Balance charging lithium cells

[paulca]

Something I have looked into before.  It came down to, yes, you can do non-resistive balance charging but it's (1) much more complicated and (2) takes much longer.

However, I wanted to point out that those who believe that you don't need to balance charge very often obviously do not use multi-cell packs in the real world very often.  EVERY time you charge a multi-cell pack you MUST use a balancer.  On a good pack you might only see 100mV difference during charge or discharge, however with a straight 12.6V charge across a 3 cell pack you could end up with 200mV inbalance and charge a single cell to 4.35V or higher easily.  If you are high current charging the pack they tend to explode in flames at that voltage.

My experience is in high current LiPos discharged at 10C and charged at 2C, 100mV imbalance is not uncommon and depending on use they can be much, much higher.  Consider that LiPo alarms or other voltage monitors tend to take their running power form cell 1.  It's also not completely uncommon to pull 8.4V off cell 1 and 2 on a 3S or 4S pack to power radio gear, although generally frowned upon, it's still done.

Resistive balancing is loseful.  The total amount of loss would need to be calculated.  It would be interesting to see how much power is actually lost using a set of LiOns such as Adam Welch's solar shed setup.

[thm_w]

Something I have looked into before.  It came down to, yes, you can do non-resistive balance charging but it's (1) much more complicated and (2) takes much longer.

However, I wanted to point out that those who believe that you don't need to balance charge very often obviously do not use multi-cell packs in the real world very often.  EVERY time you charge a multi-cell pack you MUST use a balancer.  On a good pack you might only see 100mV difference during charge or discharge, however with a straight 12.6V charge across a 3 cell pack you could end up with 200mV inbalance and charge a single cell to 4.35V or higher easily.  If you are high current charging the pack they tend to explode in flames at that voltage.

...

But some real world packs do not use a balance circuit, as I was told by Siwastaja. Even in microchips multi-cell 18650 charger app-note they left it out.
I just checked a two cell pack here I have been using for a few months (numerous charge and discharge): 3mV difference in cell voltages (3.880 3.877V). Of course the capacities are very close, I can't remember the specifics but I think within 50mAh.

High current charging for sure may be a different story, this is only charged at 1A max.

[Nitin25]

So which logic you have used to sense individual cell voltage.
how to sense effectively individual cell voltage with microcontroller if there are more number of cells in series. As microcontroller works only 5 volt.

[paulca]

But some real world packs do not use a balance circuit, as I was told by Siwastaja. Even in microchips multi-cell 18650 charger app-note they left it out.

Yes, but this is cheating you.  What will happen is as soon as the first cell reaches 3.0V the pack disconnects.  As soon as a single cell reaches 4.20V the charger disconnects.

This is exactly why laptop packs without balance circuits have such a short life span.  It's obviously for one reason.  It makes companies more money.

How many laptop packs have been taken apart by cell refurbishers to find one single parallel cell set at 2.9V but the rest are absolutely fine, yet the pack is unservicable.

[OM222O]

there seems to be different schools of thought here like always 

As I mentioned earlier the packs that I plan on making are DIY packs made from recycled laptop packs which as pointed out earlier, only have one or two bad cells but the rest are fine.
so there will be no guarantee of capacities being equal or even the cells having exactly the same chemistry.

A few pointed out that if a cell is not charging like the rest it would only mean that they are band and will fail soon so replacing them is best, but I'm still not sure how a pack can remain balanced when cells have different capacities  yes, I know I compared them to super caps which are different, but the main reason that they charge to different voltages and need balancing every time, is exactly because they have different capacities. maybe I should do some real world tests after I build the packs and do 10 full charge / discharge cycles using both methods and compare the voltage levels of each cell at the end of each charging / discharging cycle to see after the initial balancing, if the cells drift again or remain close together.

[paulca]

Generally, as a rule of thumb, do not mix different capacity cells in series... or parallel for that matter.

If you watch through some of this guys videos he provides links to software designed for what you are doing.  Basically you capacity test all your batteries, enter them into the software and it gives you the best way to lay them out in parallel and series.  He also shows various ways he has used to balance them.  His latest videos show an elaborate system where each parallel cell pack has a BMS itself and these chain together over I2C (or similar) to communicate with monitoring software.

[paulca]

Also, just a polite reminder that Lithium cells are genuinely dangerous.  Unprotected cells can get pretty upset if overcharged or shorted.  The metal 18650s can still swell, rupture and spray flames.  Just be careful and don't charge them unsupervised unless you are totally sure of your setup.

[OM222O]

the protection is already taken care of. I'm just not sure about balancing them or not and if I decide to go down that route I want to minimize losses

[Siwastaja]

there seems to be different schools of thought here like always 

This is the classical difference between hobbyists consuming "for hobbyists" products (RC toys), and li-ion battery analysis, pack construction and management professionals, such as myself.

The mantra of "absolutely needing balancing" is ubiquitous on forums, yet hilarious if you have ever looked at real-world products. Apparently the world's biggest and most well known companies have no idea what they are doing. HobbyKing School of Engineering knows better!

Of course, if your target is just a hobby, then it might be wise to listen to what other hobbyists say, and use the designated hobby products; they have a wide audience testing the products.

If you are actually designing a product, go with the real-world information from the industry. It's a long road anyway. Completely ignore any RC hobbyist information, Battery University, and similar sources. Peer reviewed scientific studies are better, although slow and inefficient way to learn. If someone is talking about "li-ion and LiPo", it's a good telltale sign already they have no idea what they are talking about.

Even better, do teardowns of commercial and industrial li-ion packs. You'll see many different approaches!

As said earlier, non-balanced, even non-monitored packs are absolutely ubiquitous in consumer and industrial products; especially in 2s and 3s, but fairly commonplace up to 6s. I have measured a few BOSCH power tool packs that, after End-Of-Life of years of hard use and highly increased DCR, are still in balance within 10mV at top, using Samsung SDI cells. They do not include any electrical connections to the series taps. It's highly possible some units are worse, and do imbalance, but it seems they still don't go up in flames. OTOH, we have seen quite a few incidents of flames and early destruction of cells in laptop packs which do include balancing.


In general, if you don't know what you are doing and need to design a safe system, I recommend you do neither kind of a system. Keep learning, and in the meantime, use something off-the-shelf that everybody uses already; it's a good way to hobby around, but having used a battery analysis toy from HobbyKing to measure a Chinese toy battery from the same place doesn't magically make you an expert on battery tech.

[paulca]

If you don't know what you are doing, I recommend you do neither kind of a system. Use something off-the-shelf that everybody uses already; it's a good way to hobby around, but having used a battery analysis toy from HobbyKing to measure a Chinese toy battery from the same place doesn't magically make you an expert on battery tech.

So if I show you batteries out of balance I would clearly be making that information up and we are both hallucinating as clearly they don't go out of balance in your world.  Laptop packs don't fail either because of cell imbalance. 

Also, "Hobby" batteries are most often used in different ways, surprisingly far more intensely.  You show me a power drill that will pull 100A, 200A peak and drain it's battery pack in under 7 minutes.  Then show me the person who expects that pack, cooled, charged and flying in the air gain in under 30 minutes.

I'm sure that if you buy expensive matched, low resistance, batch matched cells for £4 each and you charge them at low rates and discharge them at low rates and seal them in a can preventing anyone accessing them, they MIGHT remain in balance for years.  However not all of us live in the world where we want to spend £65 on a new battery pack.

On your advice I will now use "CHARGE" mode at 2C rate and not "BALANCE CHARGE" mode when I charge my LiPos.  As you state I am completely safe doing this.  So I will.  I assume it's okay to do this while my daughter sleeps upstairs and leave them unattended, yes?

Also, we had the conversation before on LiIon versis LiPo, we agreed I think that while it has nothing to do with the chemistry, that marketing and branded has polluted that completely (LiPo just means pouch cell these days) and there is little to now way to tell, asides a datasheet what the actual chemistry is in both cases.  In consumer grade batteries, typically LiPo = High current, fast chargeable pouch cell.  LiIon is generally a low current, hard cased, slow charge battery.  Neither really are those chemistries/technologies anymore, but they definitely NOT the same type of cell either.

[Siwastaja]

but I'm still not sure how a pack can remain balanced when cells have different capacities

They will be balanced only at one point - typically at "top". Everywhere else, they'll be out of balance, until you are at top again - necessitating per-cell* cutoffs for charge (OVP) and discharge (UVP). But they do remain balanced (at that same point) because same current runs through every cell, and they all charge and discharge the same number of Coulombs. (Coulombic efficiency being around 99.99%, with little difference between the cells.)

So when we talk about "balance", what we really mean is balanced at one point.

*Cell as in series element. Multiple parallel cells count as a single cell

Now, to came back to your question about redistributive balancing; if you really have a pack where capacity differences are large, you would (theoretically) want "dynamic balancing" - balancing current which is high enough to make large balance differences during discharge cycle. I.e.:

Cell 1: 100Ah
Cell 2: 90Ah
Cell 3: 110Ah

"Normal balancing": all are at 4.20V at the same time. Pack capacity is 90Ah, when Cell2 hits 2.80V (or whatever you use).

"Dynamic redistributive balancing": All are at 4.20V at the same time. While discharging, charge is moved from Cell3 to Cell2. Pack capacity is 100Ah. All cells hit 2.80V at the same time.

Now, unless you discharges are super slow, you'll quickly see you need enormous balancing power to achieve this. Making the cost higher (easily at least around $5-10/cell, even if optimized for cost in masses), also weight (transformer magnetics). So, such systems are mostly of academic interest, they almost never make economic sense; even in second-life applications.

The largest completely repurposed pack I have built was a 85V, 250Ah (around 21kWh) pack for an EV, some cells were damaged by overdischarge (by a redistributive, expensive, broken-by-design BMS, by the way!), meaning I had bigger-than-normal differences on leakages. Still, 20mA balancing current (resistive) was enough to keep the system in balance. In any case, I carefully tried to match the capacities, just by measuring all the cell capacities beforehand and working up combinations where the capacities match as closely as possible. This is the preferred way instead of dynamic redistributive balancing: you'll have much lower-cost system, with less extra weight and complexity. So, instead of 90, 100, 110Ah, you would have 99, 100 and 101Ah, resulting in a 99Ah pack.

If you are building a large second-life pack out of small cells, such matching is fairly easy to do. Remember that if you parallel different cell brands/types/conditions, you risk problems with current sharing, so may need to derate both charge and discharge current.

[Siwastaja]

So if I show you batteries out of balance I would clearly be making that information up and we are both hallucinating as clearly they don't go out of balance in your world.

That's not at all what I wrote; a complete strawman. I have developed several li-ion balancing BMS systems; including the redistributive ones asked by the OP, because, yes, balancing is often needed. Please reread carefully; if you are unable to understand, it will be your issue, not mine. Thank you.

[paulca]

So if I show you batteries out of balance I would clearly be making that information up and we are both hallucinating as clearly they don't go out of balance in your world.

That's not at all what I wrote; a complete strawman. I have developed several li-ion balancing BMS systems; including the redistributive ones asked by the OP, because, yes, balancing is often needed. Please reread carefully; if you are unable to understand, it will be your issue, not mine. Thank you.

It just seemed from how brutually you attacked "hobbiests" on their understanding of lithium batteries that they don't go out of balance.  You seemed to suggest that professionally developed products don't even consider balancing as there is no need to.

I frequently (after a sunny week) run a laptop from RC LiPos charged off a solar panel to use up excess solar power from a battery only system.  They are always balance charged, but as they are unprotected high current packs (I don't know if you can protect packs capable of 100A constant).  I run a LiPo alarm on the pack which cycles through cell voltages so I can literally watch the pack unbalance.  Within about 5 minutes of running the laptop they will go out of balance by a few tens of mV.  By the time the pack is flat the total output voltage is greater than 3.0V per cell, but one cell will set off the alarm as it's below 3.0V per cell.

In fairness the lipo alarm (a voltage sensor, LED display and smoke alarm sounder) is running off cell 1.  But without that they still go out of balance and cells 2 and 3 are not equal either.  Maybe this could come down to cheap hobby lipos.  However I have seen similar tests with expensive LiPo packs from top name brands.  Maybe if you approach LG or Samsung and make a commerical order for exactly matched and batched cells you might get better, I don't know.

Once thing to consider and you might experience this in an EV, when tightly packed into a pack the middle cells are exposed to higher heat and this, I believe changes their internal resistance which under high load will cause imbalance to build up pretty quickly.  Although, if you are dealing with 18650s this is probably less of a concern as their total pack density is poor compared to tightly packed LiPo flat "slab" cells shrink wrapped into packs.

[Siwastaja]

brutually you attacked

Sorry 'bout that, I can see my poor choice of words...

You seemed to suggest that professionally developed products don't even consider balancing as there is no need to.

... very hard to see how you would make this interpretation, though. Your claim, instead, was absolutely clear: balancing is absolutely necessary, always. This is untrue.

I frequently (after a sunny week) run a laptop from RC LiPos charged off a solar panel to use up excess solar power from a battery only system.  They are always balance charged, but as they are unprotected high current packs (I don't know if you can protect packs capable of 100A constant).  I run a LiPo alarm on the pack which cycles through cell voltages so I can literally watch the pack unbalance.  Within about 5 minutes of running the laptop they will go out of balance by a few tens of mV.

I see. You are mixing up two things here (see my previous reply). Having different cell voltages outside the "balance point" (typically top), doesn't mean the battery should be called "unbalanced". Try charging it up without balance charger, and you'll see the voltages match again! It was (top) balanced all the time! Now, do it 1000 times (and/or wait for 5 years) and you may be developing an actual imbalance; for this, balancing is needed. It would never happen during one cycle, unless the battery is dying in your hands. Now, if you started to balance during this discharge, you would be out-of balance from the top, and would need to balance again in opposite direction during charge! This would be a massive waste.

To avoid confusion, it's better to talk about capacity mismatched pack, and not use the term "balance" to describe what you are seeing.

By the time the pack is flat the total output voltage is greater than 3.0V per cell, but one cell will set off the alarm as it's below 3.0V per cell.

.

Yes - necessitating either cell-level monitoring (Low-Voltage Cutoff), or safety margin (i.e., use 3.10V instead of manufacturer-specified 2.50V or whatever). The smaller the number of cells in series, the more a single cell contributes to the total voltage; hence, 2s untapped packs are very common, and over 6s, all packs tend to use cell-level monitoring (highly recommended unless you have a very good reason not to). Neither case still requires balancing.

Once thing to consider and you might experience this in an EV, when tightly packed into a pack the middle cells are exposed to higher heat and this

This is true. Higher temperature increases leakage (self-discharge) current, causing slow imbalancing. Large systems are difficult to thermally couple, again necessitating balancing. Luckily, large packs are expensive anyway, and tend to have some complexity in any case, so adding a simple cell-level management system with simple resistive balancing tends to become a no-brainer.

Note that still this is a phenomenon that happens slowly (during years) if not handled. Don't mix it up with varying voltages during a cycle; it will be back "in balance" again, even without balancing. In fact, trying to balance "mid-cycle" causes imbalance; hence balancers typically work with full battery, during the end of charge cycle, or, if they are smart enough (such as one of my BMS design was), they sample the balance at the end of charge, then can do the actual dissipation work at any state. This was my trick of reducing the balancing current from the 500 mA of the competitors down to 40mA, yet still do the same work (or even more), making thermal design and safety easier to achieve. Yes, I saw a report of stuck-on 500mA balancer resistor burning a car. I was actually able to fuse the cell connections with 100mA fuses.

[paulca]

I my case my LiPos are all high current discharge packs with no cell protection.  I'm not sure you can get cell protection for a 5Ah 30C cell capable of 150A, I'm not sure how you would switch that kind of current without heavy bulky mosfets.   I have seen people take the LiPo alarms I have and disconnect the sounder and connect a relay, but that's also bulky and heavy.

In their real application the brushless motor speed controller connects to the balance lead and watches cell voltages.  When one falls under 3.30V (normally, but configurable) it starts a "soft shutdown" in that it limits your power progressively.  This usually gives you 30 seconds to a minute to make a controlled landing.  Hard cut off happens at a lower voltage and forces you to do a dead stick landing which can be messy with a plane and very difficult and usually fatal with a heli.

If I was to try and test "top balance" I don't think my charger would permit it.  In "CHARGE" mode with no balancing it will charge until a single cell his 4.20V and then end the charge, regardless of the other cell voltages.  I could I suppose try this and if it is top balanced, the others will be with in 10mV or so.    Obviously not on the pack running the cell monitor from cell 1.

[OM222O]

so what I'm getting at here is to test the capacity of each cell, then make the packs so they are as closely matched as possible, then balance at the full charge point (4.2V) and actively monitor cells to protect against over discharge 

as you can see from my first post, my main goal was to reduce the waste as well, so can you please share how you managed to reduce the balancing current using resistive balancing?

[Siwastaja]

so what I'm getting at here is to test the capacity of each cell, then make the packs so they are as closely matched as possible, then balance at the full charge point (4.2V) and actively monitor cells to protect against over discharge 

Yes, matched for same capacity, then balanced so that voltage matches at 4.20V. It won't match perfectly at low end, hence the cell-by-cell low voltage cutoff.

So, monitor each cell and stop discharging when any hits LVC (I'd recommend 3.00V as a baseline). Charging is a bit different and depends how well you can control the charger. If you can control the current, do a proper CV phase based on the highest cell voltage - dynamically adjust current to maintain the highest voltage at 4.20V, stop when your current tapers off (C/20, for example). If you can't do that (i.e., if your charger has fixed CV output voltage), just let the charger do the CV phase based on total pack voltage, and use the cell-level OVP as an extra safety measure, but you'll have to adjust it a bit higher (say, 4.25V) so that it won't cut off prematurely, preventing full CV charge.

If you don't need 100% charge, just skip the CV phase completely, and you'll have a battery at around 90% charged, and a simplified procedure.

as you can see from my first post, my main goal was to reduce the waste as well, so can you please share how you managed to reduce the balancing current using resistive balancing?

By using longer balancing time compared to the competitors. Instead of dissipating for a a few minutes during the end of the CV phase, like most products do, I just measured the voltages at the end, calculated required shunting times per cell based on imbalance voltage measured - this could be hours, and the balancing could run while discharging, or doing whatever you do. The required amount of balancing charge is the same, but during a longer time, with lower current. This makes sense, because very seldom you need a lot of balancing at all.

[OM222O]

oh! so it doesn't actually reduce the waste, just makes it over a longer time (presumably for heat dissipation factors ? and safety like the fuses you mentioned?)
by that definition I would just choose the normal method (higher current over shorter time) if the balancing is not required that often (lets say once a month or something) as it does not produce a huge amount of waste 

I watched the video about the "DIY Powerwall" that was posted and it leaves a lot to be desired (using cheap pre made BMS? 7 segment leds for voltage indication?) and it's pretty rough around the edges but I had a similar idea so might take some inspiration. My end goal is to have my bench running all on DC from this sort of DIY power wall so there won't be massive current draw (100s of amps or so) and they won't be abused, maybe something like 20A would be a worst case scenario (for charging super caps  ). currently I'm just using switch mode PSUs that you can buy online but they are a bit too noisy for some of the analog projects  it might end up becoming a modular system like he proposes but that's later down the line. it'll also save me quite a bit as I can run it from solar and have free electricity in my work area. Thanks everyone for their advice.