How to Parallel Balancing. (YEP 99% of us is doing it wrong)(PART#1)

[upnorthandpersonal]

If it's that cold, how much sunshine is actually available?

Quite a bit actually. Cold winters tend to have a lot of sun. Think of places like Alaska etc.

 

[SolarRat]

how is it very very low, but still above BMS trigger point? Isn't this a contradiction?

Not at all. One BMS might cut off all discharge at 2.2v, another at 2.6v, another at 2.8v, and others might not shut off at all at the cell level. If I only want to cycle my battery down to say 3.1v, those BMS's would be useless to stop a single cell from going much lower. And if running an inverter straight from the battery, bypassing the BMS (which is common here), the BMS won't do squat to protect the one low cell. Any way you look at it, a battery that frequently is drawn down to low levels is probably safer with a bottom balance.

 

[electric]

Not at all. One BMS might cut off all discharge at 2.2v, another at 2.6v, another at 2.8v, and others might not shut off at all at the cell level. If I only want to cycle my battery down to say 3.1v, those BMS's would be useless to stop a single cell from going much lower. And if running an inverter straight from the battery, bypassing the BMS (which is common here), the BMS won't do squat to protect the one low cell. Any way you look at it, a battery that frequently is drawn down to low levels is probably safer with a bottom balance.

A guy from this post would beg to differ. Such approach creates false sense of security and often leads to completely destroyed battery. Not saying it's not possible to manage it your way, but it seems to me that an actual good BMS that does not need bypassing would be a simpler and more reliable solution regardless of how the battery is used.
Those hypothetical "BMSs" you described above cannot be called a "good BMS", so how does it help any argument? Just don't buy crappy BMS, get a good one that fits your application. BMS must be able to disconnect the battery before any one cell goes outside of safe range. If a BMS does not meet this simple description, then it's not a real BMS, so you don't need to work around it, just don't buy it in the first place.
With all that said, I totally agree with you on small LiPo packs used in RC world. This application is better off without a BMS onboard, just set conservative limits on your ESC and only use BMS integrated in the charger to prevent overcharge. If your expensive plane or chopper or drone is at stake, sacrificing a battery is a much better choice than crashing into the ground.
I also have a rule of thumb, if BMS costs more than 20% of your battery cost, consider a simpler BMS or a larger battery.

 

[ghostwriter66]

Quite a bit actually. Cold winters tend to have a lot of sun. Think of places like Alaska etc.

I thought Winters in Alaska had very little Sun?? I was in Anchorage one time and YES it was very COLD but we literally had 45 minutes of daylight ... well maybe a few minutes more ... but we all know what you are saying ... The colder it is the more amps the panels cranks out ...

 

[SolarRat]

A guy from this post would beg to differ. Such approach creates false sense of security and often leads to completely destroyed battery. Not saying it's not possible to manage it your way, but it seems to me that an actual good BMS that does not need bypassing would be a simpler and more reliable solution regardless of how the battery is used.
Those hypothetical "BMSs" you described above cannot be called a "good BMS", so how does it help any argument? Just don't buy crappy BMS, get a good one that fits your application. BMS must be able to disconnect the battery before any one cell goes outside of safe range. If a BMS does not meet this simple description, then it's not a real BMS, so you don't need to work around it, just don't buy it in the first place.
With all that said, I totally agree with you on small LiPo packs used in RC world. This application is better off without a BMS onboard, just set conservative limits on your ESC and only use BMS integrated in the charger to prevent overcharge. If your expensive plane or chopper or drone is at stake, sacrificing a battery is a much better choice than crashing into the ground.
I also have a rule of thumb, if BMS costs more than 20% of your battery cost, consider a simpler BMS or a larger battery.

I never said it's smart to bypass the BMS for high loads, just that it is VERY common. Not all of us work at the largest battery manufacturer in the country and have access to their high dollar stuff. Not all systems are perfect, nor do some have to be. Even with the perfect BMS, why, in the scenarios I describe, would you want your cells balanced at the top rather than the bottom? You'd be leaving a lot of capacity on the table if top balanced older batteries are used, and it will get worse and worse until they are bottom balanced. I've had most of my lithium batteries for between 5 and 12 years...and I don't believe the "good cells will stay in balance" crap. With new good cells, for the first few years they will, but older batteries can and will drift. Especially when one or two cells are constantly in the lower knees when the rest are not.

A BMS is a failsafe, NOT a charge controller. Even with a great BMS it's best to have it intervening as little as possible, which means wisely choosing whether to top or bottom balance for the specific application.

 

[GVSolar]

I'm not sure if that's a good thing or a bad thing...

"he was among the first to emphasize scientific accuracy in his fiction, and was thus a pioneer of the subgenre of hard science fiction. "

Exactly my point. But the discussion rocks.

 

[electric]

Even with the perfect BMS, why, in the scenarios I describe, would you want your cells balanced at the top rather than the bottom? You'd be leaving a lot of capacity on the table if top balanced older batteries are used, and it will get worse and worse until they are bottom balanced.

Regardless of top or bottom balance your usable battery capacity is limited by the smallest/weakest cell. Whichever balance strategy is used, you still line up your charge and discharge by the bookends of the weakest cell, so how am I leaving anything on the table? My argument against bottom balance is that people still attach a BMS to the battery and most commercial BMSs are top balancing, so from day one your BMS will work against your initial bottom balance, making it useless. Bottom balance is only good when you run without a BMS, even though you are likely to kill your smallest cell by being oblivious to its voltage shoot at the end of charge, assuming that by lowering your charge target you somehow protected the first cell from grossly overshooting at the top. Even new cells come with 1-2AH spread, which is enough to damage the smallest cell while others are barely reaching the upper knee.

 

[nebster]

You don't charge to 3.4V, it is not practical, you'd be wasting most of your battery investment.

I agree with most of what you've written in these threads, but this is simply not true in my experience. You can slowly charge LFP to 96 or 97% SOC at 3.40Vpc, any day of the week. (And it might well take all day, because the rate will be quite low at that CV.) I don't think it's fair to say that that is "wasting most of your battery investment."

 

[nebster]

assuming that by lowering your charge target you somehow protected the first cell from grossly overshooting at the top. Even new cells come with 1-2AH spread, which is enough to damage the smallest cell while others are barely reaching the upper knee.

I'm going to disagree, again, here. I use this exact strategy. My smallest cells (which are 3Ah smaller than the largest) are not being damaged. Have you actually tried this? Please share your actual data... because I have mine, and I'm curious what you're doing differently (and whether I have made a mistake).

The only thing I can think of is that you are assuming that someone wants to run up to 100% SOC. But since that itself is detrimental to longevity, that would be silly with this approach. Instead, we just oversize the pack a little bit and live simply by stopping at 90%.

 

[nebster]

Can't speak for everyone, but I personally don't comment on "WHAT IF" posts describing something that is not a real thing or something that never happens in practical applications. Life is too short to engage in useless pondering.

While I agree that it is unlikely that many DIY packs today will use a BMS with this strategy, there definitely exist BMSes that actively rebalance while the pack is in every state (charge, discharge, at rest). Tesla uses this approach. It has a certain degree of elegance to it, if balancing is required... which I think we agree, it is not for ESS applications.

 

[electric]

Tesla uses this approach.

Do you have a source I can look into? I have analyzed Tesla BMS boards taken from a wrecked car and did not see any evidence of active balancing. All I saw is a classic top balance scheme with tiny amount of balancing, which is all you really need when using top quality perfectly matched cells.

Active rebalance just doesn't make any practical sense to me even if done properly, which these cheapo BMSs can't possibly do with their horrible voltage sensing inaccuracy.
Think about it, active balance schemes are all based on reducing voltage differential, as if that is the goal. When battery is actively used, voltage differential can be perfectly normal due to spread in IR between adjacent cells. Voltage differential does not mean difference in SOC, so if you are trying to equalize voltage, you are actually hurting the SOC balance, which should be your real goal. As such true SOC balance, either top or bottom, can only be achieved at the vertical knees of the curve, where relationship between SOC and voltage is known. Anywhere in between this relationship is moot and so you can't use voltage as a measuring stick.

 

[electric]

Instead, we just oversize the pack a little bit and live simply by stopping at 90%.

How can you tell you are stopping at 90%? Does your charger know the SOC of your smallest cell? How can you set your charger's pack voltage precisely matching your weakest cell's voltage corresponding to it's 90% SOC? You'd have to experiment by carefully monitoring end of charge and stopping when your smallest cell reached the desired voltage, then recording the pack voltage at that moment and setting your charger to that level, so next 1000 charge cycles will end precisely the same level relative to that particular cell? This implies you even have a charger where you can control the target voltage with such precision, if at all. Not many chargers out there with flexible voltage settings.
Yes, all of this can be done, but why bother with all this complexity when you can just slap on a decent BMS, pick any decent charger, even if you can't tweak it's voltage, and move on to the next project without ever worrying about any drift or imbalance.

 

[electric]

You can slowly charge LFP to 96 or 97% SOC at 3.40Vpc

Yes, you can. But I can't as I value time more. The best quality of LFP is ability to accept fast charge, I don't want to give it up. I design all my batteries to be able to charge fast, so people can take advantage of short solar hours or engine runs, reduce generator runtime, etc. Everyone wants a fast charge. This is why I said not practical, but not impossible.

 

[nebster]

Do you have a source I can look into? I have analyzed Tesla BMS boards taken from a wrecked car and did not see any evidence of active balancing. All I saw is a classic top balance scheme with tiny amount of balancing, which is all you really need when using top quality perfectly matched cells.

You can find a detailed discussion of the balancing strategy on the Tesla forum. It is pretty simple: a computer watches the cell's voltage rise in the knee and estimates a quantum of energy that needs to be dumped. That value is written into a timing circuit that proceeds to slowly bleed off that energy, possibly over many hours or even days.

This means that the "active" aspect of the design is independent of the cycling of the pack. It also means you need independent bleeders and timers for each discretely-managed cell or block of cells.

Think about it, active balance schemes are all based on reducing voltage differential, as if that is the goal. When battery is actively used, voltage differential can be perfectly normal due to spread in IR between adjacent cells. Voltage differential does not mean difference in SOC, so if you are trying to equalize voltage, you are actually hurting the SOC balance, which should be your real goal. As such true SOC balance, either top or bottom, can only be achieved at the vertical knees of the curve, where relationship between SOC and voltage is known. Anywhere in between this relationship is moot and so you can't use voltage as a measuring stick.

I'm no fan of active balancing, but my understanding is that the "good ones" do make their balancing efforts at the very steep slope at the top of the charge curve. That that would require those batteries to be held in that regime is reason enough for me not to be excited about this strategy.

To your point, I'm not sure how to differentiate between internal resistance and SOC. To some degree -- and I'm not an electrochemist, so I'm probably bludgeoning this -- I think Ri and capacity must be interrelated. A deterioration in one seems like it must present as the same falloff in the other, over time.

 

[nebster]

How can you tell you are stopping at 90%?

I did it the "easy" way, by empirical observation. I gradually increased the CV threshold, a few millivolts at a time, while watching the weakest cell in each string. It turns out I was able to get to about 92% of rated capacity without the weakest cell diverging enough to be an issue. At that point I was at my design goal (which was itself a bit arbitrary -- seemed like 90% was totally safe, and >95% was unknown but possibly deleterious over the long term).

Does your charger know the SOC of your smallest cell?

No, it is dumb. All it can do is regulate the voltage at the battery busbars.

How can you set your charger's pack voltage precisely matching your weakest cell's voltage corresponding to it's 90% SOC?

The 92% is a pack SOC expressed a percentage of the nameplate storage. My cells are 100Ah on the stick but between 102 and 107Ah in real life. So the total pack is storing 92Ah beyond a stable bottom 2.75Vpc when I stop the charge.

This implies you even have a charger where you can control the target voltage with such precision, if at all. Not many chargers out there with flexible voltage settings.

I'm not sure how repeatable the charger is in terms of its measurements, but monitoring stopping voltages over a long time, it looks like it's getting very close each time. Definitely within +/-5mV on a 16s battery.

Yes, all of this can be done, but why bother with all this complexity when you can just slap on a decent BMS, pick any decent charger, even if you can't tweak it's voltage, and move on to the next project without ever worrying about any drift or imbalance.

That's not a bad argument, but my particular pack is very large and cumbersome, and it's actually made up of many discrete, paralleled strings, so the BMS cost would be pretty high. I also don't like the BMS wiring complexity (adding sense wires). Stopping early is pretty easy, and it only took a handful of test charges to determine that I could safely hit the 92% with the balance my cells arrived with. I'm certain that wiring multiple hundreds of voltage sensing wires would have been more work in my scenario.

 

[nebster]

Yes, you can. But I can't as I value time more. The best quality of LFP is ability to accept fast charge, I don't want to give it up. I design all my batteries to be able to charge fast, so people can take advantage of short solar hours or engine runs, reduce generator runtime, etc. Everyone wants a fast charge. This is why I said not practical, but not impossible.

I don't mean to sound like I'm quibbling with your terminology. But I have lived full time off of my battery for 2.5 years now, and I have a lot of incentive to charge fast, to not run my generator longer than I need to, and to harvest all the solar I can get. And those desires are not incompatible with a conservative CV threshold.

Here is a very typical graph from a recent charge. This charge began at about 55% SOC and delivered energy at about 0.16C:



You can see the charger was in CC until about 7pm, after about 2.5h. The labels show that about 18.4kWh was shoveled into the battery in total. For the last 45 minutes of CV, the total energy delivered there was about 2.5kWh, or 13% of the total but using 25% of the time. That last bit of charging was only 50% as time-efficient.

So, yeah, that kinda sucks? Except:

(a) I could always just decide to stop at the CV threshold and skip that last sliver if I really hated listening to the genset. That "wasted" bit is only 6% of pack capacity.

(b) Even with a BMS and a top-balance, one still hits a CV regime -- and the higher the rate, the earlier. As soon as the charger shifts into CV, the characteristic curve should look similar to the one above... and thus, there is "waste."

(c) In many scenarios, I have unlimited low-level charging available. That occurs when on shore power, or on sunny days when there is excess PV. In those cases, the very small CV portion is not wasted, because there is no opportunity cost for that energy.

I respectfully submit that conservative stop-charging voltages and a slightly-oversized pack can produce a pretty darn good, practical, workable result. I'm right here living in proof of it!

 

[ghostwriter66]

I don't mean to sound like I'm quibbling with your terminology. But I have lived full time off of my battery for 2.5 years now, and I have a lot of incentive to charge fast, to not run my generator longer than I need to, and to harvest all the solar I can get. And those desires are not incompatible with a conservative CV threshold.

Here is a very typical graph from a recent charge. This charge began at about 55% SOC and delivered energy at about 0.16C:

View attachment 5999

You can see the charger was in CC until about 7pm, after about 2.5h. The labels show that about 18.4kWh was shoveled into the battery in total. For the last 45 minutes of CV, the total energy delivered there was about 2.5kWh, or 13% of the total but using 25% of the time. That last bit of charging was only 50% as time-efficient.

So, yeah, that kinda sucks? Except:

(a) I could always just decide to stop at the CV threshold and skip that last sliver if I really hated listening to the genset. That "wasted" bit is only 6% of pack capacity.

(b) Even with a BMS and a top-balance, one still hits a CV regime -- and the higher the rate, the earlier. As soon as the charger shifts into CV, the characteristic curve should look similar to the one above... and thus, there is "waste."

(c) In many scenarios, I have unlimited low-level charging available. That occurs when on shore power, or on sunny days when there is excess PV. In those cases, the very small CV portion is not wasted, because there is no opportunity cost for that energy.

I respectfully submit that conservative stop-charging voltages and a slightly-oversized pack can produce a pretty darn good, practical, workable result. I'm right here living in proof of it!

Are you running solar for you home, RV, work?? Just curious -- your knowledge is outstanding and the way you explain things is really understandable ...

 

[electric]

@nebster my hat is off to you, you did a wonderful job of a highly specific and customized system that perfectly fits your needs. But you cannot argue that it's practical for most DIY projects? I bet you have at least some EE education and experience to pull this off.
I am ranting on forums because I see plenty of Youtube videos and forum posts where people follow some half baked advice to bottom balance their cells and then proceed to charge them at a max pack voltage, oblivious to some cells shooting thru 4V and beyond, because they never even monitor on cell level, and then after a while when the pack is damaged they blame getting used cells or Chinese conspiracy, etc.
Also, you have a 16S battery, so tweaking charge voltage is easier. Much harder on most common 12V systems, especially when alternators are used for charging, which are terrible at voltage regulation.
I'm curious why many parallel strings? Is it due to physical space layout? Or you don't trust parallel cells in a single series string?

From my experience, I prefer to charge to 3.5V-3.55V to get a good trade-off of fast enough charge, yet still plenty of charge cycles. Calendar aging and temperature will do them in before cycle life anyway. Oversize AH as much as reasonably possible to reduce average DOD. Not a fan of bottom balance because I always use a BMS and find top balancing just so much easier to do.

Thanks for sharing your project, much appreciated. Sorry I can't share much of mine due to NDAs.
I do have a 10kWh home battery setup in a fully automated peak shaving cycle, shifting 10kWh every day from night hours to peak day hours. Has been cycling for a year now, more than 3 MWh time shifted by now. Used CALB cells rescued from previous EV application. Cells are 5 years old, 4 years in EV and one year in daily 90% DOD cycle, still running strong at about 90% of original sticker rating.

 

[nebster]

Are you running solar for you home, RV, work?? Just curious -- your knowledge is outstanding and the way you explain things is really understandable ...

You're very kind. I live full time on the road in the RV. Solar is limited by the roof space, but we have just enough to cover our baseline when the weather is nice and it is not winter.

 

[nebster]

@nebster my hat is off to you, you did a wonderful job of a highly specific and customized system that perfectly fits your needs. But you cannot argue that it's practical for most DIY projects?

I think a simple BMS and a 4s pack are much more practical for a lot of projects. But it can be instructive for any DIYer to understand the range of choices, even if it's likely they'll choose one of the simpler paths.

I bet you have at least some EE education and experience to pull this off.

It just became a hobby. I'm a software guy. But they did make me go to physics in college...

I am ranting on forums because I see plenty of Youtube videos and forum posts where people follow some half baked advice to bottom balance their cells and then proceed to charge them at a max pack voltage, oblivious to some cells shooting thru 4V and beyond, because they never even monitor on cell level, and then after a while when the pack is damaged they blame getting used cells or Chinese conspiracy, etc.

I completely understand and agree.

Also, you have a 16S battery, so tweaking charge voltage is easier. Much harder on most common 12V systems, especially when alternators are used for charging, which are terrible at voltage regulation.

This is a good point -- the 4x extra, "free" resolution is really nice. And we have no alternator-based charge sources, with which I have very little experience.

I'm curious why many parallel strings? Is it due to physical space layout? Or you don't trust parallel cells in a single series string?

Great question -- it is a mix of reasons. The main one is that this pack is a test run for a liveaboard boat one day, and on the boat I want redundancy so that a cell can take a string out but we can maintain house power. The only reasonable way to do that is to have relatively high parallelism (because the load will redistribute and blow fuses if n = 2, for example).

But, as it turned out, in the space I have there was no straightforward way to wire 7p and then 16s. I just couldn't make things fit and have a clean wiring topology.

I was originally concerned about inter-cell fusing, because certainly in theory this is the tradeoff for doing parallel-first construction. Only by reading, and not by experimentation, I have since decided that I don't think that's a very likely failure mode with prismatics. I think I would be okay with a parallel-first design in a future battery, the above requirement for redundancy notwithstanding. (What I would really prefer is fewer, higher-capacity cells. I didn't have a good option for those when I built the pack 3 years ago.)

From my experience, I prefer to charge to 3.5V-3.55V to get a good trade-off of fast enough charge, yet still plenty of charge cycles. Calendar aging and temperature will do them in before cycle life anyway. Oversize AH as much as reasonably possible to reduce average DOD. Not a fan of bottom balance because I always use a BMS and find top balancing just so much easier to do.

Seems reasonable enough. I was prepared to install 7 BMSes, but after testing, it just turns out they're not really necessary in a healthy, relatively new pack that's carefully managed.

Thanks for sharing your project, much appreciated. Sorry I can't share much of mine due to NDAs.

I hope when you all get something to market you can tell us more. You clearly have spent a lot of time with this chemistry.

I do have a 10kWh home battery setup in a fully automated peak shaving cycle, shifting 10kWh every day from night hours to peak day hours. Has been cycling for a year now, more than 3 MWh time shifted by now. Used CALB cells rescued from previous EV application. Cells are 5 years old, 4 years in EV and one year in daily 90% DOD cycle, still running strong at about 90% of original sticker rating.

Nice. If the next owner of this RV doesn't want it, I plan to repurpose this pack for something similar one day. (I confess: I built this pack in my garage first, and I did wire the whole thing straight into the panel and run our previous house off of it for some of the load testing!)