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

[NahanniV]

Originally Posted by goboatingnow

My experience with cylindrical LiF ( and other LI) is that at fractional C discharge and re charge, cell voltage seem to converge and do not diverge.
Dave

What do you think is the mechanism for this convergence ?

[dlentz]

Two questions as to theory. If you're top balancing by charging cells in parallell to keep voltages equal across the cells, would not a bank consisting of 8 cells in 2p4s be preferable to larger cells in a 4s string? As it would at least pair 2 cells together to help avoid an over voltage situation in a single cell. Obviously it would not totally prevent it and it introduces additional issues with connectors and such, but it also would provide a potential back up in the case of a cell failure (just reduce the bank to 1/2). Would there be any advantages?
And second, while it is mentioned that the cells may be oriented in any position, is it possible to stack cells on their sides, maybe 2-3 cells high? Are the cases sturdy enough and would the interior pouches be affected by lying sideways? All of the installations I've seen have the cells upright.

[goboatingnow]

Originally Posted by NahanniV

What do you think is the mechanism for this convergence ?

I dont know, its seems to very dependant on the level of charge and discharge. I dont know enough about the cells at a chemical level to speculate .

What I do observe ( and Ive been using various Li clyindric cells for about 2 years in various designs) is that low C use and low C charge seems to result in the cells staying or converging on a common point. Additionally never float charge and aim for an 80-90% SOC limit point, though sometimes thats hard to fathom.

Though I have to say all my recent designs use cell level recharging as it eliminates many issues and allows the user to ignore initial balancing which is useful in a real product environment.

The use of current counting is fraught with errors as well as you have to decide what is the top cutoff and the bottom cutoff, ie to reset the count, otherwise the amp counting gets quite rapidly out of sync. In practice I use 10 -12 bit ADCs to measure voltage and find that more reliable , certainly in lower capacity cells , where the voltage slope is more defined.

Dave

[diugo]

Originally Posted by dlentz

Two questions as to theory. If you're top balancing by charging cells in parallell to keep voltages equal across the cells, would not a bank consisting of 8 cells in 2p4s be preferable to larger cells in a 4s string? As it would at least pair 2 cells together to help avoid an over voltage situation in a single cell. Obviously it would not totally prevent it and it introduces additional issues with connectors and such, but it also would provide a potential back up in the case of a cell failure (just reduce the bank to 1/2). Would there be any advantages?

Cells in parallel are at the same voltage, so both can still be overvoltaged or undervoltaged simultaneously---in which case you may destroy two cells instead of one. A good cell in parallel will not "protect" a bad one---the bad one will bring the good one down to its level, making both vulnerable.

The only advantage of parallel topology is the redundancy it affords in the event of a single cell failure. However, this can be mitigated simply by keeping a fully charged unconnected spare cell in reserve. (On a boat in the middle of nowhere, I would strongly consider this idea, BTW.)

And second, while it is mentioned that the cells may be oriented in any position, is it possible to stack cells on their sides, maybe 2-3 cells high? Are the cases sturdy enough and would the interior pouches be affected by lying sideways? All of the installations I've seen have the cells upright.

The only unacceptable orientation is upside-down---as this might interfere with proper venting. Cells are often banded together under considerable tension (to prevent bloating). This would suggest that the cases have enough internal strength to support several times their weight. If it's still a concern, rigidity can be increased by banding, even more so by banding with steel plates on two or more vertical sides.

[Maine Sail]

Originally Posted by T1 Terry

Once the cells are full, they are full, the voltage will rise rapidly. This means if one cell reaches full even 0.1Ah ahead of the others it will hit 4v easily if a simple system of replacing the Ah that came out with the same number of Ah back in. This is why I recommend restricting the charging voltage to 13.8v float and 14v end of boost charging AND have the secondary back up of the cell logger with the alarms set at 3.6v and 2.8v, simply to protect the cells from harm.

T1 Terry

This is not what I am seeing. The cells are still well below the knee at about 3.5V. I am seeing a cell voltage drift that seems only superficial and not egregious. Once you hit it with a load the cells all wind up back within a few mA of each other. Just don't understand why this happens...

[mrm]

Originally Posted by Maine Sail

This is not what I am seeing. The cells are still well below the knee at about 3.5V. I am seeing a cell voltage drift that seems only superficial and not egregious. Once you hit it with a load the cells all wind up back within a few mA of each other. Just don't understand why this happens...

My understanding (maybe flawed) is, that there are two possible energy storage modes in such cells. The primary mode is chemical energy and this is what provides the bulk of the capacity. However, the secondary mode is a static charge. Once ions stop flowing what we are left with are large areas of positive and negative electrodes separated (isolated) by some material, with a certain dielectric properties.

Maybe, just maybe, you are beginning to see the secondary mode charge starting to accumulate? This would get discharged pretty fast with any serious load and bring all cells to their potential resulting from chemical storage mode?

A far shot, admittedly...

added: you meant mV, not mA, right?

[Maine Sail]

Originally Posted by mrm

added: you meant mV, not mA, right?

Yep mV not amp

[goboatingnow]

Originally Posted by Maine Sail

This is not what I am seeing. The cells are still well below the knee at about 3.5V. I am seeing a cell voltage drift that seems only superficial and not egregious. Once you hit it with a load the cells all wind up back within a few mA of each other. Just don't understand why this happens...

The knee voltage is at 3.7 for LiF is it not , its around 4.2 for LiCo.

Dave

[diugo]

Originally Posted by Maine Sail

This is not what I am seeing. The cells are still well below the knee at about 3.5V. I am seeing a cell voltage drift that seems only superficial and not egregious. Once you hit it with a load the cells all wind up back within a few mA of each other. Just don't understand why this happens...

Your cells all have similar internal resistance and charge, but differ slightly in capacity. When the same current is applied across cells in series with similar resistance, all the cells display similar voltage. Ohm's Law. Or as some here like to put it, Ohm's Theory/Opinion

When cells are tied in parallel together, charge moves between the cells until current drops to zero all the cells acquire the same voltage. Does that mean all the cells have the same charge? Yes---because otherwise there would be current. Does that mean all the cells have the same state of charge? Only if all the cells have the same capacity.

When the cells are removed from parallel, positive voltage drift in one cell suggests lower capacity than the others.

[Maine Sail]

Here's an example.

I just charged the 400Ah "bank" (in series) to 13.98V and let the current drop to about 1.2A in CV mode.

Cell voltages with the charger still on are:

#1 = 3.498V
#2 = 3.521V
#3 = 3.497V
#4 = 3.468V

The cell spread from lowest to highest is 0.053V and will hang out close to that spread, even without the charger, until I load the cells or wait a week or more.

I apply a 100A load for 30 seconds and my cell spread from lowest to highest is back to 0.004V or 4mV...

I just found it interesting and wanted to see of others see the same thing. If charging to a higher voltage I can see where this could set off BMS alarms even when the cells are "technically" balanced. It seems that in order to get an accurate voltage spread between cells they need to be "cleared" of any surface type charge...? I am familiar with this with LA batteries but not read about it with LiFePO4...

I am messing around with the charge voltages I plan to use on-board and so far 14.0V keeps me well below my max and still gets me the 400+ Ah's of capacity I want. This bank will rarely get cycled below 50% DOD so I have been cycling up and down to 50% and all cells stay in safe range when charged to 14.0V even with the superficial (perhaps) cell voltage drift.....

[diugo]

If cell voltage imbalance is removed by briefly applying a load, that simply sounds like surface charge to me. The false voltage is just due to the varying capacitance of the cells.

[hellosailor]

Maine-
You guys are measuring voltages to the ten thousandth of a volt. What are you metering them with, with what accuracy, precision, or repeatability on the measurements? i.e. if you measure the same cell four times, do the numbers float?

[charlie p]

Earlier in the thread, Maine Sail stated he was using a Fluke 179 meter. It has an accuracy spec of 0.09%. Also, he is only measuring to a millivolt, not a tenth of a millivolt. For the 179, the millivolt digit is not exact but it does contain information.

The resolution of the 179 is a millivolt, meaning that relative measurements, which are what you need for detecting balance drift, are better than the absolute accuracy.

[Maine Sail]

Originally Posted by charlie p

Earlier in the thread, Maine Sail stated he was using a Fluke 179 meter. It has an accuracy spec of 0.09%. Also, he is only measuring to a millivolt, not a tenth of a millivolt. For the 179, the millivolt digit is not exact but it does contain information.

The resolution of the 179 is a millivolt, meaning that relative measurements, which are what you need for detecting balance drift, are better than the absolute accuracy.

All cells are compared to the same meter "control" for measuring drift. I am only measuring to the thousandths of a volt not ten thousandths as HS suggested..

Sure the last digit is likely off a couple mV, but I don't really care. What I care about are the cell to cell differences using the same meter as the known control for differences between cells.

The meter repeats the same measurements with no drift no matter how many times I place the leads on a particular cell.

[hellosailor]

Charlie, not knowing what the low scales are on the F179, if he's got four digits that's arguably a 10-volt (9.999) scale, so on the whole scale, that's still ten-thousandths of the scale. Not ten-thousandths of a volt, but as a percent of the whole scale, since accuracy is ofen expressed by the percent.

So in this case an accuarcy of 0.9% on a 9.999 volt scale? means it is only accurate to plus-or-minus 0.0899 volts, or 0.09 volts more practically. With a 3.5 volt reading, OK, that's still going to be about 0.03v of uncertainty. That last rightmost digit means something--I just don't know how much faith (or concern) I'd put in it.

I'm not knocking anything here, I've just become more attuned to putting numbers in context. Or at least trying to. "What I care about are the cell to cell differences using " Understood and agreed.

I wonder if simply banging on the cells, which would compress the contents briefly, would change the voltage output readings? "Contents may settle in shipping" as the potato chip makers say.