Battery voltage dropping far too low under test

[TheBikeMan]

Hi,
I test a large number of batteries with a computerized load tester and usually when a battery is bad the capacity is very low. In this case the capacity is fine but the voltage dumps.
I'm testing a 36V 15Ah battery. it fully charges to 42 volts and it fully discharges measuring a capacity of 15Ah. However as it comes to the end of the load test the voltage drops to 14V rather than the expected 30. I'm inclined to think that this is a BMS problem but not really sure. Any thoughts? Thanks guys.

 

[vfr400]

Doesn't your tester have LVC to save damaging the battery?

is the battery still providing current at 14v? If not, it's pretty normal to see that sort of voltage after the BMS has cut off. It's just leakage through the mosfets.

 

[TheBikeMan]

Thanks for your reply,
Yes, the tester stops loading the battery as soon as it hits the expected 30V. However, when the drain on the battery took it to about 32v it nose dived to 14v. This of course caused the tester to stop loading the battery. Before this happened it had measured a loaded capacity of 14.57Ah or 499.11 w. Once the battery hits the expected drain voltage (or lower as in this case) it goes in to a 100 minute cool down period before recharging the battery. At the end of the 100 minutes the battery voltage had recovered to 33V before recharging began and then it correctly recharged to 41.8 V. It's that sudden drop when everything else measures so well that has me uncertain. Any thoughts will be helpful

 

[vfr400]

I can't see what you're doing, but, as I said, the battery would normally switch off at 30v. After that, you can measure something like 12 to 18v leakage on the connector. The cell pack is still 30v.

What current were you testing it at?

 

[TheBikeMan]

I can't see what you're doing, but, as I said, the battery would normally switch off at 30v. After that, you can measure something like 12 to 18v leakage on the connector. The cell pack is still 30v.

What current were you testing it at?

Thanks, The tester I have doesn't register the leakage as it stop within seconds of hitting the expected low voltage. It will sometimes show a volt or two below 30 but this time it dumped to 14V before the tester could register that the expected voltage had been achieved. when the test is running it peaks at a 16Ah load. it does this in waves and reads the resistance as it is running through the cycles. I've attached the PDF of the test results with the values and the graphs. You can see from the graph that it takes a sudden, radical drop to 14V before the tester even has time to put it in cool down mode when it hit 30V.

 

[vfr400]

Maybe I'm misunderstanding you. "The voltage dumps" is not a technical term and I have no idea what you mean by it unless you mean that it drops from 30v to 16v when the BMS switches off, which is what normally happens.

1. You don't seem to be understanding what I'm saying. Your tester is measuring the voltage at it's connection to the battery. When the BMS is switched on, the voltage it sees is the cell-pack voltage. When the battery reaches LVC, the BMS switches off.

People would expect to see 0v on the connector at that point, but the mosfets that do the switching don't 100% block. Some charge still leaks through, but it's tiny.

The voltage you read with any sort of meter depends on the impedence of the meter. Typically, you see 12v to 18v with a normal DVM, but it could be anything between 0v and 30v. the voltage you see is absolutely meaningless. Effectively, it's zero volts because you can't draw any current from it. As soon as you try to take current, it collapses to zero.

2. In case you don't know, the battery's capacity depends on how much current you draw from it. The specification for the individual cells' capacity is normally based on a current draw of 0.2C, which would be 3A for a 15Ah battery. If you discharged at 16A, the capacity would be much less than the nominal 15Ah. How much less depends on the type and specification of the particular type of cells that are in the battery. I've seen as much as 20% less capacity when drawing 10 amps compared with the specified 2A. That's from a brand new and healthy battery.

In other words, your test results show a very healthy battery that's probably performing above its specification for capacity.

If you measure the voltage on the charge socket, you will probably get the true cell-pack voltage rather than measuring on the discharge terminals, as the charge mosfet doesn't normally shut off at 30v unless it's a crappy BMS that shares the mosfets between charge and discharge. Maybe your has shared mosfets, which accounts for the delay before you could charge again.

 

[TheBikeMan]

Yes, everything you are saying makes sense but for that I've tested hundreds of batteries with this tester and this is not the usual reading for a healthy battery. When I used the word Dumps I just meant that it drops significantly, quickly and unexpectedly.
I've been using this tester for a very long time on all types of batteries and this is the first time that it has produced a reading like this. Usually if the cells in the battery (Samsung 18650 in this case) are bad, the test will indicate a significant reduction in the capacity even if the voltage readings are fine. In this case it is the other way around. I've attached the test results from a healthy battery where you can see what the usually expected reading of the voltage at the end of the discharge test presents. This has been the case in all healthy battery test to date. Thanks for following up.

 

[vfr400]

Your battery tester should stop the test when it reaches your chosen LVC. I use 31v because there is very little charge between 31v and 30v. In no circumstances should you let it go further than 30v nor should you let it run with no limit and rely on the BMS switching off, otherwise you risk damaging some batteries becauses the BMS LVC can't be relied on. That's why controllers have their own LVC normally set higher (typically 31v). A typical BMS LVC is 29v, but can be as much as a couple of volts higher or lower.

Your second test report that shows the voltage at 28.89, which means that the BMS hasn't tripped. In the one where the voltage went down to 16v, it did trip. That's the difference between the two tests.

Does your tester have the feature where you can set the LVC to terminate the test? If so, what have you set it to?

Samsung cells doesn't mean anything. It's the type of Samsung cells that matter. They make many different types with different capacities, different discharge rates, different life cycles, etc.

Rechargeable batteries - 18650: Li-ion and Samsung | NKON

Li-ion and Samsung

eu.nkon.nl

 

[TheBikeMan]

Yes, I can create whatever profiles I want with the LVC determined. I've always run my 36V at 30V LVC. Once the battery is down to 30V the tester does not draw anymore amps from it. When it dropped to 14V at the end of the load test that was a static reading with no load or at the most a few seconds of load.

 

[vfr400]

Yes, I can create whatever profiles I want with the LVC determined. I've always run my 36V at 30V LVC. Once the battery is down to 30V the tester does not draw anymore amps from it. When it dropped to 14V at the end of the load test that was a static reading with no load or at the most a few seconds of load.

The BMS tripped at 30.1V (or higher), which isn't unusual. As I said 31v would be a better LVC for testing.