Lithium Top or Bottom Balance

[Sunking]

An excellent idea... forces the choice of an appropriate charger as well and most importantly, the very careful use of that charger to perform the balance charging.

Still a very expensive experiment as a 4S 40 AH battery will cost you $250 to $300 when you can get the same FLA for $100.

[sahucker]

T
As always, instead of just typing, a simple 12v / 40-60ah 4S pack can be built from GBS, CALB what have you for hands-on before spending any big $$.

An excellent idea... forces the choice of an appropriate charger as well and most importantly, the very careful use of that charger to perform the balance charging.

[PNjunction]

The one big downside of lithium batteries, is aging is accelerated by heat, cold, high charge/discharge rates, over charging, and the big killer is over discharging. All it takes is one over discharge and you have a brick.

Don't forget TIME. You can discharge a lifepo4 dangerously low, but ** provided you get to it in time - like asap and not a few days later *** , you can recover IF you recharge at less than C/50 until the cell reaches a normal voltage (3v or so) whereupon you can apply normal current. Basically if you apply too much current when the cell is super-low in voltage, ie below the normal lvd levels, a traffic-jam occurs since you are trying to force too much lithium through the SEI layer too fast for it to deal with (think a soccer crowd trying to go through some tiny entry gates at the world cup.) Probably not the best or most accurate analogy here. Prof Dahn would probably groan, but it's the best I've got.

In the end, most let this bad discharge condition go on for too long, and/or may apply too much initial current for a sane recovery. Not that you've done the cell any favors, so kids don't think this is the norm.

One of the simplest and most effective ways to extend cycle life is to only use 80% of the batteries capacity, an dnever fully charge or discharge them. So with a 100 AH battery we want to only use 80 AH.

Yep - that's why I instantly DE-rate the labeled capacity to 80% for capacity power budget calcs. Seems to work fine. Those that expect the rated capacity usually find themselves at the lvd of 2.5v or so, which is too far into the discharge curve for *regular* cycling. In our storage application, typically in a "Sub-C" environment, then about 3.15v or so is around the 80% DOD level. If something unforeseen should happen, then the normal LVD is there - but like a Pb based LVD, that is a dead-man's catch and should not be a cheap way of automating your DOD.

Also, being a bit conservative is insurance against the unforeseen - manufacturing tolerances, system monitoring voltage calibration / drift, and accidental environmental situations, like losing air conditioning in your battery vault etc. Nothing like walking into your vault and finding that it has been cooking gear like that for weeks - running lifepo4 conservatively gives you a bit of a hedge against full-charge / high ambient temp cooking -- although high heat is NEVER a good thing which can be waved off with conservatism. It just might buy you a little more time that's all.

[PNjunction]

The path of learning involves being able to repeat back the concepts in their own words, not as a parrot. I'm am grateful for feedback this forum provides.

That is very wise. I often try to gauge my own understanding by trying to rephrase it in my own words and still be accurate and understandable enough for my friends who aren't into batteries per se, yet hold up to the scrutiny of peers. That is VERY hard to do and on a forum one takes the risk of "typing oneself smart".

I like your analogy since you have taken into account real-world variables that exist between buckets - which is sometimes missing in various charts and whatnot.

As always, instead of just typing, a simple 12v / 40-60ah 4S pack can be built from GBS, CALB what have you for hands-on before spending any big $$.

A 4S system like I run (in a "sub-c" environment) is easy to bottom balance. Once the initial bottom balance has been achieved (covered elsewhere), then for only 4 cells, you may get away with pack-monitoring. Initially, charge the pack, and keep checking the voltages from cell to cell to identify the first one to reach 3.6v. STOP. Then quickly take note of the total pack voltage as your normal stopping point - or set your charger to stop a bit lower if desired. If the cells are quality and reasonably the same in both capacity and IR, then the cells should be close. If any cell is less than 3.45v when one has reached 3.6v, you've got some issues. They should be a bit closer than that. NOTE - this is assuming you are charging higher than .05C, otherwise you may never reach 3.6v!

Because I can get away with it with only a 4S pack, and since I'm running "sub-c", I conveniently top-balance or top-charge. I still do it pack-level, but I initially brought all cells together at the top, and since I'm conservative and not pulling EV-like current from acceleration, even a pack-level monitoring to stop at around 12.75 total works. My system has TIME in a sub-c application to operate that way without risking a very quick cell reversal since no cell is taking a precipitous dive at relatively low current. It is NOT the most accurate way to do it, but since I operate conservatively, I get away with it.

High-voltage house-bank setups would probably do best with bottom-balance in my opinion, even in a sub-c environment, but if you are a battery cowboy like I am with relatively inexpensive setups, then top-balance if your wallet is open.

[Sunking]

Any flaws in a Lithium battery will come to surface within a few cycles which is a good thing as you get a free replacement early in the battery life.

The construction method of large prismatic cells is no tan exact machining process. If you open one up it basically looks like a large sheet of metallic plastic film with a gooey gel. Toleranc range is a bit loose, but the error favors the customer. If you buy a 100 AH cell you are guaranteed the minimum is 100 AH. but the real range is on the order of 100 to 115 AH. Again a good thing.

The one big downside of lithium batteries, is aging is accelerated by heat, cold, high charge/discharge rates, over charging, and the big killer is over discharging. All it takes is one over discharge and you have a brick. On the plus side when a lithium cell fails, it fails shorted. So in a EV of say 96S if a cell fails shorted means you get to at least get somewhere in Limp Home Mode rather than stranded. However with some lithium chemistry as Tesla owners have found out a shorted cell forced to limp home likes to catch fire.

So what can you control? Depends on who is making the EV. Manufactures use highly automated BMS systems with thermal management of heating and cooling. One of the reasons they do this is because they are using High Energy Density types like Cobalt which are very finicky and just out right dangerous. They have to use high Energy Density Kg/wh and high w/Kg to get the range up to something acceptable to the public. But enough of the manufactured products.

It would be very foolish for someone to use high density lithium batteries for solar or a DIY EV. They cost way too much, dangerous, and require a lot of automation like BMS and thermal management, If you use LiFeP04 they are the safest most stable, longest life, and the lowest cost per watt hour. They are stable enough no BMS is required if you Bottom Balance thus keeping cost low. Many and I do mean many DIY EV typres have learned BMS systems are the root cause of all failures. In other words BMS is the problem, not the solution.

So now to your question of aging. One of the simplest and most effective ways to extend cycle life is to only use 80% of the batteries capacity, an dnever fully charge or discharge them. So with a 100 AH battery we want to only use 80 AH.

OK if you Bottom Balance we set two know reference points. We establish 0% SOC where every battery is at 2.5 volts or exactly 40 volts on a 16S pack. (16 cells in series) We also have established the know capacity of each cell at 0 AH. So to charge and stay below 100% SOC we only pump in 90 AH. Every cell with have 90 AH from the weakest cell of 101 AH to the strongest cell of 115 AH. The voltages of each cell will not be equal, but is not important and irrelevant. The weakest cell will have the highest voltage of around 3.4 volts at 90% SOC and all the others slightly lower SOC. So when charged we have achieved of not charging any one cell more than 90% to extend cycle life. On the discharge side we just monitor pack level voltage and set the LVD to 48 volts or roughly 10% SOC where at 10% SOC the cell voltages will all be roughly equal to 3 volts well above the danger of 2.5 volts per cell.

Result minimize risk to ever over charging or over discharging, and have not spent a dime on a expensive BMS that likely will cause a failure.

[sahucker]

FLP bottom balance discussions

Why? Bottom balance is essentially a "once and done" thing, although checking occasionally that they all do reach your LVD at the same time with a discharge would be advisable.

The 'once and done' is one aspect that makes this method look very attractive in trade for some effort and time to get it balanced at the beginning. As for why, I'm thinking about the end of the life. Let me digress a bit, as forum members span a range of backgrounds. I work in manufacturing where Mr. Normal Distribution is alive and well. Nothing manufactured is exactly the same dimensions, but, all the good parts are within the product tolerances. In addition, some components have features (dimensions) that are special. Let's call them key product characteristics (auto industry). For those very few dimensions, the customer will perceive a benefit if that feature is 'centered' within the tolerance instead of just being "in range".

I've wondered (no personal data in the battery world) if a battery bank reaches the end of it's life because the 16 (or many more) cells cyclically age differently from each other. Eventually, that cell which came from the tail of the distribution takes the whole pack out of action. The range of usable DOD for LFP, high cost, and the manual aspect of setting up BB (so the implementer is not a typical user) had me wondering if finding and removing that 'odd cell' early in the life would then extend life of the whole pack enough to make up for the cost spent on the extra cell.

Start with quality, and not gray-market trash. Run good wiring infrastructure (ie, clean and tight to spec - normal stuff). I wouldn't go nuts but *checking* once a year would be prudent I suppose.

I understand and agree.

No. It would only be the odd man out at the top end. In fact, that is the one that if you monitor cells individually, would be the one you use to trigger the charger to stop.

Ah, yes. I had not thought of that. Is it common to have a bank charger that that can tap into a single cell to use that cell's voltage for the control decisions? If not common, do you think a study of the naughty cell's voltage compared with the total bank voltage would give one the bank value that corresponds to the value that you aren't able to tap into?

Marrying a new cell into an existing pack is of course doable, and will get you back up and running, but then you should re-do your bottom balance and determine the weakest of the bunch again anyway, and use that for the trigger to stop the charging if you do individual cell monitoring. But like any battery chemistry, you now have a mix of old and new inventory and will have to keep track of that just in case.

Moral - do it right, and you'll never have to be marrying in new cells anyway - BUT if you can spare it, perhaps keep one loose cell around maintained at 50% DOD just in case - stuff happens.

True.

For the 17th cell... how about this extra credit question/thought... since we would not want the cell bank to be charged > 90%, and if all devices connected to the bank can handle an extra 1.7V (back of scrap paper estimate) when at 90%, then, just store the extra cell in the string. You already need to make changes to settings to the charge controller to implement a standard BB strategy... thoughts? Eventually, one extra cell would be pulled (and settings changed) when the first cell begins to deviate from the pack.



I've been thinking about Sunking's bucket of water analogy. I am thinking that the series of buckets are connected at the bottom from bucket to bucket with short pieces of hose. The bottom balance setup operation is the same as shimming the buckets up and down so that the flow out of each bucket has the same 2.7V of potential. I realize I was letting differences in resistance per cell distract me, forgetting that current has to flow through every cell in the same amount (still thinking about how heat generated complicates the picture). Since the buckets have a range of size from manufacturing variation, the top balance people who have their hoses connected along the top have to deal with bridging the flow over any cell that is too short to let it flow though the bucket to get to the next bucket. The bottom balance avoid the overload (overflow) of any one cell by staying < 90%. Still correct in my understanding?

The path of learning involves being able to repeat back the concepts in their own words, not as a parrot. I'm am grateful for feedback this forum provides.

inMichigan

PS my first thought of buckets looked like the ones from Sorcerer's Apprentice

[Sunking]

If one bottom balanced yearly while recording the voltage differences at both LVD and full charge, then
I expect to find the same battery # being the odd one out at both ends (true?)

False. When you Bottom Balance all batteries will have the same voltage of 2.5 volts, and the same capacity of 0 AH. Think of it as say ten slightly different sized buckets with no water in the, When they are empty, you know how much water is every one of them. When you charge, everyone gets the exact same amount charge, so you know exactly where how much water is in each bucket. The smallest bucket gets filled first and you stop.

In Top Balance the only place where anything is equal is at the TOP at 3.6 volts. But that does not tell you how much capacity you have. Back to the slightly different sized buckets. Everyone of them is completely filled up. You do not know exactly how much charge is in them except 100% SOC. On the discharge side, all drain equally but the smallest bucket is falling faster than the others. As you get near the Bottom you can have a cell or two go below 2.5 volts and never know it because your Inverter or Gizmo LVD only sees Total Pack Voltage and thinks all if fine when in fact your destroying cells from over discharge. At the bottom of a Top Balanced battery voltages are all over the place. You can have some of the stronger cells still up around 3.2 volts and others well below 2.5 volts which destroys them. If you look at the Total Pack Voltage everything looks good when in fact it is not.

[PNjunction]

If one bottom balanced yearly while recording the voltage differences at both LVD and full charge,

Why? Bottom balance is essentially a "once and done" thing, although checking occasionally that they all do reach your LVD at the same time with a discharge would be advisable. Start with quality, and not gray-market trash. Run good wiring infrastructure (ie, clean and tight to spec - normal stuff). I wouldn't go nuts but *checking* once a year would be prudent I suppose.

I expect to find the same battery # being the odd one out at both ends (true?)

No. It would only be the odd man out at the top end. In fact, that is the one that if you monitor cells individually, would be the one you use to trigger the charger to stop.

In the ~3rd year, would the overall health and life of the pack be improved if this 1 ( or 2) battery is replaced with a new one?

Nope. Marrying a new cell into an existing pack is of course doable, and will get you back up and running, but then you should re-do your bottom balance and determine the weakest of the bunch again anyway, and use that for the trigger to stop the charging if you do individual cell monitoring. But like any battery chemistry, you now have a mix of old and new inventory and will have to keep track of that just in case.

Moral - do it right, and you'll never have to be marrying in new cells anyway - BUT if you can spare it, perhaps keep one loose cell around maintained at 50% DOD just in case - stuff happens.

[sahucker]

replacing the odd one out

Hello,

If one bottom balanced yearly while recording the voltage differences at both LVD and full charge, then
I expect to find the same battery # being the odd one out at both ends (true?)
In the ~3rd year, would the overall health and life of the pack be improved if this 1 ( or 2) battery is replaced with a new one?

Thanks

[Sunking]

Maybe not Grandma, but there are wheelchair users who are definitely interested in that kind of power.
Check out http://www.wheelchairdriver.com/ for some very interesting info ranging from ultra-high performance to conservative, hosted by an engineer who knows the details of the target application.

That is some funny stuff. I did not read everything, just some points I was looking for. Not sure a 16 mph wheel chair is safe for any age group, but some of his claims are pretty far fetched like 10 to 15 year battery life. Like everyone else I hope this latest generation of LFP batteries can get 2000 to 3000 cycles, but to date no battery has ever done that in real world application.

[PNjunction]

That engineer's downright honesty and get-to-the-point tactics reminds me of somebody who is also interested in saving time, money, and safe practices for the readers.

Be sure to visit the message board too, and not just the main page. The transition from Pb (mostly Optima, later Odyssey, and now Lifepo4) was watched with fascination over the years, and is quite inspirational.

[inetdog]

I can't see Grandma popping wheelies and burning rubber on her motorized wheel chair.

Maybe not Grandma, but there are wheelchair users who are definitely interested in that kind of power.
Check out http://www.wheelchairdriver.com/ for some very interesting info ranging from ultra-high performance to conservative, hosted by an engineer who knows the details of the target application.

[Sunking]

PN glad you chimed in. In all honesty you are correct the A123 cells are high C-Rate cells, but you understated them and Large Format cells.

The A123 cells are rated for continuous 20C, and 50C for 10 second burst.

CALB on the other hand is rated for continuous 3C, and 10C for 10 second burst.

As for using A123 cells for EV, wheel chairs, or high C-rate applications I am not in full agreement with you as that depends on AH to Controller Current ratio.

As an example my little golf cart hot rod uses a 650 AH controller with 100 AH cells which is more than enough. 650 amps is the peak for acceleration. For continuous the current is 190 amps. So a 100 AH large format is more than adequate for the job. I sort of wish I could use A123, but the expense, complexity of assembling and operation using 540 cells is insane IMO. Think Tesla EV 8000 cells. Heck 16 cells is enough eggs in the basket for me to worry about. 540 is unthinkable.

I am not sure a wheel chairs or ebikes can justify A123 cells or need the cells. A123 main advantages are higher energy density and of course higher C-Rates. I am not sure a E-Bike or wheel chair requires the higher energy density and C-Rates.

Don't most E-bikes use 36 to 48 volt 10 AH batteries? Don't Ebikes use 250 to 400 watt hub motors? A large format would be adequate on a 100 amp controller on a 900 watt hub motor. That would yield up to 4800 watts or 6 peak horse power. I can't see Grandma popping wheelies and burning rubber on her motorized wheel chair.

[PNjunction]

Whether you chose to Top Balance or Bottom balance is a personal choice. As a manufacture I would insist you Top Balance because I make a lot more money initially with the extra equipment needed to pull it off, and and make more money from more frequent battery replacement.

No question about it for EV use - bottom balance is the way to go. I hear rumor that some powertool applications using lifepo4 may have switched to this scheme also - which makes total sense since a consumer is just going to run that high-current tool to the dregs. What is frustrating is identifying the manufacturers who actually do this.

Solar DIY can do this as well, but for our low-current (relatively) needs, you can top balance and at least get in the ballpark if you don't overdo it, and run conservatively. At a comparatively low current discharge rate, your lvd be it cell-based or pack based (provided you start out with quality) is a decision / risk the owner will have to choose.

Ideally I'd be running bottom-balance too, but all my gear is essentially a top-balance environment. Fortunately, I don't balance much with my "sub-C" application!

[PNjunction]

BUT, even at 80% SOC the low cell will give you fewer cycles over its lifetime than the ones which you max out at 75%.

On paper, this is true - the weakest, lowest capacity cell will get worked the hardest. That is one reason when you buy large expensive cells, you do so from a quality dealer who makes sure that all the cells are nearly the same in capacity and internal resistance. You may even get a printout identifying these paramaters matched to the barcode on the cells so you can keep track of this baseline measurement from the factory.

And, internal resistance is also a factor. You may have a cell with higher overall capacity than the rest, but it may have the highest IR of the pack, changing operational paramaters during use a bit. I think at the end of the day, just buy good quality stuff, and not try to go for absolute perfection. Being conservative either top or bottom gives one that option.