Dead dyson battery

[najrao]

I repurposed a red-blinking Dyson battery, substituting a Chinese general purpose BMC. Some hacking work and I got it to power a big cordless driver. Working nicely, I put it away maybe six months ago.
Today, the battery is dead -- as a dodo. Every one of the six cells is at 0 volts, and will not pass the slightest charging current. None shorted, all OPEN. I believe they are LG cells.
The BMS still passes checks, and does not leak current at any tap.
Uncanny.

[timeandfrequency]

Hi najrao,

When a well working BMS detects a low voltage condition on any cell, the whole pack is disconnected from the load.

Try to check if that still works.
Hook a lab PS (set to 21VDC and current limit 300 mA) to the cell assembly.
Solder a bunch of identical resistors (220 Ohm to 1k) in // to each cell.
Connect a load (82 to 220 Ohms 5W) to the battery pack, and measure the voltage across the load.
Lower the PS voltage slowly down to 10VDC : check if the low voltage condition actually disconnects the load. It should happen between 12 and 18VDC (2 to 3VDC per cell).

In the affirmative, would this mean that the BMS was drawing so much current for it's own power supply that it emptied all of the cells down to zero ?

Once the low voltage condition disconnected to load, the residual voltage across the load gives an information (using Ohm's law) on the leakage current that still flows from the battery pack towards the load. It should not exceed a few µA.

Do you have the reference code printed on the cells ?

[wraper]

Looks like CID (overpressure protection) built in into batteries disconnected the positive terminal. Li-ion does not like to be left at full charge for a long time. Probably batteries were somewhat overcharged too.

[najrao]

Thank you, timeandfrequency.
I have done this sort of BMS proving before. But not on this number, will do in a day or two. With the (dead) cells still connected to the measuring taps, the BMS draws 5uA at any voltage between 5 and 25 applied at the P+and P- . The mosfets (6+6 in parallel!) are not shorted or leaky. Individual taps draw <1uA at up to 4.5V.
I am concerned for the top balancing circuit built in (2//201 R switched by bipolars). But then at least 1V should have remained on the cells.
The cells are marked < LGAAHD2C1865> on each and another line <K307K036A7>, this latter being different from cell to cell.
Will get back soon.

Wraper, thank you too. I had done a monitored charging to 25.2V before putting away, and the current at termination was  <C/50. I routinely leave batteries fully charged for long periods in order to have them ready on demand. Not a good thing to do, may be, but I accept some reduction of life or capacity as a price of this service condition. Have never seen any other pack pack up altogether like this. There is no indication of ruptured cell seal, the cells still look pristine. I had used the prewelded nickel tabs to solder connections to the new BMS, no soldering to cell direct.

[wraper]

Leaving them charged at ~80% is an order of magnitude safer. Batteries which are left at full charge for a long time generally fail within a few years.

I had done a monitored charging to 25.2V

4.2V per cell is on a high side, there is basically no good reason to charge above 4.1-4.15V as you won't get more than a few percent of additional charge at best but greatly increase wear and the risk of batteries going bad.

[RoGeorge]

If you take a few datasheets from various Li-Ion cell manufacturers you'll find some are specifying the long term storage charge must be about 50%, or about 3.7-3.8V.  Expensive standalone chargers have a special charging mode for storing.  Same for factory charging, the cells come charged from the factory at about 30%, because they might stay on the shelf for a long time before being sold/used.

I didn't know that last year and charged a lot of phone batteries leftover to the max, then disconnected them.  One year later, most of them were bulged.  The more a Li-Ion stays at higher voltage, the more damage it gets.  Charge at 4.2V only when you know you'll start using them immediately.

For long term storage (months), charge them at max 3.7-3.8V/cell.  That's what is written in some datasheets.  (random example https://www.farnell.com/datasheets/2634739.pdf "preferred charge state for long time storage: 30...50%")

I don't know what happened to your batteries, my point is storing them for months at 4.2V is not good for batteries.

[NiHaoMike]

Charging to 4.2V/cell and then storing it would cause accelerated aging, but shouldn't cause complete failure. I think the only likely failure cause would be the BMS going into some high quiescent "latchup" state, probably buggy firmware.

[najrao]

I am sorry for not picking up on time, other urgent calls came up.

I did the BMS test suggested by timeandfrequency, with 2k divider resistors and load. The cells were disconnected altogether. The load switches ON only if a momentary short is applied between B- and P-, at 18V total and rising (this unacceptable behaviour is seen on many other BMS's as well). Raising the voltage gradually to as much as 30V showed that the BMS simply failed to turn off. And this on a new and unmolested part sourced from China!

If I may return to the original point of the battery's sad  demise, I did not depend upon the BMS when I charged it to 25.2V @down to 50mA on CV. I had even checked the cell balance to 50mV. Fitted to the load (driver), it had worked well too.

Even now, the shitty BMS for all its other faults does not draw over some 5uA at any voltage up to 30V. What I do not know is if the balancing resistors are turned on (possibly left turned on) inadvertently. How do I test the BMS for this?

I have to go with NiHaoMike on BMS 'latchup', but only in the balancing area.

Maybe it is best to get rid of balancing altogether, as recommended by Siwastaja. Or, even the entire BMS,  as I can monitor continuously while charging and frequently while discharging, a la Bosch. Live with the risk!

[timeandfrequency]

Hi najrao,

Well, this type of BMS design doesn't seem to be successful.

Even now, the shitty BMS for all its other faults does not draw over some 5uA at any voltage up to 30V. What I do not know is if the balancing resistors are turned on (possibly left turned on) inadvertently. How do I test the BMS for this?

If you used 2k resistors for the divder, replace the resistor at the cell you wanna test the balancing feature with a 2k2 or 2k7 value. The voltage at this cell will be higher than the other ones, but the balancing feature should always be able to limit the value to a bearable value (I would say something between 4.20 and 4.26 VDC) .   

One could certainly have a long discussion about the 'must have' features of a good Li-ion BMS, but here's my (wish) list :
a) disconnects the load if any cell faces an undervoltage condition (let's say below 2,5V)
b) disconnects the load in case of load overcurrent or dead short (the actual trigger value is clearly application specific, but in most of the cases we can expect something between 3A and 40A)
c) basic balancing feature : gently bypassing each individual cell when its voltage exceeds something like 4.225 V
d) no lock-up condition at startup
e) The BMS' own current consumption when a fault condition is triggered should never exceed a few µA
f) disconnects the battery pack from the charger if its total voltage exceeds N x 4.225VDC (N = number of cells in series).

The latter is clearly what the (new) unit you tested does not, as you could set the total cell voltage up to 30VDC (= 5VDC per cell !)