% SOC due to no perfect balancing so not significant loss of capacity. Also no effect of cells. Forcing perfect cell balancing means spending more time at the end of charge and thus higher degradation.

Yes you need MPPT in your situation but this has nothing to do with cell balancing. The MPPT can be controlled ON/OFF same way that DSSR50 can.

new NEEY Gen 4 balancer has an even better level of parameter adjustment control than the SBMS0 balancer with adjustable start and stop voltages, voltage delta down to 1ma required to start, balancing amperage, and Bluetooth comms for monitoring and quick adjustments.

The SBMS0 has all those adjustments that you mentioned. But the most important one is missing from the NEEY and that is knowing if battery is being charged or discharged. If balance happens during discharge it will be counterproductive.

Also, unless I misunderstood your testing graphs, it looks to me like the MPPT is the clear but slight winner for raw output in your tests.

Yes MPPT is a net winner but it wins in average by 3% so not something that will ever make a difference. But in regards to shading DSSR will do better. Still not important for you as you need to use an MPPT due to using 12V battery.  I just mentioned that shading is not something that MPPT is better over DSSR50.

But for the many reasons I mentioned above, we and many other experienced users prefer to have the MPPT terminate its charge first with the two levels of SBMS0 HVD mentioned in the charging system as backup.  If closely balanced, the MPPT, using its adaptive charging feature, will allow the battery to reach full capacity at 14.0v which is only a very few ahrs short of SBMS0 14.2v/3.55vpc full capacity.  In order to reset the SOC, it is an easy job to once in a while let the SBMS0 perform a HVD.  And again, we need to have the MPPT active in the system, as there is a mismatch between our system and panel voltage which helps our MPPT perform its function.

If current shunt is very well calibrated and the battery capacity is correctly set you will still get a 2 to 3% SOC error accumulated by the SBMS0 so it can exceed 10% SOC error in a full month without the SBMS0 getting to HVD.

As long as all cells stay below 3.6V and charge terminates as it seems it does it should be no problem. When will the MPPT start charging again ? I only see the first charge just after noon. I will be curious to see when charging is re-enabled for the rest of the day. If it spends the rest of the day charging multiple times it may be detrimental long term for the cells. 

 

I'm collecting now data from my system (one of them) and when I have a full year of data I will probably make a video showing what is produced by a 9kW array in each month of the year and maybe get in to the details of how the battery balancing is maintained around 0.5% SOC for that year. I will also do a capacity test to see the capacity loss for that year. I think I started around summer 2025 logging everything at 1 second interval so should have all the data this summer. The winter is the more interesting part as I heat the house and use much more energy than I do in summer. Today I got about 40kWh from that PV array most of that for heating. 

It is exactly -25C outside now and +25C inside so 50C delta.