Which parameters show a cells capacity (V/A/IR/temp)?

[powersupply]

I try to identify the parameters needed to match twelve 4s1p packs arranged as 12s4p.
The goal is to match these packs in the most optimal way so that the (4xparalleled) cell capacities of each of the 12 (serial) stacks are as equal throught the stack as possible.

How that is done is another topic and I will address that later, an Excel/VB implementaion is in the works, so PLEASE do not get into the how and so on for now.

Right now, I only need to find out which parameters (volts/internal resistance/heat?) identify each cells capacity/capabillity.
And if voltage would be sufficient in the descirbed scneario.

For instance:
I have 12x 4s1p 5Ah packs from hobbyking arranged as 12s4p.
http://www.hobbyking.com/hobbyking/store/__18631__Turnigy_5000mAh_4S1P_14_8v_20C_hardcase_pack_USA_Warehouse_.html

Did follow recomendations (as far as I had the equipment) breaking them in with light load and charging them slowly, three times.

Then used them 3 times on my 750W scooter, discharging them down to 3.6V or higher.
Then put ALL packs in parallel (4s12p) and charged them to 16.4V (4.1V).
Their balance wires were connected to each other during both discharge and charge (on cell level).

Then used them 2 times without their balance wires connected to each other and noticed some larger differences (up to 100mV), prompting me to slowly adjust them to each other before paralleling them ALL and charging them to 16.4V, like above (their balance wires were connected to each other during charge only).

After his, I used them twice (down to 3.507V for the lowest), and charged them as they were (12s4p) with a 48V CC/CV charger (the original SLA charger modded to 48V). These two times, their balance wires were not connected to each other during both discharge and charge.

Measured their (cell) voltages and found even larger discrepancies between them as expected.

Now I wonder, is the residual cell voltage good enough to identify the cells capacity?
I mean their voltages do represent their charge amount after the discharge, right?
Giving an indication how much capacity that particular cell had in relation to other cells?

Equipment I have are one CellLog and an older Fluke DMM (while the CellLog has a higher resolution than the Fluke).

Thanks!!!

 

[Hillhater]

I would do a IR check on each cell to initially weed out any "outliers" and do an initial assessment for any problem cells.
see this.. http://endless-sphere.com/forums/viewtopic.php?f=14&t=48739

 

[dnmun]

my understanding of how you match capacity, is to charge the cell to full charge. usually 4.2 but some now say in the pdf pdf i read 4.23V and then discharge to a fully discharged state which will depend on the cathodic material. but in general the full discharged resting voltage is usually 2.7V (but is 2.5V for the new NCA cathodes http://www.samsungsdi.com/battery/cylindrical-INR18650-13Q.jsp) . but since you would be using the same cathodes with carbon anodes the 2.7V would give you final capacity.

i don't think anyone actually does try to do capacity balancing though. they just use the cells that arrive as they are. i find about 5% variation in the effective capacity of the cells that arrive from hobby king and after they are balanced they stay very close to the labeled capacity, and repeatable. i get essentially the same capacity that the hobby king cells are rated at too, not short when they arrive like some have claimed. i account for that in the 5% variation i see on the cells coming from the formation charge.

jmho

 

[powersupply]

Thanks for the input!

Hillhater, in that video he says it is not scientific, and I agree, as he leaves out the variation of the load (halogen bulbs do vary quite much depending on their load), and also the 10 seconds seem a bit arbitrary.
But most important is that the test is just made at the beginning, however the end (or close to the end) resistance is more important, as it is during that time that the LVC gets triggered.
And the internal resistance does change quite during discharge, doesn't it?


Dnmun, the reason why I'm concerned about capacity balancing are exactly those percentages which would trigger the LVC first.
If I would use the best in parallel with the best, and the worst with worst, I come to over 12% difference around LVC.
I am already trying not to charge to full capacity in order to extend their life, but these percentages here are avoidable.

Considering that I have only one CellLog, I'm thinking of doing 2 test runs (with my Fluke to record the amps) and draining them till LVC.
Would need to do that 14 times though.
And record the amps through the pack, which is a big problem, as I would need to record it manually.

I know this has been asked many times but I can't find it right now, how much current can the balance leads take?
If it is a few amps, then I could use the highest input of the CellLog as an amp meter.
With a shunt instead of a battery and a modded divider (to give that input a higher sensitivity) at the input.

Has anybody done that?
Does it look doable?

 

[powersupply]

Instead of measuring the actual current I could use a constant current source, why didn't I think of that right away?
Just have to find a resistor (lamps fans etc) capable of handling 1c (14v@5A).

Any objections to that idea?

 

[dnmun]

i don't usually use the cellogs when doing the discharge to measure capacity. i discharge into a regular oil filled radiator type electric heater. i measure the current out of the pack through the watt meter and keep track of it at the end as the first of the cells goes to the LVC. i follow the voltage with a regular voltmeter. it is pretty clear when it drops off the map. goes from 3V to 2.7V very fast. put it back on the charger immediately when the LVC on the BMS hits.

i found that the 5% range that the cells have when they arrive can be balanced in one or two cycles of the BMS. one pack had a cell at 4.12V when the BMS hit the HVC of 4.225V so i had to discharge the pack down under 4.05V and start again to balance it but eventually i got it from 7.4Ah to 8Ah after balancing it to 4.2V. then all cells out of 24 were within the 15mV spread.

 

[Hillhater]

Hillhater, in that video he says it is not scientific, and I agree, as he leaves out the variation of the load (halogen bulbs do vary quite much depending on their load), and also the 10 seconds seem a bit arbitrary.
But most important is that the test is just made at the beginning, however the end (or close to the end) resistance is more important, as it is during that time that the LVC gets triggered.
And the internal resistance does change quite during discharge, doesn't it??

The video is just a guide as to one easy way of finding a value for RI.
You dont have to use the halogens, ( but they are a simple solution for a load bank) others have used heaters, motors etc etc.
The 10 secs is just a suitable period to allow the current to stabilize.
Unless you have a lot of time and resources, any test you devise will have some compromises and inaccuracies, ..but at the end of the day all you are looking for is a COMPARISON between the cells, so providing you are consistent with your test method, ( same starting voltage, same load, same temp', etc ) you will be able to weed out any bad cells and any cells with significant differences.
Absolute IR values ( +-10% ) are not what you need.
Any test is better than no test !
PS, ..You dont actually need to measure the current if you keep a standard load resistance and test voltage, You can simply compare voltage drop data for each cell.
So just charge/balance your packs up to a set voltage, attach the cellog and record each cell voltage.
Then hook up the load, wait for the voltage to stabilize, then record the cell voltages again.
Work out your Vsag data,
Rinse and repeat for each pack
Easier still if you have a logging cellog .
TIP: .. pick a test load that is close to the magnitude you expect to see in normal use.

 

[powersupply]

Very useful tips, and they make sense, thank you!
Timewise I have as much time as to plug in the next battery and start a test and leave, a timer/LVC condition (from the CellLog) will stop the test an hour later.
The draw should be around 1C (5A) from these batteries.
The scooter draws around 15A (from four parallel batteries, i.e. 3.75A per battery)

The reason why I want to use the CellLog is to identify the capacity of each cell within the 4s1p battery pack.


What I still don't understand is the time when the cells are nearly depleted, isn't that period the most crutial?

Keep in mind that 4 packs will be paralleled, so it won't be only one cell carrying the current, but four.
I noticed that the cells behave slightly different under different loads and at different capacity stages.
Drkangls weird voltage/current/mA curves come to mind.
Meaning the stronger battery will be exhausted later than the (parallel) weak.

But if they are really as stable as dnmun said, then I might worry too much indeed.

Yes, I have a Cellog 8S with memory. Only one functioning unfortunately.
The other turns on now but shows no voltage,

.


Thanks!

 

[dnmun]

you have to sit with the watt meter at the end of the capacity discharge so you can read the total Ah discharged.

if the pack goes to LVC then the ground for the watt meter is lost and the meter loses it measurement so if you did not read it on the last cycle you miss it altogether.

but you can tell with the volt meter or the cellogs which is the one that is gonna reach the LVC first. they stand out usually, especially after about 70% discharged.

 

[Hillhater]

What I still don't understand is the time when the cells are nearly depleted, isn't that period the most crutial?

Yes, the end of the discharge is critical, but what dont you understand ?
simply dont discharge below 3.6 v .. there is little useful capacity below that, and great risk of damage to the cells.

Keep in mind that 4 packs will be paralleled, so it won't be only one cell carrying the current, but four.

If you are parallel linking the balance taps also, then each cell will also be "balanced" by its coupled cells ( IE they will each maintain the same voltage during discharge )

.. I have as much time as to plug in the next battery and start a test and leave, a timer/LVC condition (from the CellLog) will stop the test an hour later..

I would advise against letting cell testing run unmonitored .. unless its in a fireproof location. !

 

[powersupply]

dnmun, I was thinking to hook up the CellLog to the balance wires, and the load to the main battery power cables.
When LVC or timer hit, it will disconnect the main power.
But I could also use a USB power supply or hook it up to the PC to keep it powered, right?

And as mentioned, I'm planing to either use a constant current, or as Hillhater mentioned just take some current measurements and use them as ballpark, as long as it is the same resistance. Don't know yet.
Must sleep over the 10 second test method.
Also, I thought to LVC at 3V, maybe even 3.5V, not at 2.7V.
That cliff happens way before that with these batteries, right?

Hillhater, yes, the cells will be paralleled via balance wires, but eventhough they are all at the same voltage, the current they deliver/contirbute will be different. It is basically the same with the traditional balancing, just instead of serial balancing, I try to do a parallel matching/fitting before I wrap the cells up and stow them in their compartement.Agree with 3.6V or so. Scared of 2.7V.