Does the charger always start dropping current when it's almost full?

[ASICS]

My lithium battery has a small capacity. When using a 16-series LiFePO4 58.4V 5A charger, it charges up to 58V and 5A, but most of the cells only reach slightly over 3.4V, with one cell reaching 3.65V and triggering protection. Is there a charger that starts reducing the current earlier? Ideally, the transition from constant current to constant voltage would be more gradual, starting at around 56V and reducing to 0.5 or 1A by 58V for a more balanced charge.

I have various chargers, ranging from tens to hundreds of yuan (with adjustable voltage and current settings), but they can only adjust the final voltage and maximum current. I also have modified power supplies with similar issues.

For single-cell charging, different buck modules behave differently; some don’t reduce current until 3.6V, while others start at 3.4V. Is there a charger that reduces current earlier, not just the maximum charging current?

[Berni]

This is why proper battery chargers have a balancing connector to access individual cell voltages.

For larger battery systems this job is usually separated out. So the BMS on the battery does the cell monitoring and balancing, then the BMS communicates the maximum charging current back to the charger. When a cell starts going too far out of balance and the BMS is not strong enough to balance it out, it will request a lower charging current from the charger.

If you don't have any of this, then this causes this exact problem you are having.

[Siwastaja]

One cell reaching 3.65V should not trigger "protection", but it should communicate "please lower the current" to the charger, such that charger follows CV not based on the pack voltage, but based on the highest cell voltage.

Further higher voltage (e.g. 3.75V) could then trigger a protection mode.

Now is there any cheap hobbyist friendly BMS-charger combination which does that, I don't know, but anyway this is what a proper BMS - charger combo needs to do.

[tom66]

It sounds most like your cells are out of balance and you might want to replace the cell that reaches 3.65V too soon.

The conventional lithium ion battery charging curve, which includes LiFePO4, is to charge CC then CV. This works okay for charge rates of 0.5C - 1C and does not wear the battery too much.  If you want to charge faster you need to use more complex curves that minimise the energy loss in the cell.  The cells also need to be warm.  You can see some of these complex charging curves in electric vehicles, for instance, Porsche Taycan is charging like this:



However if you do not need rapid charging, then stick with conventional CC/CV. 

Any complex charging curve likely needs to be implemented in software. The simple charge controllers are based on analog logic but these complex multi-step charging sequences require computer control and monitoring of battery cell temperature, balance, and state of health and charge.

[nctnico]

One cell reaching 3.65V should not trigger "protection", but it should communicate "please lower the current" to the charger, such that charger follows CV not based on the pack voltage, but based on the highest cell voltage.

I've never seen such a charger. At least not for consumer devices. They all are using CC-CV with a fixed voltage and no communication with the battery pack.

More to the question of the OP: if one cell reaches its maximum voltage way sooner than the other cells, the battery pack is likely to be end-of-life or constructed from non-matched cells. There is not much to be done about that as the worst cell determines a battery pack's capacity when cells are placed in series. Maybe discharging the pack fully (at a slow rate) before a charge can help to balance the cells.

[Berni]

I've never seen such a charger. At least not for consumer devices. They all are using CC-CV with a fixed voltage and no communication with the battery pack.

All the big chargers i seen used in electric vehicles(cars, boats, planes..etc) have some communication to the packs built in BMS, often via CAN. But these are very big battery packs that are often made out of Li-Ion cells, so you really don't want to mistreat those spicy fire potential filled cells. Or the smaller packs that have no BMS at all (usually used for RC models)typically bring the cell voltages to the charger to do balancing.

But yeah if one or two cells are quickly going out of balance, that means the pack is on it's way to death anyway. These bad cells get stressed more and more until they completely die and render the pack dead as a result. You can often bring packs back to life by replacing just the bad cells.

[Siwastaja]

One cell reaching 3.65V should not trigger "protection", but it should communicate "please lower the current" to the charger, such that charger follows CV not based on the pack voltage, but based on the highest cell voltage.

I've never seen such a charger.

Which only proves you have not worked with stuff like EVs, battery energy storage systems etc. They all do this as a fundamental feature. How they exactly communicate, varies; usually by CAN or RS485.

And, EVs and home battery systems are consumer products, just that these technical details are of course hidden from the end user so you are not aware of it happening.

Small battery systems e.g. laptops would not care. They just have BMS top-balance the cells and the charger regulates pack voltage during CV phase, which it can do independently. This works out in modest size (<10s or so) packs pretty well so no need to overcomplicate.

Now, does the OP need this feature or not is another question. They could try to start by just top balancing their cells properly, bring down the CV voltage a bit so that cell-level overvoltage detection does not trigger too early, and call it the day.

[Siwastaja]

It sounds most like your cells are out of balance and you might want to replace the cell that reaches 3.65V too soon.

... or, just... balance it. Imbalance isn't necessarily a sign of failure; maybe it was never properly balanced, maybe the BMS can't balance it for a reason or another.

Then, if after balancing the problem resurfaces, that would signify a cell quality problem.

Initial gross balancing is simple: just charge every cell separately, fully, using the same conditions (e.g. 3.65V, C/20 stop). A separate single-cell charger is a good instrument for this, and if you can top up two cells per working day, it only takes 8 days to finish the 16s pack this way.

The BMS might choose to not offer balancing at all, or offers only very low current maintenance balancing which cannot correct large differences in initial SoC when the pack was constructed. Therefore separate full charge cell-by-cell is a good starting point.

[tooki]

One cell reaching 3.65V should not trigger "protection", but it should communicate "please lower the current" to the charger, such that charger follows CV not based on the pack voltage, but based on the highest cell voltage.

I've never seen such a charger. At least not for consumer devices. They all are using CC-CV with a fixed voltage and no communication with the battery pack.

One example: laptop batteries. They invariably have SMbus communication with the charger, with a chip in the battery pack that performs safety functions (like undervoltage lockout) as well as monitoring individual cell voltages and pack temperature(s) and performing state of charge and battery health monitoring.

[nctnico]

One cell reaching 3.65V should not trigger "protection", but it should communicate "please lower the current" to the charger, such that charger follows CV not based on the pack voltage, but based on the highest cell voltage.

I've never seen such a charger. At least not for consumer devices. They all are using CC-CV with a fixed voltage and no communication with the battery pack.

One example: laptop batteries. They invariably have SMbus communication with the charger, with a chip in the battery pack that performs safety functions (like undervoltage lockout) as well as monitoring individual cell voltages and pack temperature(s) and performing state of charge and battery health monitoring.

Those are an exception indeed but only to a limited extend. From my experience with those 'smart' batteries I can tell you that those battery packs are prone to giving false information where it comes to SoC. So in the end these are not as ideal as you may expect and using a dumb CC-CV charge strategy is a good idea.

Either way, there is always a chip inside a Li-ion battery pack doing safety checks as it is impossible to construct a Li-ion battery pack which passes UN38.3 testing without such a chip or other circuit which monitors short-circuit current, under voltage, overvoltage, under temperature and over temperature conditions.

[tooki]

One cell reaching 3.65V should not trigger "protection", but it should communicate "please lower the current" to the charger, such that charger follows CV not based on the pack voltage, but based on the highest cell voltage.

I've never seen such a charger. At least not for consumer devices. They all are using CC-CV with a fixed voltage and no communication with the battery pack.

One example: laptop batteries. They invariably have SMbus communication with the charger, with a chip in the battery pack that performs safety functions (like undervoltage lockout) as well as monitoring individual cell voltages and pack temperature(s) and performing state of charge and battery health monitoring.

Those are an exception indeed but only to a limited extend. From my experience with those 'smart' batteries I can tell you that those battery packs are prone to giving false information where it comes to SoC. So in the end these are not as ideal as you may expect and using a dumb CC-CV charge strategy is a good idea.

Ummm… that conclusion does not follow from the observation, which may or may not even be true. Lithium packs configured in series practically always have per-cell taps to the BMS chip, because that is necessary to ensure safety. Literally the whole point is to make sure you’re not overstressing the weakest cell in the pack.

State of charge monitoring accuracy is an unrelated function (normally called a “fuel gauge”) that is just often combined in the same chip. Many algorithms are used, and to be accurate, the battery pack and load need to be carefully characterized. Additionally, as the pack ages, the fuel gauge needs to be given the opportunity to update its parameters, which is why you need to occasionally run your device all the way until it turns off, and then charge it all the way up, so it can learn where “full” and “empty” now are. This has nothing to do with the charging algorithm.