LiFePO4 balancing.

[Siwastaja]

Whatever happens, it always tries its best to equalize the voltages.

Which is a wrong thing to do. You are assuming this is the right metric, but it's not, because voltage is not indicator of SoC under charging/discharging current, especially on LFP mid-curve. The balancer is doing needless work and wasting energy as a result. When on-grid, maybe the wasted energy doesn't matter, but if the "balancing" is going on while discharging, the wasted energy then reduces the usable pack capacity. This might be more serious than just the power consumed by the indicator LED.

It is impossible to say if the net effect is positive, neutral or negative by following the voltages alone. Much more sophisticated monitoring and analysis would be needed, involving accurate current measurements (for charge integration).

A proper algorithm would monitor a full cycle without any balancing to figure out the amount of active balancing needed during cycle, and then repeat just that pattern. All the small details need to go right.

This is clearly a cargo cult engineering product. As an expert, I would not recommend it.

[paulca]

Some of the balancing boards I am looking at provide a whole set of parameters allowing things like, only enabling the balancer when certain conditions are reached.  Such as maybe only enabling the balancer while in "absorption" phase for an hour before cutting charge or maybe bring the balancer back in at the bottom when your pack starts on the descent again.  Many options.

The Victron "Smart shunt" thing sounds interesting, but it also sounds expensive and proprietary.  It's basically a learning fuel gauge type affair, with a coulombs counter maybe, you can get it to watch the pack for a full cycle, then pick what you consider 100% SoC and it goes from there. 

[shapirus]

Which is a wrong thing to do.

Depends on multiple factors. In my case, which is to keep well-matched cells equalized and not allow them to drift apart over time, and equalize them initially after they are assembled into a multi-cell battery, it's exactly the right thing to do.

I didn't care to add the circuitry to turn off the balancer during the discharge stage or only turn it on when the charger is on. Doesn't make any difference in my case: a few mA of the balancer's quiescent current or tens or even a few hundreds mA equivalent of energy wasted at charging and discharging the capacitors don't make any difference compared to the typical 10-30 A discharge current.

It may be a good idea to do it in the general case though.

Yes, these balancers may not work well (and, the rest of the variables being equal, neither will the resistive ones) with poor quality batteries made of cells that are unmatched in terms of capacity or internal resistance or both.

Those will obviously require more sophisticated logic to get the best performance of them and extend their life as much as possible, but it's a special case: normally it's preferred to pick matched cells of decent quality to build a battery.

[Siwastaja]

Which is a wrong thing to do.

Depends on multiple factors. In my case, which is to keep well-matched cells equalized and not allow them to drift apart over time, and equalize them initially after they are assembled into a multi-cell battery, it's exactly the right thing to do.

It's totally wrong thing to do, although as a side effect, if you equalize the voltages all the time, indeed the cells can't drift apart in SoC very far. However, this is very inefficient way to keep a pack balanced and reduces usable capacity, and the primary reason to balance a pack is to increase the usable capacity.

Especially if your cells are well-matched, then all you need is a small classic dissipative balancer.

You have absolutely no idea what you are talking about. You are a consumer buying some Chinese shit and based on that "experience" you come to teach experts who actually have researched and developed such systems. This is just getting ridiculous. I'm out of this bullshit discussion.

[shapirus]

Yes it's difficult to admit that a dirt-cheap Chinese board can now in most situations solve the problem which previously required a skilled engineer and costed a fortune to solve. I understand.

However, there are still many cases where complex solutions are required, and skilled engineers are still in demand there. So there's no reason to get frustrated, not before AI replaces them in a few years, anyway.

(but balancing a damn battery pack is not one of them.)

[paulca]

Assuming we are talking about the same active capacitve balancers...  aka these:
https://www.aliexpress.com/item/1005005050192601.html

I gather it's a series of logic counters and mosfets.  The tests I seen, A*V in > A*V out.  By a fair margin of double figure percentage.

That is what you would expect with a PWM on a mosfet gate every rise and fall it goes through the variable resistive region.  Every time you turn the mosfet on you have to charge the gate.  Everytime  you turn it off you dump that charge to ground.  It's all wasted.

If you have a LiFePO4 cell at 3.30V and it's neighbour is at 3.20V and you attempt to balance them, you could end up transferring 25% of the capacity of the cell via a 80% efficient balancer.  I think the flat portion on most cells is about 0.5V from 0% to 100% resting OC.  10mV difference can amount to quite a chunk of the capacity.  Its that reason balancing is not recommended specifically for LiFePO4 within that voltage curve region.  Li-Ion is more linear and affords balancing during charge.  I would always balance charge my high current LiPos, unless I'm in a real rush, when "fast charge" canes them at 2C until one cell hits 4.20V and takes half the time.

[paulca]

Back on topic.

I have been using the lead-acid boost charge profile at 14.40V to "effectively" top balance the 4S pack through "absorption".

It's been a week or so now and ... it's not exactly working out as planned.

I have one cell stuck in it's flat charge region while the other 3 are climbing up over the 3.65V target.  The highest got to 3.900V while the lowest was still dragging in the flat region at 3.460V

While I was expecting them to appear unbalanced at the top undercharge the self discharge back to the 3.6/3.7 region would effectively equal them out.

That is assuming however that all cells reach their upper voltage curve on every charge.  Once that stops being try the cell imbalance will probably continue to get worse until there is a cell taking far too much voltage on it's own.

While 3.9V is not terminal for a LiFePO4 cell, it's not optimal or recommended.

The only balancer I have had a total "capping" current of 100mA per cell.  Thankfully, the sun is now out of scope on the panel and that 3.9V cell is rapidly falling back under 3.7V.

I suppose it bumps the BMS board up a notch in priorities.  I was hoping to get away with it until the other 4 cells arrive mid april, then I don't need a 4S BMS, but can go to the main range which start at 8S.

I have a 2 Amp balancer/BMS but it's 3S.

When I finish work I suppose I will have to go out and manually top balance them all again... and probably repeat that at least weekly until the other 4 cells arrive.

We don't have much sunny weather planned anyway

[shapirus]

I'm not sure there is such thing as absorption when it comes to lithium cells. You can't "overcharge" lead-acid cells, because once they reach a certain voltage, they won't be able to get any higher, because this "extra voltage" is "spent" for dissociating the electrolyte, which is known as so-called "boiling". Overcharging them simply means slowly (how slowly? I don't know) destroying their electrolyte, which, in non-sealed (or manually unsealed) batteries, can be restored by bringing the acid-water solution back to the required density.

This is why it is possible to balance lead-acid batteries by charging them at a higher voltage: when some cells begin producing hydrogen and remain at a certain voltage level, the others will eventually catch up and become equalized. This is my understanding of this process, at least.

It apparently doesn't happen like this with lithium batteries, which is why it is important to have their cells balanced, especially as you're nearing 100% SoC, or, rather, as some of the cells are reaching their maximum allowed voltage.

[paulca]

There definitely is absorption during charge. 

Take an average small LFP cell.
Resting voltage, lets say 3.4V, close to 90%+ SoC.
Apply 1 Amp of current to it and the cell might rise to 3.8V
Remove that current and it will settle back to 3.4V.
Apply 5 Amp of current to it and the cell might rise to 4.0V
Remove that current and it will settle back to 3.4V.

It's that "boost" differential I was hoping would scale in the same way as the upper voltage curve scales.  When cells have 14.40V across them a lot of those volts are accounted for by the "boost" differential across the cells.  Because there is so little actual capacity up there I was assuming it is caused by the resistence to charge rising as fast as the voltage.  Thus a reasonable amount of time spent unbalanced in that region, assuming of course that no cell exceeds 4.2V seemed likely to balance the pack.

I think ultimately it would/will, but with such high charge currents over 0.1C it does not appear to be working out.

[Zucca]

This is why I decided to use my LiFePO4 from 20% to 80% SoC and to avoid any possible chinese or cheap or professional BMS.
BMS does cause more harm than good on the long run.

Yes I had already some fun with thermal events and white smokes events. All because of (poor designed) BMS.
BTW designing a good BMS is a nightmare.

I mean instead to built clever protection in the battery we should use clever chargers and clever loads.

That said, I think it is crucial to have a stupid passive monitoring in the battery pack that reports the voltage of each cell to the outside.
Don't know but I am a fan of stupid battery packs that are not a pain to replace (look at here what a horror movie), and have all the protection outside.

And yes, for portable application where 100% of the SOC needs to be squeeze out, then and only then, I will bent over and install a BMS.

PS: Very interesting discussion, thanks for all the good info here. Especially thanks to Siwastaja for sharing his knowledge!

[paulca]

I have seen people white smoke BMS's by testing the overcurrent protection.  I believe the device had 20 mosfets (some odd back to back "bridge" config) 10 x 170A brutes in parallel and claiming something like 300A breaking current.

Turns out it did break the current path, but unfortunately for one poor little mosfet at the end, it was the slowest.  Puff!  Cracked through the case, skid marked the board.

Chinese electronics, halve the stated ratings.  If you "need" 300A breaking current, buy 600A.

I personally plan on only really using 40A, but technically the battery is 100A capable and should be cabled and fused accordingly.  Thus a 100A BMS.  The only time the over current protection which I believe is 300A rated is going to be a question is in a short circuit or serious overload scenario and the BMS popping a few mosfets trying to halt that, I'm ok with that.  £50 BMS is cheaper than the potential damage a short will ultimately lead to.  To be honest, I might consider setting the overcurrent protection on the BMS above the battery DC breaker rating, which is probably cheaper still to replace if it dies breaking 300A short circuit.

I don't need to worry about overcurrent under charge, my solar panels have known limits.

On the "balancing" aspects, I am picking a BMS that has fine grained settings over the balancer.  When it comes on, when it goes off, what imbalance it will tolerate etc.  Having it come on as soon as one cell hits 3.65V and remaining on while there is a >20mV imbalance, using it's full 2A active balancing path... ONLY for top balancing.  Then disconnecting 99% of the time.

The BMS would also have solved todays 3.9V mishap.  The "charge mosfets" would have shut down the incoming current from the solar panel as soon as a single broke a 3.75V limit (by example).

[SiliconWizard]

I mean instead to built clever protection in the battery we should use clever chargers and clever loads.

Uh, no. I certainly would not second that.

While some BMS in cheap (read: crap) batteries may not work well, adding a reasonable level of protection inside the battery shell itself is a must IMO.
LiFePO4 are not as dangerous as standard LiIon, so they can take a good amount of beating, but still.

That said, for sure if you buy "random" chinese batteries from shaky sources, the odds of getting something reliable is pretty low, and I'm not surprised about your experience with the built-in BMS.

Generally speaking, you protect what needs to be protected (here the battery) as close to it as possible, rather than rely on "clever" stuff connected to it.
Besides, your clever stuff is never going to be as clever as you thought.

Maybe you can also rely on clever users rather than add any protection to a product, in the same vein.

Just my 2 cents.

[paulca]

Maybe you can also rely on clever users rather than add any protection to a product, in the same vein.

Just my 2 cents.

I tried this recently.  I put zero protection on a PCB I had printed.  Willingly.  Believing, "It's a dev board, only I will be using it.  I'm not stupid."

Well.  Guess what?  2 voltage regulators and an Optical->I2S module later, I'm regretting that decision.

On my other battery builds, most appropriately the one in context is the "growing, budding, house battery style" with LiFePO4 cells.  It's pay day in a few days. 

The shopping cart contains
* a BMS (the best reviewed one from AliExpress) JK-BMS.  Which includes RS-485 monitoring and control.
* a small dual way din enclosure
* a 100A DC compatible breaker (not from aliexpress!)
* 2 large 100A capable 'female' bulkhead style screw terminals, ideally with safety caps.

The DC Breaker will be overcurrent protection and (assuming I can find a dual pole breaker) the isolator switch.  I know using a breaker RCBO as an isulator switch is frowned upon, but much like your air-cond or heating system the battery will very, very rarely get turned off.

As I have the mechanical skills of a lettuce that has been microwaved for 30 seconds, I am placing the battery inside a "Really Useful Box" with the only exposed bits being the female + and - terminals and... the BMS display/rs-485 cables.

[Zucca]

I mean instead to built clever protection in the battery we should use clever chargers and clever loads.

clever stuff is never going to be as clever as you thought.

for "clever" I meant KISS or appropriate solution.
Example to limit the current and prevent fire, a simple fuse is a very clever solution.
To me better than any µP or high tech BMS.

Why the closer to the cells the better?

I mean a good protection circuit will work either if inside the battery pack as well as integrated in the load or charger.
Of course a temp sensors on the cell can`t be avoided in some application, but why this general concept to put a protection or BMS or whatever in the cells pack?

A battery pack is a consumable device, we should be able to replace it throwing away the less components as possible....

[Zucca]

zero protection on a PCB I had printed

IMHO the problem here is the zero protection of course.

[paulca]

As I am also spec'ing a new battery for "Audio electronics", because I gave up on understanding power supply filters, I have been looking at some smaller, lesser, BMS's.

That puts me in the market for some of those "cheap and possibly nasty" BMS boards that seem to get put into every multi-series cell battery pack.  They even sell kits sans the cells.

The issue I see with them, well, some of the issues I have seen with them.

Breaking current.  Most use bridge mosfets and the smaller board only have 4 in parallel.  The BMS may be advertised as, say, 30Amp.  But also claim a breaking current of 100A.  However under test you will find the over-current protection has a delay, sometimes that delay is quite long, several seconds.  Even a large lunch box full of 21700 or 18650s in 3 or 4P could exceed 100A in a short circuit ,especially if the battery has been wired "beefy" as it should.  The issue is that those 4 mosfets trying to break that fault current may fail or more likely will succeed releasing white smoke from several of the mosfets that never switch off at exactly the same time.

Balancing current versus charge current.  None of these cheap ass BMS's are actually "charge" aware.  They are completely passive to the charge current.  The only thing they can do about it is disconnect the whole battery if they don't like it.  So when a solar panel is pushing 10-15 Amps into the pack and it only have 0.5A balance current, one cell will hit 3.65V and the whole pack will disconnect.  Then a few minutes later when the cells rest down it will turn on again.  On, Off, On, Off, On, Off until those poor little mosfets die or overheat.  That battery disconnect is absolutely NOT recommended for anything else on either side of the battery, or for the battery itself.  Solar charge controllers, for example, detest having their battery removed, with such large inductors and buck converters removing that huge balast from them, I'm guessing, causing huge transient inductive spikes and removes any "system voltage" reference the charge controller had.

This is something that only caught up on me recently.  I was too familiar with using combined balance chargers.  These ARE charge aware and in the scenario a cell gets to 3.65V they will engage the balancer, but they will also lower the charge current such that the balancer can keep that cell under 3.65V while giving the others a chance.  This can prolong charging, but gives you a full absorption cycle at 3.65V.

Large battery BMS's are not like this.  The only thing they can do with charge current is (a) ship it between cells with capacitors or (b) resistively cap high cells (c) disconnect the whole pack.

[paulca]

BMS's are essential to maintain the "DO NOT EXCEED" values of the Lithium cells.

Mostly because almost all consumer batteries sold today are lithium.  However, correct adherence to those "DO NOT EXCEED" values, cannot and should not be placed into the hands of the layman.

People out there will watch a video on TikTok on how you can charge an 18650 with nothing more than a 5V USB charger!  Seriously.  Go search YouTube and TikTok and you will find hundreds and hundreds of similar videos.

Each and everyone of them a fire harazard.

When "lads" buy one of the larger "replaces a lead acid" 100Ah lithium pack and it turns out to be LCM not LiFEPO4 and they... without reading a single line of the packaging least any manual and without the slightest understanding of what they are doing, go ahead and charge it with the 20Amp lead acid charger.... leaving it completely unattended for days.

That is why lithium batteries come with BMSs inside.  I would not be surprised is that becomes a law/regulation in places like the UK.  That all lithium battery packs sold much include a BMS.  Of course that BMS must pass a certification programme which will cost the manufacturer thousands, so UK sold BMS's will tripple in cost and everyone will go to china anyway.