LiFePO4 Batteries: Discussion Thread for Those Using Them as House Banks

[Maine Sail]

Originally Posted by Jd1

Should you loose power to the meter, although it keeps history data, it does NOT keep your current SOC data. Instead, you are greated with a flashing '-----' for everything that relies on SOC data. At this point your only choice is to manually re-sync the display which brings the SOC to 100% no matter what the actual state is. To get the meter to read accurately, you have to drive the battery SOC to your reset point or to where your real 100% point is.

In an installed system this should never be an issue... If you are working on the bank charge to full first or, using the batt SW parallel in the reserve and use a jumper to maintain the SOC setting until you reconnect.

As I have said many times before a Coulomb counter is only as accurate as it is programmed and used. You can not pay enought me to utilize the auto-synch feature on any of them...

Unfortunately a Coulomb counter is currently the only good way of tracking LFP SOC and even on LFP they will drift over time....

[T1 Terry]

Originally Posted by Maine Sail

In an installed system this should never be an issue... If you are working on the bank charge to full first or, using the batt SW parallel in the reserve and use a jumper to maintain the SOC setting until you reconnect.

As I have said many times before a Coulomb counter is only as accurate as it is programmed and used. You can not pay enought me to utilize the auto-synch feature on any of them...

Unfortunately a Coulomb counter is currently the only good way of tracking LFP SOC and even on LFP they will drift over time....

I think if you are realistic about the capacity, check and reset yrly, set the auto reset to the equivalent of 99% full, set the Peukert's factor to 1 and the charge efficiency to 100%, the Victron doesn't stray very far from a realistic reading. Each time the battery voltage climbs above 13.9v, the current drops to around 1 amp to 5 amps depending on capacity, but say 1% CA, and it maintains that for 1 min then you can be fairly confident you have reached or exceeded to 99% SOC point.
A high voltage set point of say 14v will achieve the above figures if you have a decent regulator that uses a PWM control of 0.5v or better. The only time a reset will fail to occur is if you have a run away cell and the HVC acts to protect a cell before the PWM control acts to limit terminal voltage.

T1 Terry

[diugo]

Originally Posted by OceanSeaSpray

Remember that if you are using resistive voltage dividers against ground, having equal sensing currents in the sensing wires still doesn't result in balanced drain on the cells!
This is because the first cell is still contributing current to make up the sensing voltage of the others. It takes more than 8 resistors to make a balanced sensing circuit.

Thanks for pointing that out.

Just to clarify, you mean more than 8 real-world resistors---but only to improve precision. In other words, the math may call for 2014.3 ohms, which is unlikely to be found in a single resistor; yet I might get away with a single 2K 1% resistor that's "close enough"...

[OceanSeaSpray]

Originally Posted by diugo

Thanks for pointing that out.

Just to clarify, you mean more than 8 real-world resistors---but only to improve precision. In other words, the math may call for 2014.3 ohms, which is unlikely to be found in a single resistor; yet I might get away with a single 2K 1% resistor that's "close enough"...

No, I really mean more than 8 resistors regardless of how you achieve the required values. It takes 11 resistors to construct a balanced sensing network. You need to write the equations for current and voltages for all 4 cells and all branches of the circuit. I will only write the node current equations here to illustrate the problem:

Let's call Cell 1 the one against ground and Cell 4 is the one at the top. Similarly, we can call the current out of Cell 1 i1, and the sensing current for the Cell 1 voltage is1. Same for cells 2, 3 and 4.
Now Cell 1 is supplying not only is1, but also is2, is3 and is4 because all of those are sensed against ground and therefore they all have to flow through Cell 1. We have:

i1 = is1 + is2 + is3 + is4
i2 = is2 + is3 + is4
i3 = is3 + is4
i4 = is4

As you can see, no matter what we do, Cell 4 will have a lower current than all the others and Cell 1 will carry the heaviest load.
Unless you artificially increase the current drain on cells 2, 3 and 4 without going to ground in order to match what cell 1 is supplying, the system won't be balanced. This requires introducing 3 more parasitic resistors to achieve i1 = i2 = i3 = i4. On the other hand, whether is1 = is2 etc doesn't actually matter at all.

A neat consequence of such a sensing circuit is that it actually promotes auto-equalisation between the cells in the sense that if one cell voltage is a little higher than the others, it will also draw slightly more current from that cell compared to the others until they all reach the same voltage again! In practice the current differences are too small to be really effective, but the point is that it favours convergence over time, not divergence and as such it is technically better than "no sensing".

Best regards,

Eric

[diugo]

Originally Posted by OceanSeaSpray

No, I really mean more than 8 resistors regardless of how you achieve the required values. It takes 11 resistors to construct a balanced sensing network... This requires introducing 3 more parasitic resistors

Thank you Eric for taking the time to write out the node equations. So basically I just put resistors in parallel to cells 2, 3, and 4 and then use Kirchoff's Current Law at the six nodes to get all the currents.

I still contend these tiny currents are negligible enough to be ignored---kinda like scramming the reactor at Palo Verde by turning on a desk lamp---but it is a piece of cake to change out four of the eight existing sense resistors and add the three parasitics. So I will do so, if just for the elegance.

If the cells remain perfectly balanced a year from now, I will eat my hat

[OceanSeaSpray]

Originally Posted by diugo

...
I still contend these tiny currents are negligible enough to be ignored---kinda like scramming the reactor at Palo Verde by turning on a desk lamp
...
If the cells remain perfectly balanced a year from now, I will eat my hat

It depends how tiny! 1000 hours is only 42 days when you keep at it continuously, and even at 1mA only, there goes 1Ah already. You would need to sense with microamps literally to be able to ignore these currents.

Now they just won't do worse than they would normally do by themselves. LiFePO4s have self-discharge rates of 2-3%/month, so around 4mA internal leakage per 100Ah capacity. Any cell-to-cell variation in this figure will cause a bank to slowly drift apart over time and that variation is guaranteed - manufacturing, aging...

In your case your cell voltages looked far too suspiciously progressive from cell 1 to cell 4 to be attributed to random factors, so I think you could be eating at least the rim of your hat!

[T1 Terry]

Originally Posted by OceanSeaSpray

It depends how tiny! 1000 hours is only 42 days when you keep at it continuously, and even at 1mA only, there goes 1Ah already. You would need to sense with microamps literally to be able to ignore these currents.

Now they just won't do worse than they would normally do by themselves. LiFePO4s have self-discharge rates of 2-3%/month, so around 4mA internal leakage per 100Ah capacity. Any cell-to-cell variation in this figure will cause a bank to slowly drift apart over time and that variation is guaranteed - manufacturing, aging...

In your case your cell voltages looked far too suspiciously progressive from cell 1 to cell 4 to be attributed to random factors, so I think you could be eating at least the rim of your hat!

That 2% to 3% per mth is only from fully charged, and anyone with lithium batteries knows better than to leave their batteries fully charged for mths on end. It's fine to leave a working system for mths, but it's asking for capacity loss to store them fully charged. At 50% SOC the self discharge is so small it would take a number of yrs for it to accumulate into an amount sufficient to accurately measure in regards to a % figure.
If the system is in storage then why would you leave a monitor connected, if it's a working system the 1Ah over a mth wouldn't ever eventuate as the system would include that loss in it's recharge cycle.

T1 Terry

[OceanSeaSpray]

Originally Posted by T1 Terry

That 2% to 3% per mth is only from fully charged, and anyone with lithium batteries knows better than to leave their batteries fully charged for mths on end. It's fine to leave a working system for mths, but it's asking for capacity loss to store them fully charged. At 50% SOC the self discharge is so small it would take a number of yrs for it to accumulate into an amount sufficient to accurately measure in regards to a % figure.

Self-discharge is driven by voltage and internal leakage resistance, so it would self-discharge at 2-3% if OCV is say 3.5V, but hardly at all with OCV at 3.3V???
Manufacturers rate them at 2-3% self-discharge per month, but they do auto-magically stop self-discharging once they are flat and dead. In the meanwhile it still gives them 1.5-2 years shelf life from initial storage at 50% SOC.

Originally Posted by T1 Terry

If the system is in storage then why would you leave a monitor connected, if it's a working system the 1Ah over a mth wouldn't ever eventuate as the system would include that loss in it's recharge cycle.
T1 Terry

We were talking about voltage sensing currents unbalancing the cells over time, not storage.
If you put it in passive storage at 50% for a few months, just disconnect everything. Come back 3 years later and you will be making a trip to the nearest rubbish bin. If I was putting a system in unattended storage for well over a year, I would be making other arrangements or think a lot more carefully about initial SOC.

[SV THIRD DAY]

Look, if I'm leaving my boat in storage for more than a year, I'm having someone come take a look at things anyway and that could just include turning on and then off the battery charger and I'm done. What's the big deal here? Why all the trash bin panic talk?

I know people just can't stand it and have to make this more complicated than it is...I get that...it's what people do in chat rooms, but really....I'm to worry about leaving my boat for several years because of my 2% self discharge on my LiFePO4 battery bank down to 50% then it stops ...ha ha ha....folks....a lot of other systems on a boat sitting unused for a year or two will have bigger problems than that!

I see this as a "I can think of a better negative than you can that shows how smart I am" contest in all honesty. But what do I know anyway...I'm just a guy with a 400AH LiFePO4 bank with none of the fancy control stuff that everyone says you gotta have or your boat will blow up and you will die, except it won't and I haven't.

As a data point, my bank sat for 4 months at a 90% SOC and what did I find when I got back to it....Bingo...nothing. No change, no self discharge of 2%...the exact same voltages on all 4 cells. So when I see some "OMG you will lose 2-3% a month"....well....you know what I see....

[hellosailor]

The self-discharge and storage advice appears to violate conventional logic and physics.


If a battery that is fully charged lost 3% per month, then after a convenient 15 months it would have discharged to 50% SOC all by itself, and it would then be at the "perfect" SOC for further storage. So what, if anything is lost? Nothing. According to what has been said, it should or could be stored fully charged and it would simply adjust itself [sic] over the next 15 months into the same 50% SOC that it should have originally been in.


?? Is there something illogical here or something being omitted ?? Some other consideration that supposedly will happen along the way fro 100% SOC to 50% discharged during those 15 months?

[goboatingnow]

I think this storage debate is getting a little confused.

Ive just finished some very long term tests on Li ( various chemistries inc ferrous) in 18650 form factors.

The 50 % storage is a function of maximising the life of the battery, Li will degrade merely sitting still

the life time degrades faster if the cell is left discharged or fully charged ( it degrades very fast if left discharged )

My test show that under and over excursions( limited not continuous ), don't necessarily damage the battery, ( well not in a discoverable way ) , merely that they may effect either the capacity or the cycle life.

what is comforting is that this technology is remarkably resilient as long as you stay away from a few areas.


PS those expecting long storage lives from LI are kidding themselves, These batteries have a life span, after that irrespective of how they were used you will see lower capacity.

PS those taking about voltage sensing systems affecting battery life are incorrect, its easy to make such sensing system draw microamps. ( thats excluding any other electronics running) There is no issue with long term voltage sensing if designed right


dave

[bcn]

Originally Posted by goboatingnow

PS those expecting long storage lives from LI are kidding themselves, These batteries have a life span, after that irrespective of how they were used you will see lower capacity.
dave

Dave ...

can you explain this further please?

bcn

[goboatingnow]

Originally Posted by bcn

Dave ...

can you explain this further please?

bcn

First while I have some larger prismatics , my experience is on smaller cells

My experience is that Li cells have a defined life span. to some extent its not that dependant on cell activity. There are operating factors that effect that life span of course . ( what I mean by life span is a defined loss of capacity ) .

Hence these batteries are great , but don't expect to be passing them on to your grandchildren

My own tests would seem to indicate a potential 5-7 year life at a capacity reduction of 30%. but I stress this is for the cells Ive tested and its based on projecting effects into the future. I have also seen significant variability in such life spans, ranging from very poor, i.e. well below the mean point to better then average


Sorry I can't release the underlying data, this was a paid for commercial project

hence the worst thing you can get is "old cells", or buying cells to day and thinking Ill install them in three years time .

Even if you maintain the SOC you will get the life span effect.


In my designs I now derate the capacity of Li to allow for life span effects where I want long life ( i.e. no user replaceable batteries) this leads to the funny position that the cell efficiencies in Li begin to approach LA again. ( obviously better KWH/Kg is better and remains so.)


What this means is that buying Li for occasional use boating is not a great idea. For heavy users it is. Just like the diesel engine, these are batteries that need to be used,

dave

[goboatingnow]

We were talking about voltage sensing currents unbalancing the cells over time, not storage.
If you put it in passive storage at 50% for a few months, just disconnect everything. Come back 3 years later and you will be making a trip to the nearest rubbish bin. If I was putting a system in unattended storage for well over a year, I would be making other arrangements or think a lot more carefully about initial SOC.

This doesn't make sense, voltage sensing can be virtually negligible, the LI has a defined life span in my experience, its that life span that will do damage, even if you maintain the SOC.

dave

[tsmwebb]

Originally Posted by goboatingnow

In my designs I now derate the capacity of Li to allow for life span effects where I want long life ( i.e. no user replaceable batteries) this leads to the funny position that the cell efficiencies in Li begin to approach LA again. ( obviously better KWH/Kg is better and remains so.)

dave

Does the Li battery just look like a smaller and smaller Li batter over time or are you seeing other changes to the battery behavior over time?