Seeking Normally Closed Latching Contactor and Q&A about balancing my Li Ion battery bank

So here are the test results of the second bank of cells. Again charging from a pack charged in parallel to 3.6v.

Similar results. Discharged the bank to the highest cutoff voltage on my inverter; 48.8v (3.05v per cell). Cells stayed in great balance all the way down, but started diverging near the bottom, around 3.1v. When the inverter cut off at 48.8v the cells had about a 180mv divergence.

Then attempted to charge pack to 3.45v per cell (55.2v). At around 54.2v the cells started to diverge to such an extreme that I had to stop charging. 4 cells were going higher than 3.45 (one up to 3.8v), while the other cells were still in the 3.3x range.

I think @jonr makes a good point here about efficiency (which is the same point I think @karrak was making earlier, although he didn't use the same term. I am coming to the belief that it is not necessarily so much an issue with capacity, as it is with a divergence in how efficiently the cells provide and absorb power. Two pieces of evidence may support this notion:

1) Both banks tested as having at least the rated 72AH capacity. Not even one cell was below 72AH. Although capacity may (and probably does) vary higher, the issues with some cells exploding higher after discharging the pack to only 3.0v or higher does not seem like a capacity issue.

2) I have placed the balances back on the cells after seeing they were diverging in voltage near the top of the RE-CHARGE cycle. The balancers did seem to even out the energy in the cells (per narrowing voltage readings), yet when I activated the charger again the same cells soon started to diverge and go high, while the other cells lagged.

Conclusion:
Although I may be able to get these 32 New Energy cells to usable in a pack by cutting off the discharge cycle even earlier, before the  divergence occurs, which may allow the charging to complete around 3.4-3.45 without some cells exploding higher (MAYBE), I am going to open a warranty claim with the manufacturer. My vendor has agreed to help me pursue this.

In the meantime I'm going to test above theory to see if the less extreme discharges that karrak has recommended allow the pack to be charged to at least 3.4v per cell. We will see.

Raj174 said:

Do you believe all the cells have issues, or just some? 

Well if the divergence in efficiency theory pans out then no cells may have issues other than the fact that as a whole, the 32 cells have broad differences in how efficiently they handle energy. So what I need to negotiate is to swap cells until the pack is closer in that regard. I'm hoping all the documentation from the testing will give my vendor enough ammo to convince the battery maker to provide a solution.

And again, I think this is the metric that Simon has suggested will smooth out over time.

Perhaps the maker will agree that the cells they were selling in that time frame were too far out of factory spec. I'll post the results of that conversation here.

jonr said:

This is why I have an interest in the BMS24T  BMS.  It claims 1.2A of balancing.

Yeah, I checked that out. Looks interesting. Would love to talk to someone who has tried out that solution. Not feeling too trustworthy of the reviews they posted. ;>

----------------------
Note to everyone following this thread: As I run the tests a second time, to verify findings and get scrupulous documentation for a warranty claim to the battery maker (New Energy), I am learning that the extreme divergence in cell voltages is only occurring during charging, not on discharging. Furthermore, the cells are recovering any divergence that occurred at the low end of the SOC, and staying in balance until nearly fully charged. It is only in the last stages of charging (I am currently determining that exact threshold) that cell voltages radically diverge. This seems to be a few cells absorbing the charge far faster than other cells.

I therefore must RETRACT everything I have stated in this thread regarding the balancers. These balancers may well do all they claim. I doubt though that they were designed to react as quickly as required at the nearly full SOC when voltages diverge quickly and sharply.

jim

Bill, thank you. I am wondering how best to do this. I certainly don't want to delete anything, but asking you to do a strikeout spree would be a chore because only sections of text within a post would be applicable.

I have always wished I could rename the thread, because it morphed into a completely different topic. If I could rename the thread, then add a disclaimer to the initial post that I made, that could tip off someone reading the post for the first time. Not sure if either of those are possible though.

Fair enough.

[EDIT: the balancers were NOT connected for any of the following tests]

OK, picking up from post #186, I am retesting both banks to verify for sure that the cells are diverging in voltage near the top of the charge. One reason for this is that it never hurts to verify, and, I don't think I logged the first test well enough to map an over-charging cell noted on the celloggers to the specific cell number it mapped to (cells are numbered 1-32). So this time I was careful to note which specific cells were diverging high.

SO ON THIS SECOND TEST A LARGE CLUE HAS EMERGED... I just don't know what to do with it.

After discharging the freshly balanced bank (charged on the imax 3.6v charger), and observing about a 100mv difference when discharged to 48.8v pack voltage, I then recharged the pack with the inverter's charger set for 55.2 (3.45v). As noted before, near the top of the charge curve a few cells (specifically 4) climbed high (over 3.5v) relative to the other cells. What was of interest however was that it was the same cell numbers on both of the two respective cellogers; cell 7 and cell 8. This mapped physically to cells 9&10 for the cellogger monitoring cells 16-9, and to cells 1&2 for the logger monitoring cells 8-1.

I found this coincidence striking. Both celloggers had the same two cell numbers spiking in voltage. I therefore decided to repeat this test #2, only with the overcharging cells swapped around. I swapped cells 9&10 with 11&12, and cells 1&2 with 3&4. Re-balanced the pack and repeated discharge/charge procedure. I was of course expecting the overcharging cells to move to from positions 7&8 to positions 5&6 on both celloggers.

NOTE: Not finding the photos for this first pass of test #2. No photos of first pass for now.

To make a long story short I was surprised to see that the overcharging cells did NOT move to positions 5&6 on the celloggers. You can see the swapped cells in the attached photos of the second pass. They were originally in numerical order for the first pass (of test #2).

How can this be? How can cells in a specific physical location on the pack be subject to charging higher than the other cells? See attached photos for second pass results. (I stopped the charging before the 4 errant cells went into dangerville).

The first answer is that the celloggers MUST be out of calibration. Positions 7&8 on both loggers must be reading the cell voltages incorrectly, and the pack is in fact still balanced. Except, look at the photos of the voltmeter. My voltmeter read the same numbers.

I'm going to run this same test on the bank now (cells 17-32). But not sure what to make of this mystery.

jonr said:

It would be less confusing if cell 1 in the first cell logger was physically cell 1.

I know! I was crossing my fingers that  the explanation could even be followed. Sorry. :<

jonr said:

Your connections are all clean and low ohms?   

Good question. I should add that the braided bus bar strips were completely mixed up in all the tests. I pulled them off and threw them in a pile, because each time I charged from the imax charger I used longer bus bars to do the parallel charging.

Same with the bolts. I don't see how there could be mixed resistance with all the mixing up... plus everything is new.

I should add that I was just looking over some photos of the celloggers in prior testing, and again it is logger positions 7&8 that go high.

jonr said:

Does your imbalance change with a high load vs no load?

Well, these measurements are all under no-load conditions.. while charging is occurring, except that the solar charge is powered up so that I can get the pack voltage remotely in my house.

When the charger is turned off the bulk of the cells that are running around 3.45v quickly drop back to 3.4v (which as a side bar seems evidence that those cells were still not fully charged. ?), while the 4 errant cells also drop back, but not nearly as much. They still read higher than the others.

jonr said:

What happens if you leave your cell logger completely disconnected and only use your voltmeter?

I will need to check that and get back to you. Good idea.

I just pulled the celloggers off, including the connecting leads to the high cells, and voltage remains high.

Also, attached are pictures of the cell voltages after sitting under no load and charger off for about an hour.

Here is a picture of the top of the array.