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

[goboatingnow]

Originally Posted by Viking Sailor

Internal resistance increases as SOC decreases. Thus, when two cells with unequal SOC's are connected in parallel, there will be less surge current with two low SOC cells then when the cells have a high SOC.

Not with Li technology. Attempting charge a very low SOC ie near the reverse voltage from a cell with medium SOC would result in huge currents flowing.

On the other hand cells in the flat portion of the dv/dSOC curve would transfer little of no current. In fact due to manufacturing tolerances. The lower SOC cells could actually slighty bleed into the higher SOC.

Dave

[Marqus]

Originally Posted by T1 Terry

A Li cell is fully charged at 3.4v rested but it requires 0.05v over voltage to create a charging condition. A cell at 3.35v rested is somewhere between 80% and 90% charged, a cell at 3.0v could be anywhere between 100% discharged and 70% discharged. State of charge can be measured by cell volage to a degree but it's in the single milli volt range, a large imbalance has occured by the time there is sufficient difference between cell voltages for it to tranfer. I remember reading some where on the EV DIY forum where someone linked all their cells together and left them for 48hrs, then connected them is series and charged them, they were still all over the place when the 3.45v mark was passed for each cell. This is the method the majority of BMS active cell balancers use, when a cell reaches say 3.8v a controlled rate short circuit discharge across a resistor in switched on and burns the excess voltage down and turns off at 3.5v. Unfortunately manufacturing being what it is some will turn on at between 3.85v and 3.75v and turn off between 3.55v and 3.45v. some drift into the higher end and cells are damaged, some drift under the 3.45v mark and then the balancer is trying to discharge the real charge capacity, this overheats the little units either burning them out or causing them to latch on gradually dragging the cell dead flat.
As balance is really only an issue at the the very top of charging, a cell running away and overheating, or at the bottom and a cell being dragged into reverse voltage. The second one is extremely unlikely to happen in a hose battery situation so top balancing is the choice of methods for house batteries.
Now, if the cells are dragged up to 100% full the over voltage doesn't happen, if the 100% charged point and top end balancing is left to a time the operator can watch what's going on then if a cell goes high a load can be applied to that cell to bring it back down, once all cells are 3.45v to 3.5v and the charge rate has dropped to zero the cells are considerd fully charged, the 100% is reset on the SOC meter and the system returns to being charged to 98% each time till anoher top balance is required. If the cells stay within 50 milli volts while working then a top balance isn't required.

T1 Terry

Thanks for writing Terry; there's a lot of good info in there.
Allow me to extract just a few points from your write-up, and ask a follow-up question (correct below if wrong):

1) For a typical boat House Bank situation, balancing is only required initially and every now and then.

2) Cell re-balancing should be done when the batt is near fully charged (as opposed to when it is half way or discharged). This is Top-balancing.

3) If a BMS malfunctions (and it's relay burns/fails) it may kill one/more cells.

4) A cell imbalance of less than 50millivolt can be ignored.

5) To charge a cell, you have to apply a voltage that exceeds the cell voltage by a minimum 50millivolt.

Let's focus on this last point, that you need to exceed a cell's terminal voltage by 50mV to get current to flow into it.

Let's assume we have just charged a 4-cell battery to a point where the HVC stops charging. Now we find an individual cell measures over by, say, 100mV.
If we re-connected the 4 cells in parallel, we should expect current to flow from this highest-voltage cell into other cell/s until our cell's voltage drops down to within 50mV of the others.

Is that what will happen, or is life not that simple?

[goboatingnow]

Originally Posted by Marqus

Thanks for writing Terry; there's a lot of good info in there.
Allow me to extract just a few points from your write-up, and ask a follow-up question (correct below if wrong):

1) For a typical boat House Bank situation, balancing is only required initially and every now and then.

Yes , typically at the receipt of the cells from the manufacturer and in fact in fractional C charge and discharge regimes, may never need cell balancing

2) Cell re-balancing should be done when the batt is near fully charged (as opposed to when it is half way or discharged). This is Top-balancing.

Cell balancing can be top or bottom, since in boats the batteries are generally kept near full charge if possible, top balancing is the only realistic option

3) If a BMS malfunctions (and it's relay burns/fails) it may kill one/more cells.

I would disagree with Terry , here, A good BMS should do charge divertion, rather then bleed off balancing. IN practice since you dont need cell balancing, you dont have the potential failure mode anyway

4) A cell imbalance of less than 50millivolt can be ignored.

No, you are missing what Terry says. Cell imbalaqnce is exhibited by the different time it takes each cell to reach the completion voltage, its not really detectable by measuring voltage difference. Cell (a) is nominal 3.4 fully charged, does not exhibit say 3V to show its at 60%

5) To charge a cell, you have to apply a voltage that exceeds the cell voltage by a minimum 50millivolt.

Yes you have to overcome the EMF of the battery

Let's focus on this last point, that you need to exceed a cell's terminal voltage by 50mV to get current to flow into it.

Let's assume we have just charged a 4-cell battery to a point where the HVC stops charging. Now we find an individual cell measures over by, say, 100mV.
If we re-connected the 4 cells in parallel, we should expect current to flow from this highest-voltage cell into other cell/s until our cell's voltage drops down to within 50mV of the others.

Youve asked this before , NO it doesnt work like that you cant balance LI cells by connecting them in parallel. You can only balance cells connected in series( with individual cell charge diversion or bleed off) or not connected together at all.

Cell balancing works by sensing each cells as it charges. As each cell reaches the completion voltage , the charge is diverted away from that cell ( in a sophisticated BMS), Each cell therefor arrives at teh completion voltage in a slightly different time.

Without cell level monitoring all you can is remove the bank charging when you see the cells completion voltage x number of cells. ( ie 4x3.4). at this point IF there is imbalance some cells will be above the completion voltage and some maybe below it. If you just connect them in parallel, The over completion cells will drain slightly into the under completed ones, but once the voltage reaches the flat part of the curve, its stops.

BUT the cells are still out of balance as in reality outside fully charged and fully discharged the cell terminal voltage really doesnt change with SOC. In fact with manufacturing tolerances, you could actually bleed the lower SOC into the higher SOC just as easily.

Is that what will happen, or is life not that simple?

Why get too worked up about it

[s/v Jedi]

We ought to look at designing a charger-converter which takes the output of our regular boat-charging sources (alternator, AC charger, Solar/wind controller etc) and which charges each cell individually. This might not be very difficult...

cheers,
Nick.

[T1 Terry]

Heavens Dave, I even had trouble following that

Let's assume we have just charged a 4-cell battery to a point where the HVC stops charging. Now we find an individual cell measures over by, say, 100mV.
If we re-connected the 4 cells in parallel, we should expect current to flow from this highest-voltage cell into other cell/s until our cell's voltage drops down to within 50mV of the others.

If a cell is charged beyond 3.45v it's still only 100% full, the 100mV is surface charge and does not equal any Ah of capacity. If you connected all the cells in parallel yes the 100mV would dissipate across the other 3 cells but they would still be only 100% charged. If it was the other way around, the HVC stop charging point was reached but one cell was 100mV below the 3.45, so it read 3.35v, then the pack was connected in parallel the high voltage cells would drain down to the low voltage cell till the 50mV point was reached, the voltages would measure the same across all the cells while they were linked together but the state of charge of the low cell would still be lower than the other cells, when the links were removed the voltage of the low cell would still be less than the others but now the state of charge of any of the cells is unknown. The only definites are 100% charged if the cell voltage is greater than 3.45v and close to zero state of charge if the cell voltage is 2.8v or less, these are all rested voltages measured after 10 mins.

Quote:
3) If a BMS malfunctions (and it's relay burns/fails) it may kill one/more cells.

I would disagree with Terry , here, A good BMS should do charge divertion, rather then bleed off balancing. IN practice since you dont need cell balancing, you dont have the potential failure mode anyway

Great in theory Dave, yet to see one that can do it in real life. Where is it going to divert the charge too? The only way it could be done is if the high cell is discharged into a capacitor and the capacitor is discharged into the low cell, possible but very high tech and very expensive and asking for thinks to wrong. The BMS manufacturing trade is mostly electronics techo's building units in their garage, it's not a high tech mass produced by factories with a gazzillion dollar R&D budgets. Factory EV builds don't use a single cell BMS so they aren't building them for a big market, Tesla don't even have single cell management. It’s only the home built EV brigade that use these things and even they are divided as to their worth. Many that fitted them are now removing them.

As far as individual cell charging, there was a set up made by one of the constructors on DIY EV forum to charge his Tractor Pull vehicle batteries, worked great as long as no one sneezed near it, not moisture resistant, not quite boat ready yet. It required a 48V battery to power it anyway so not really house battery stuff.

Seriously, top balancing isn't the boogie man it sounds, it is very simple and straight forward and once done thy stay that way for ages.
Balance is only an issue at the 100% fully charged mark, the only reason to 100% charge is to reset the state of charge meter. If charging is limited to 98% full by the SOC meter balance should never be an issue. If the HVC trips it will tell you it's time to do a 100% balance again but rather than that come as a surprise at an inconvenient time I'd recommend weekly balance checks to start with the try stretching it out to mthly balance checks. This will tell you how much drift your SOC meter has and how much drift your battery pack has and will point out if a particular cell pack is a repeat offender.
Keep in mind, for the first 30 cycles or so the cells go through a settling in period.

There is another method (isn't there always :lol Dave mentioned earlier about the cells drifting together rather than apart, this happens if you used PWM voltage control and charge to 13.8v and hold it there. I've had a 360Ah pack I built for a friend start out at 380mV out of balance, within a week of cycling powering my fridges via an inverter they had drifted to within 130mV, last time I checked they were 14mV. This method needs a good HVC to ensure no cell goes over voltage and it needs to be set at 3.6v and auto reset when the run away cell drops back to 3.4v. This system uses the principle that there is no additional capacity in any charge voltage above 3.45v so milli amps are used out of the battery to bring the run away cell back to 3.4v, then the charging starts again, each time milli Ah are being added to the low cells but not to the 100% full cell, finally the low cells are filled up to the 100% mark. There are other reasons why they drift together, to do with the higher voltage cells providing more Wh to the load than the low voltage cells but if I tried to explain how that worked I'd end up confused at the end.
This method requires 100% faith in the charge controller, solar charging to do the final top up and 100% faith in the HVC successfully stopping any cell going into voltage run away. Not as risky as it sounds but out at sea is very different out on the highway, you can't just stop and check out what's happening if you need too, you need the system to be as close to fit and forget while in operating mode and leave the maintenance time for when it's convenient. I guess it's a call up to each operator, only they can decide which system suits better.
I've ramble on way too much now, if anyone want's any of that explained better I'm happy to do so.

T1 Terry

[OceanPlanet]

Originally Posted by s/v Jedi

We ought to look at designing a charger-converter which takes the output of our regular boat-charging sources (alternator, AC charger, Solar/wind controller etc) and which charges each cell individually. This might not be very difficult...

cheers,
Nick.

Which would be less expensive than simply using a real BMS (with alternator protection, etc)?

Right...

[Sparohok]

Originally Posted by goboatingnow

Youve asked this before , NO it doesnt work like that you cant balance LI cells by connecting them in parallel. You can only balance cells connected in series( with individual cell charge diversion or bleed off) or not connected together at all.

This keeps coming up because what you are saying seems contrary to laws of physics. Granted, LiFe cells in parallel will not self-balance perfectly if they are in the relatively linear voltage range of (say) 20% to 80% SOC. However when top or bottom balancing, voltage clearly does vary significantly with SOC. Even the linear range is not perfectly flat. If there is a voltage difference, even a very small one, then significant current will flow between parallel cells through the extremely low internal resistance of LiFe cells. I'm curious what could prevent this from happening.

This is amenable to experiment. If you hook up a LiFe cell at 20% charge and one at 80% charge in parallel (thru a resistor please!) what SOC do you expect them to end up at? Or, what about one at 80% and one at 100%? I don't have any large cells handy but I think I can scare up some small ones to do this experiment.

Finally you certainly can top-balance multiple LiFe cells in parallel using a CC-CV charger. As long as you limit the current to the safe charge rate of one cell, this is the simplest and safest way to balance LiFe cells.

Martin

[goboatingnow]

Originally Posted by s/v Jedi

We ought to look at designing a charger-converter which takes the output of our regular boat-charging sources (alternator, AC charger, Solar/wind controller etc) and which charges each cell individually. This might not be very difficult...

cheers,
Nick.

Working on it

Dave

[T1 Terry]

Finally you certainly can top-balance multiple LiFe cells in parallel using a CC-CV charger. As long as you limit the current to the safe charge rate of one cell, this is the simplest and safest way to balance LiFe cells.
Martin

It is essential the voltage is limited to 3.45v, current isn't really important as long as 3C isn't exceeded. The reason for this is if one cell is near or at 100% and the voltage is higher, say 3.6v, the electrolyte in the 100% cell/s will overheat while the other cells are still accepting charge wrecking these cells.

This is amenable to experiment. If you hook up a LiFe cell at 20% charge and one at 80% charge in parallel (thru a resistor please!) what SOC do you expect them to end up at? Or, what about one at 80% and one at 100%? I don't have any large cells handy but I think I can scare up some small ones to do this experiment.

Why through a resistor? The cell voltage can't go higher than the donor cell voltage, these cells aren't effected by 3C or less charge or discharge, it's highly unlikely you will get that much current from with such a small voltage differential.
As far as the 80% SOC cell and 20% SOC cell, how will you determine these capacities? what state of charge do you expect to end up with? Could be anything between close to what you stated with to a 60/40 balance but highly unlikely to end up wit them evenly balanced, insufficient voltage differential in the middle of the state of charge curve. The voltage of the cell at 20% charge will still be 3.2v or higher, the cell at 80% charge will be 3.3v or lower.
Here is a discharge curve chart, loking at the 0.5c discharge curve try to imagine how flat that is at zero discharge.

T1 Terry

[goboatingnow]

Sorry if were going round in circles

This keeps coming up because what you are saying seems contrary to laws of physics. Granted, LiFe cells in parallel will not self-balance perfectly if they are in the relatively linear voltage range of (say) 20% to 80% SOC. However when top or bottom balancing, voltage clearly does vary significantly with SOC. Even the linear range is not perfectly flat. If there is a voltage difference, even a very small one, then significant current will flow between parallel cells through the extremely low internal resistance of LiFe cells. I'm curious what could prevent this from happening.

This is all about degree of balance. In large format cells suffering high C charge and discharge. Severe imbalances can result.

In addition current liFe manufacturering is producing batteries with significant spreads of impedance and capacitance. Hence the voltage at any particular cell in the " flat " part of the curve cannot be used to compare cells SOC( in the main)

Cell balancing is an attempt to equalise SOC not an attempt to equalise terminal voltage

The only place on the curve we know anything is in the area of rapid dv/dSOC, ie near the fully charged state.

Hence taking out of balanced cells ( worst case when you get them new), cannot simply be resolved by paralleling them. current will flow but not necessarily from the greatest SOC to the lowest. The voltages will be the same but the SOC might not be.

Over time parallel only strings will converge that's true. And in fractional C environments this tends to be the case and is why I don't recommend in circuit BMS cell balancing. In high C situations severe imbalances have been recorded which need active balancing techniques, rather then paralleling and cell convergence.

Finally you certainly can top-balance multiple LiFe cells in parallel using a CC-CV charger. As long as you limit the current to the safe charge rate of one cell, this is the simplest and safest way to balance LiFe cells.

Martin

Well yes and no. It depends on the degree of cell tolerances. The drawbacks are that LI tech is very sensitive to high voltage. To get cell convergence you have to set the V point close to the HVC point, otherwise lazy cells never balance. Given the tolerances today in LiFe, this approach can stress individual cells and cause premature failure. Cell failure in big parallel strings is very " amusing ". ( which is why I don't recommend paralleling Big LiFe sets.


Just to recap. My experiments show the best way is individual cell charging and optional balancing. This can be done by simple charge diversion ie disconnecting each cell as it reaches the assigned completion voltage. Slight overcharging and bleedback has been shown to achieve better convergence. But I don't like the additional voltage stress. ( and the risk of exceeding HVCs )

This area is really EV stiff not domestic Boat banks as we don't generate imbalances in the first place.

The other thing is that because we have to build nominally 12 or 24 systems the discussion of only parallel balancing orore correctly convergence is actually not the problem. Balancing is a issue in series strings


In my view other then initial balancing, typical boat style banks made up if series strings then paralleled need optional HVC and LVC protection, though I don't share Terry concern over failure modes. It's worth pointing out that the LV point in Li is well below LAs. Typically it's around 2 to 2.5 volts. Hence your boat is down to 8 to 10v in most modern boats every instrument would be alarming. The lights would be virtually candles. The user would know well enough he needed to recharge.

I say this because Li technology is sometimes regarded as sensitive not robust. However in particular LiFePo4 is very robust and increasingly can withstand even HV transgressions with too much damage.

Today you could take standard yittrium
LiFe stick 4 big capacity cells in series. ( the HV point is now up at around 16v). Hence you could charge these from conventional battery chargers , solar, alternators.etc without much worry at all and no further control systems. Yes in most cases you'd only get 80 -90% SOC bit unlike LAs there's no penalty, so what

Because LAs are cheap we tend to ignore protection systems that act to ensure battery life and we accept simple charging systems and the fact that we ruin the bank every so often. ( ie few people get the expected capacity or cycle life ) As LiFe drops in price the same attitude will emerge and most of the control control electronics will fade away, except in demanding applications.

Dave

[Marqus]

Originally Posted by T1 Terry

If a cell is charged beyond 3.45v it's still only 100% full, the 100mV is surface charge and does not equal any Ah of capacity. If you connected all the cells in parallel yes the 100mV would dissipate across the other 3 cells but they would still be only 100% charged. If it was the other way around, the HVC stop charging point was reached but one cell was 100mV below the 3.45, so it read 3.35v, then the pack was connected in parallel the high voltage cells would drain down to the low voltage cell till the 50mV point was reached, the voltages would measure the same across all the cells while they were linked together but the state of charge of the low cell would still be lower than the other cells, when the links were removed the voltage of the low cell would still be less than the others but now the state of charge of any of the cells is unknown.

Thanks Terry for the clear answer.
I take your point that to monitor the battery's SOC in practice, it is advantageous to balance all cells by dragging each one into the top end (3.45v).

Also take note of your opinion on BMS's. To some it would sound attractive to run a LifePO house bank with as little "intelligent electronics" as possible. I know the LVC and HVC units are indispensable, so looking at the feasibility of a relatively simple and fail-proof balancing regime is of interest.

I have had a long list of failures in electronic "smarts" on boats, so it would be nice if I can keep smart electronics away from crucial things like the engine, the anchor, the Lavac and the house battery bank.

Another forum member reported use of simple "resistor boards that help balance the cells at the end of the recharge by burning off excess amps on individual cells". These boards are not classified as a BMS. Have you come across that approach and do you know how it works?

[s/v Jedi]

About the individual cell charging... 12V charging source is well enough for a device that needs to charge a 3V cell. If others used a 48V source before, it must have been because that was what they had to work with.

Also, I can quickly draw a schematic that will do this with just a couple of dollars worth of components (just one power transistor really) so cost is not the issue either... The issue is to design a device that can do this with high energy efficiency, so without burning up excess input power. This is the problem I have with those designs that shunt to resistors... You throw away power which can be hard to come by on boats... Most try to save power, not waste it.

I'm at the point of pulling out my old PIC programmer :-)

ciao!
Nick.

[goboatingnow]

Also take note of your opinion on BMS's. To some it would sound attractive to run a LifePO house bank with as little "intelligent electronics" as possible. I know the LVC and HVC units are indispensable, so looking at the feasibility of a relatively simple and fail-proof balancing regime is of interest.

Why bother with a cell balancer at all. Hence nothing to fail.

Dave

[OceanPlanet]

Yep. Only the user...

Excellent plan.

[Marqus]

Originally Posted by goboatingnow

Why bother with a cell balancer at all. Hence nothing to fail. Dave

Indeed an interesting approach. Hence the interest in exploring an idiot-proof, quick, manual balancing method.

In that scenario, an out-of-balance indicator (eg. LED) would be useful. Because it would not intervene, it would not potentially put the house bank at risk like a faulty BMS could.

Maybe suitable out-of-balance indicators are available from the Radio Control hobby industry. You would probably want operator settable alarm threshholds, because someone may want to keep a factory-matched set in closer balance than a rough and ready assembled set.

I recall Terry mentioning an RC Logger. Terry, can your Logger serve as an out-of-balance alarm, with user adjustable threshholds?