Good thesis on preserving the life of LiFePo4 lithium batteries

[rgleason]

I see that Catnewbee posted a dual engine / alternator wiring diagram on the next page and I have saved that image for reference.


The REC-ABMS manual has a good simple diagram I like which has two buses, a charge bus and a load bus each with an ML-RBS switch this would allow separate control of discharge and charging and might simplify things, but the BMS has to be able to control two.

[CatNewBee]

Well, this is an almost complete modular solution, that I have built and configured to my needs.

It boils down to the folowing building blocks:

- programmable BMS that provides
..... static signals for OVP / LVP
..... aux signals that can be used at different set-points
..... cell voltage monitoring
..... current monitoring, SOC, SOH monitoring
..... cell temperature and BMS temperature monitoring
..... cell balancing
- Logic that translates signals to impulses
- bi-stable solenoids for high current switching
- logic that copes with legacy sources (BMV712)

I chose the REC product for this reasons:
- active balancing
- completely configurable set points
- on request 2 independent sets of signals for OVP /LVP
- self-resetting error conditions, self recovery from failures
- one single encapsulated unit, no cell modules
- takes current and cell resistance into account
- galvanicaly isolated interfaces
- display of all relevant informations SOC, SOH, time remaining, pack voltage, current, Ah drawn, cell voltages, cell resistance, cell temp, BMS temp, max/min cell voltage indicating whitch cells will be balanced, cell cycles...

It frankly contains all I needed, no additional circuitry necessary, keeps it simple.

With other systems you often need besides the cell modules some sort of monitoring / display to watch the cells or evaluate possible alarms/errors. You would need something that raises alarms before the BMS disconnects the pack. You probably need something that overrides the simple BMS based on temperature, the current usually is not considered by simple BMS modules.

As I said, this are building blocks. You are free to chose the level of protection.

The legacy sources charge relay controlled by the BMV acts completly independent of the BMS, but it is connected in series with the BMS OVP cut off, so the BMS settings protect additionally from over charging.

The legacy sources (shore charger, alternators via FET charge distributions...) are switched on or off by SOC only, estimated by the BMV712. The BMV is programmed to be synched when the battery reaches its pre-set full capacity.

What you intent is to keep the LFP at lower SOC to prevent balancing, causing the cells to drift slowly over time, without reaching a full state your BMV will never sync, so your SOC reading will also drift and you will mess up your battery.

There is a Victron BMS out there, but it is meant for Victron LFP battery packs with integrated cell logic / balancers. The so called BMS is only an interface with (weak) solenoids to turn on and of the busses. The signals are created inside the battery and thetefore this is meaningless for DIY batteries from bare cells. The cell logic is not sold separately and is also not configurable, but factory pre-set.

[CatNewBee]

You can of course built a smaller unit from cells and a simple BMS like the daisy chain LiPro modules from ECS, it will cost around 160€... 360€ ( modules are 40€ ... 90€ each depending on the chosen size and function (https://ecs-online.org), or you go for the 123BMS with bluetooth interface (https://123electric.eu) for 360€.
There are other BMS manufacturer around to chose from.

Finally you still need the logic to switch the currents (solenoids) and translate the signal outputs of the BMS to the input requirements of the relays (coil current, static vs. impulses, switch on high or switch on low...).

Btw, the Victron BMV is NOT a BMS. It is a high-level "fuel gauge" for a battery pack. If this is the only protection you have, your battery is litteraly just bare cells with no protection.

[Maine Sail]

Originally Posted by CatNewBee

Btw, the Victron BMV is NOT a BMS. It is a high-level "fuel gauge" for a battery pack. If this is the only protection you have, your battery is litteraly just bare cells with no protection.

Spot on.. On top of that it will drift out of calibration rather quickly. I have over 9 months of testing Ah counters against LFP and none of them stay accurate for very long and require manual resets when the bank is known to be fully charged.

The only SOC gauge I've tested, that works with LFP and remains quite accurate, is the new Balmar SG-200.

Unfortunately it is not yet in full production mode and is in a Beta launch stage. As soon as I can get my hands on a production model, not my Alpha testing unit, it will be replacing my Link-Pro in about a nano second.

There is a big difference between monitoring pack voltage and cell voltage. A Victron BMV can only see pack voltage. The below numbers were straight off a boat where the owner was tracking only pack level and following the "manufacturer max safe voltage" of 3.65VPC for charge.

By the time he contacted us he'd ruined two cells. In my experience it is getting more difficult to find well matched cells these days so cell level monitoring is even more important now than ever. As was stated a BMV (Ah counter) is not a BMS....

Pack level monitoring can be misleading:

14.6V = 3.65V + 3.65V + 3.65V + 3.65V

14.6V = 4.2V + 3.6V + 3.45 + 3.35V

[CatNewBee]

Hallo Maine Sail,

I got my BMV to synch each time my cells get to near full by setting the BMV 100% to 14.0V when I am using the LiFeYPO4 daily, while the solar charger is set to Absorption at 14.3V for the shortest possible setting and Float to 13.5V to run all gear on solar during the rest of the day if excess energy is available.

So the BMV syncs at approx. at 95% SOC to "full" and remains there until the battery is fully charged and then starts discharging, The BMV does not go to 105%, just stays at full and once current is drawn from the pack it starts the count down. The only disadvantage is, the BMV readings above 95% while charging are non-existent, it jumps to full.

That is the easiest way to ensure as many syncs as possible with an error of less than 5%. The battery usually gets at least once per weak to 100% during the summer time and not so nice weather, so the drift is minimal overall.

It is possible to play around with the Peukert factor to tweak it further too.

The BMS's built in SOC calculation seams to be much more accurate, it syncs at the set values, but it consideres the cell resistance, temperature and the charge current together with the cell voltage, the drift between the two coulomb counter remains in an acceptable range, so the BMV is good enough to turn on and off non-LFP charge sources at defined SOC set points or to raise audible alarms before the shoulders are reached.

It is a nice supplement to a BMS.

[Maine Sail]

Originally Posted by CatNewBee

Hallo Maine Sail,
I got my BMV to synch each time my cells get to near full by setting the BMV 100% to 14.0V when I am using the LiFeYPO4 daily, while the solar charger is set to Absorption at 14.3V for the shortest possible setting and Float to 13.5V to run all gear on solar during the rest of the day if excess energy is available.

So the BMV syncs at approx. at 95% SOC to "full" and remains there until the battery is fully charged and then starts discharging, The BMV does not go to 105%, just stays at full and once current is drawn from the pack it starts the count down. The only disadvantage is, the BMV readings above 95% while charging are non-existent, it jumps to full.

That is the easiest way to ensure as many syncs as possible with an error of less than 5%. The battery usually gets at least once per weak to 100% during the summer time and not so nice weather, so the drift is minimal overall.

It is possible to play around with the Peukert factor to tweak it further too.

The BMS's built in SOC calculation seams to be much more accurate, it syncs at the set values, but it consideres the cell resistance, temperature and the charge current together with the cell voltage, the drift between the two coulomb counter remains in an acceptable range, so the BMV is good enough to turn on and off non-LFP charge sources at defined SOC set points or to raise audible alarms before the shoulders are reached.

It is a nice supplement to a BMS.

It all depends how you use your bank and 5% off is still a bit much for what I would like to see. For our bank we have no need nor desire to hit 100% SOC, but we can pretty easily. Unfortunately we are forced to do this just to keep the Ah counter accurate.

We would prefer to simply put in what we need for energy then stop charging and discharge to 70% or 80% DOD again. Getting to 100% SOC with LFP is something we would never need to do, that's the beauty of it, but with Ah counters we do need to do this in order to sync. If you routinely get to 100% SOC an Ah counter can be decent for 5-7 days but, when you continually PSOC cycle the bank, it can get out of sync pretty fast.

For us, just getting the boat ready while on the mooring, while chilling the fridge (engine driven and 12V), and then motoring off the mooring gives us nearly a day or slightly more worth of energy. If we are already starting from 30% to 50% SOC that means we have multiple days of energy by the time we shut the engine down.

We then have the option of flipping on the PV in order to have even more time before charging again. The only reason we ever hit 100% is due to the Ah counter and drift. For us an accurate SOC meter will be a blessing and will no longer require us to ever hit 100% SOC, just put in what we need to for the planned use. Near continual PSOC cycling is what LFP is all about, to us, and the only thing keeping us from that right now is the darn Ah counter.

[rgleason]

Thank you CatNewBee and MaineSail.

The bad cell info is very informative.


It appears I did not understand the intent of the BMV712 LVD and HVD relay controls. Can you please explain why these are not appropriate for a LiFePo4 system and what they were actually intended for? As I understand it, the BMV712 with count amps and use that and voltage as the low cutoff, similar for the HVD. So why isn't that acceptable?

Also I have been confused by the different approaches taken by DIYers, and what these users do and say. I am trying to figure out a good system to use the LiFePo for a small boat. I think these different approaches are:

• Level 1 - Totally naked, manual use of the batteries, including manual top or bottom balancing.
• Level 2 - BMS and LVD and HVD protection, no balancing, with charge source regulation limit parameters set within the LiFePo4 constraints and some manual attention to limiting depletion below acceptable levels.
• Level 3 - BMS with LVD and HVD protection, no balancing, but a more automatic system operating relays for charge and load with a start battery for alternator charging.
• Level 4 - BMS with LVD and HVD protection, balancing and an automatic system for operating relays for charge and load with a start battery for alternator charging.
• Level 5 -Full BMS with balancing, automatic controls of charge and loads, perhaps disconnecting the alternator field wire.
• other variations.

I have looked at the 123BMS and the BMS from electriccarparts.com and several others, including the inexpensive system of boards that get attached to the terminals (which no matter how clever, seem problematical to me). I am very interested in finding a good one that will work well. It may be that the REC ABMS active is the right one. I just do not know enough... Entropy has a great website with great diagrams that I think I can understand but his BMS is defunct .


I am leaning towards a 270ah or 300ah Winston bank (3p4s) of 90ah or 100ah cells. I think the 100ah cells are too high for under my cabin seat. The cable runs are pretty long from the engine (72") and 40" up to the switch, but the problem is working out the system, basically with just an new 150amp alternator with serpentine belt as charger and replacing the ARS5 with MC614. We seldom have shore power, and will have a separate small reserve/starter FLA battery charged by two small solar panel(s) on the dodger.


That is about as far as I have gotten I guess.

[Delfin]

Originally Posted by Maine Sail

It all depends how you use your bank and 5% off is still a bit much for what I would like to see. For our bank we have no need nor desire to hit 100% SOC, but we can pretty easily. Unfortunately we are forced to do this just to keep the Ah counter accurate.

We would prefer to simply put in what we need for energy then stop charging and discharge to 70% or 80% DOD again. Getting to 100% SOC with LFP is something we would never need to do, that's the beauty of it, but with Ah counters we do need to do this in order to sync. If you routinely get to 100% SOC an Ah counter can be decent for 5-7 days but, when you continually PSOC cycle the bank, it can get out of sync pretty fast.

For us, just getting the boat ready while on the mooring, while chilling the fridge (engine driven and 12V), and then motoring off the mooring gives us nearly a day or slightly more worth of energy. If we are already starting from 30% to 50% SOC that means we have multiple days of energy by the time we shut the engine down.

We then have the option of flipping on the PV in order to have even more time before charging again. The only reason we ever hit 100% is due to the Ah counter and drift. For us an accurate SOC meter will be a blessing and will no longer require us to ever hit 100% SOC, just put in what we need to for the planned use. Near continual PSOC cycling is what LFP is all about, to us, and the only thing keeping us from that right now is the darn Ah counter.

What a great discussion - thanks to everyone with actual experience with this technology and taking the time to share that experience.


MaineSail, I installed a NC push button switch that resets the Link 20 by depowering it, so that after charging the LFP bank to a CAR of 5%, lets me count down Ah usage from a full battery. I guess I don't understand why you say you need to go to 100% to reset your Ah counter, or am I confused, which is generally a safe working assumption.....

Also, I'm relying on your extensive bench testing, which shows that cells don't drift if you keep charge voltages at or below 14 as measured on the pack and disconnect when the CAR drops, but it sounds like CatNewBe has a different perspective, so if you two could strip down to loincloths and settle that issue, I'd be obliged...

[CatNewBee]

OK rgleason, I try to explain it that way: The single point of truths is the LFP cell.

You want every single cell to be charged and discharged to a safe level

Winston LiFeYPO4 are specified as following:

4.00V initial charge, beyond that - cell destruction
3.65V cell is 100% full (14.6V)
2.80V cell is considered empty (11.2V), below this - cell destruction.

Usual agreement here is, this 100% - 0% settings are not very healthy for a long life beyond 3000 cycles. There is no known memory effect, nor is there a need to charge the cells to 100% frequently - in contrast to Lead Acid batteries.

The cells do have a very shallow curve between empty and full only on the corners there are steep run-offs, and this is a big problem for battery packs.
Why is this?

Well if one of the cells accepts slightly less energy than the others (less capacity), it will be charged faster and will reach the upper shoulder, the other cells are still not full and remain at a comparable low voltage. When the pack voltage hits the full threshold then (14.6V set by the charger for 4 x 3.65 all cells full - charger do not care about cells, they care about batteries), the 3 other cells will be at lower voltages, while the one cell will run off over the 4V limit and will boil.

Same for discharge. Some cells have plenty of energy and stay above 3V, the weak cell goes down below 2.8V, Battery voltage seems OK, but the cell gets damaged, it blows and heats up.

That is the reason you want balanced cells (to allow the charger operate safely at the pack based voltage cut-off) and you want cell voltage monitoring (to disconnect if something really goes wrong with the balancing). This is what BMS'es does more or less sophisticated.

The BMV is only a device, that works like the chargers, it monitors the pack voltage and current, it has no clue about an individual cell.

Now to the settings:

A safer setting would be to stop charging at 3.60 or even 3.55V (14.4V / 14.2V) and stop discharging at 2.9V (11.6V) This voltages are valid at 20°C and would deliver you 80%+ USABLE Ah of the nominal capacity.

Victrons LFP pre-set is 14.5V for stop charging and 13.5V float (3.375V)- where they consider no more energy flow into a cell and a save voltage. It will keep the Battery at 95% SOC approximately, so you have almost the full capacity available when charging has finished and you need the power.

Some argue, this settings are too high, and it would be better to cycle the cells instead of holding them at this level - or choosing a lower voltage for float. But this is another discussion about the meaning of life and the sense of a battery. Some want a longer life of the pack and commit to use only part of the available capacity, others opt for smaller batteries and shorter life, but use the full range. It is hard to decide who will be better off in 10 years.

Now to the BMS - the main meaning of this is, as stated before, to balance the cells, so no one runs off in contrast to the others to ensure all (legacy) charger and protection logic based on battery pack voltage thresholds works properly.

Second The BMS fires two independent signals (that can be joined for some purposes if you do not have separate charge / discharge buses - e.g when paralleling independent batteries with own BMS'es) - this are LVP (low voltage protection) and OVP (over voltage protection), some have different names for them, but all have the same meaning. If any one of the cells goes out of range, the according signal will be set. You then CAN and SHOULD disconnect either the chargers or the loads. This can be done grecefull by a signal to turn off the device or by brute force by disconnecting with a power solenoid. This is considered a last resort of protection - like a fuse that blows. This is not meant to control the charging sources / loads in regular operations, the settings of them must be lower than the OVP settings. IF you set OVP to 3.65V (14.6V Battery pack voltage on balanced cells), you should program your chargers to lets say 14.4V maximum, so OVP does not occur as long as the cells are almost in balance. I guess, you got the point.

Early warnings can be fired by a normal battery monitor so - like the BMV, it is not necessary to do it at cell level, the BMS takes care of the cells alone and balances if necessary. You also can use such devises also to early drop high loads or continue to use legacy chargers, that cannot be set that low to your desired maximum voltage - like alternators, old AGM charger, wind generators etc.. you name it.

Some argue balancing is bad for the cells - what is not true at all. Passive top balancing is usually invoked from a cell voltage higher than 3.55V/3.6V (that is the case when cells are charged and a current is flowing, full cells in rest after some hours are at 3.4...3.5V), so no energy is drawn from the cell, but approx 1A - a small amount of the charging current is burned by a parallel resistor. So this specific cell gets less charge than the others. Emptier cell are charged at full current. At some point the second cell hits the threshold and starts deviating energy to a resistor, than the third and than the forth - usually then Absorption of the charger finishes. Small imbalances are equalized over time with each cycle.

This type Balancer do not discharge cells - they are inactive on lower voltages where the battery delivers power, they do not add up on unnecessary cycling - but they produce heat and waste some charge energy. Active balancer do not heat the compartment but transform the energy and recuperate it to the pack, but they are more complex and more expensive. If you never charge the battery above 90%, balancing will never occur except one cell runs off.

All other level you mentioned have nothing to do with a BMS.

Your boat electric system is a meshed environment of "smart" and dumb independent devices. Some protect the battery (inverter have a low voltage cut-off), deep discharge battery protection devices - some have configurable or pre-set voltage cut-off points (Solar controller, shore charger), some just cannot deliver higher voltages (Alternator) and regulate by burning energy (Wind Gen), some simply do not care - (regular loads).

In normal operations all are safe to use, if they make the battery to run away, the BMS will intervene.

You can add at any time more smart devices for different tasks - to raise warnings or do smart things - like heat the water boiler or make water with excess energy.
One handy thing that can do this things based on SOC levels is the BMV. Another way is to use a remote switchboard and the master shunt from Mastervolt, and program them accordingly - or use the Victron GX - it is linux based and can read the SOC from the BMV and then create CAN bus messages to control remote actuators for more complex tasks...

[rgleason]

CatNewbee, Delfin and MaineSail I have bookmarked, copied parts to my worksheet and will be re-reading this many times. Thank you! CatNewbee that is a a great big picture.


I need to adjust this list

[nebster]

Originally Posted by Maine Sail

Spot on.. On top of that it will drift out of calibration rather quickly. I have over 9 months of testing Ah counters against LFP and none of them stay accurate for very long and require manual resets when the bank is known to be fully charged.

Do you have any more data you can share from that testing? Like, what kind of PSOC cycling you performed, and how long it took them to get off track, and how you determined that they were no longer accurate?

I'm interested because, although there is definitely drift, I have not found it to be enough to cause the problems you talk about in your subsequent post. (I will quote those in a moment and ask about them as well.)

The only SOC gauge I've tested, that works with LFP and remains quite accurate, is the new Balmar SG-200.

It would be interesting to learn more about how they achieve better accuracy. I could only find some hand-waving and marketing material in a quick search.

I am also interested in what "quite" means and how you evaluated it.

[nebster]

Originally Posted by Maine Sail

If you routinely get to 100% SOC an Ah counter can be decent for 5-7 days but, when you continually PSOC cycle the bank, it can get out of sync pretty fast.

I'm interested in talking about this more, since it applies in my case as well.

How many "cycles" in PSOC are we talking about? How far out of sync does it get for you? (How do you know how far out of sync it is, when you conclude that it is?)

When you talk about all of the counters being inaccurate, are you adjusting their model inputs over time? For example, on Victron BMV, do you tweak the charge efficiency and peukert exponent? Those are our two levers into the mathematical model inside that specific device, for example.

Do you use the same shunt for all the ones you've tested?

Btw, I don't mean to hammer you specifically with questions, and really we could start a whole separate thread on this topic to get broader input.

But my experience has been that current-based SOC estimation is still pretty good after 10 days of PSOC cycling. Certainly good enough to accommodate decision-making along the lines of, "hmm, we're at 40% right now; let's charge for an hour and bring the pack up to 65% for the day."

I'd like to understand what's going on that's making your experience so different.

[CatNewBee]

The SOC in pecent is something I watch and also log daily, because I am curious, but more important during the day to me are the readings of the Ah used / to be re-charged and the current at the moment.

I also notice not too big drifts between the BMS findings and the BMV readings. I do allow the batteries to recharge to full and the counters are synched frequently too, so I do not have figures about long time PSOC - cycling - it would require manual intervention to stop charging and calculations of future power needs combined with weather / sunshine estimations and vessel orientation to the sun to predict how much will be needed.

This is simply too much worrying. I prefer a set-and-forget system, that delivers power at any possible arising demand - worry free. A full battery is very much appreciated, excess energy is harvested preferrably in fresh water (watermaker) and hot water (boiler).

Regarding accuracy - I simply compare the readings of the independent systems (they share the same shunt, but calculate their own values), and watch from time to time the counters near the synch point at the next full charge to estimate the drift - between 1..3%.

In my use pattern with at least weekly sync - negligible, but if you never synch this drift can add up. Worst case 12% per month, 25% in 2 months 36% in 3 month etc...

To figure it out, set up 2 BMV in parallel with same parameters for peukert and efficiency, but set the one to sync at Absorption and the other to a higher voltage to prevent any sync. Then make a manual sync at 100% for the second counter to the sync point of the first. Let them run for several month and watch the drifting values.

The first will be your reference, the second will show you the drift per cycle.

You can play around with the efficiency and peukert to calibrate the second BMV to your battery and reduce the drift delta.

[nebster]

Originally Posted by CatNewBee

I also notice not too big drifts between the BMS findings and the BMV readings. I do allow the batteries to recharge to full and the counters are synched frequently too, so I do not have figures about long time PSOC - cycling - it would require manual intervention to stop charging and calculations of future power needs combined with weather / sunshine estimations and vessel orientation to the sun to predict how much will be needed.

Right, I have the same problem, which is why I wonder how MaineSail is getting good data on his boat. (Maybe I am misunderstanding and he is not on his boat, except he talks about some scenarios that seem to include his production battery.) I understand being able to measure on the bench in the laboratory, but underway I don't have an easy way to stop everything and directly obtain a calibration measurement.

This is simply too much worrying. I prefer a set-and-forget system, that delivers power at any possible arising demand - worry free. A full battery is very much appreciated, excess energy is harvested preferrably in fresh water (watermaker) and hot water (boiler).

I feel similarly, but mostly because I am comfortable setting my charge stopping point at a fairly low true SOC.

I first calibrate my battery monitor to read "100%"/sync when the charge strategy yields about 92% true SOC. Then, using that monitor as the input, I set my generator autostart for 20% and autostop for 90%. 90% of 92% is about 83%, so the effective stopping point when we are off grid is about 83% SOC.

My observation is that the system is very stable at repeatedly reaching roughly that true SOC over a period of at least 10 or 12 days. I mean, it might be at 81%, or 85%, but it's definitely not at 75% or 90% true SOC.

To determine where it *really* is would require me to immediately move onto shore power and isolate the pack, let the pack come to resting Voc, and then measure. We do have chances to do that sometimes, so depending on what I learn in this thread, I might make a note to try to determine where we "end" after an extended set of partial cycling.

To figure it out, set up 2 BMV in parallel with same parameters for peukert and efficiency, but set the one to sync at Absorption and the other to a higher voltage to prevent any sync. Then make a manual sync at 100% for the second counter to the sync point of the first. Let them run for several month and watch the drifting values.

The first will be your reference, the second will show you the drift per cycle.

It sounds like a good idea, especially since I already have two BMVs sharing the same shunt! I'll have to try this. Casually, I have never noticed the two BMVs off by more than 2% from each other, but right now they have identical configurations.

It seems like that would presume that the drift is static rather than varying, which I am not sure is true. Intuitively, I would expect the drift to behave like a stochastic process.

You can play around with the efficiency and peukert to calibrate the second BMV to your battery and reduce the drift delta.

My un-expert but nevertheless real world finding is that I get pretty close by first setting peukert to 1.00 (no compensation), and then use "charge efficiency" to compensate for the ohmic losses. I believe my efficiency is set to 0.98 right now.

It is possible that this simplification may only be viable for healthy cells in a relatively young state, or it may be that I have unusually good cells.

State estimation is a really interesting problem! I'm keen to learn what Balmar has done -- a machine-learning model that fits observed voltage data over time against historical data might be one way to improve on the simpler, point-in-time estimation technique that everyone else uses.

[john61ct]

I charge **to** 3.45Vpc for daily use, no Absorb.

If a "calibration 100%" is required, I have defined that as 3.45Vpc then holding that V until trailing amps taper to .03C

That is a precise objective point to reset the BM to 100% Full and should be easy to do while the rig is in use. Consistent enough even with high or low current sources.

Certainly no need to break up the bank / packs.