Best lithium configuration , parallel or serial first

This is a perennial debate.

I’m building my own BMSes so channel capacity isn’t an issue

On cruiserforum it’s long opined that parallel first serial second is best

But looking at this recently for a future boat install 300AH , 12V

3 x nominal 12 batteries , ie 4s3P

Advantage

Each “ battery “ is standalone each with a bms , and a local safety disconnect

Each cell is monitored directly by a bms there are no paralleled cells.

Each battery can independently disconnect allowing thd whole bank to degrade gracefully. Ie there is no overall safety disconnect

.
( the BMSs are all networked together anyway with common alarms etc. )

Am I overlooking anything obvious here. ?

Goboatingnow said:

Am I overlooking anything obvious here. ?

Click to expand...

Nope.

Series first. parallel second.

What you said, plus:

Typically gives higher current spec due to sum of multiple BMS when using standalone FET BMS
Expedites identification of a rogue cell. Parallel cells mask a failure and require taking the entire system offline to troubleshoot.

sunshine_eggo said:

Nope.

Series first. parallel second.

What you said, plus:

Typically gives higher current spec due to sum of multiple BMS when using standalone FET BMS
Expedites identification of a rogue cell. Parallel cells mask a failure and require taking the entire system offline to troubleshoot.

Click to expand...

Yeah exactly, the extra redundancy is well worth the cost of a couple extra BMSs (since they aren't a huge expense in the building of an LFP pack)...

Also makes monitoring / troubleshooting much easier, since every cell in the system, the user can view the individual cell voltage from within the bluetooth app management (if is a managed BMS), instead of seeing the sum of a group of cells' voltages.

And of course the other thing many here have mentioned before as well, is that cell balancing times are much lower when you don't parallel multiple high-Ah cells on the same group.

sunshine_eggo said:

Nope.

Series first. parallel second.

What you said, plus:

Typically gives higher current spec due to sum of multiple BMS when using standalone FET BMS
Expedites identification of a rogue cell. Parallel cells mask a failure and require taking the entire system offline to troubleshoot.

Click to expand...

Agreed , but there is an advantage to parallel cells , out of balance cells in a parallel configuration , share the load current and the energy extracted is the sum of the constituent cells.
in a series configuration , the bms will kill the whole series battery if one cell hits the LVC point . Leaving energy in the other cells that can’t be extracted

Goboatingnow said:

Agreed , but there is an advantage to parallel cells , out of balance cells in a parallel configuration , share the load current and the energy extracted is the sum of the constituent cells.

Click to expand...

True, but you'll still have an imbalance issue with the other parallel groups, which will manifest as premature charge or discharge termination. IMHO, the benefit doesn't outweigh the hassle of completely taking the bank offline to troubleshoot.

Goboatingnow said:

in a series configuration , the bms will kill the whole series battery if one cell hits the LVC point . Leaving energy in the other cells that can’t be extracted

Click to expand...

True, but with separate BMS (context of my response), you still have an active battery.

Goboatingnow said:

Agreed , but there is an advantage to parallel cells , out of balance cells in a parallel configuration , share the load current and the energy extracted is the sum of the constituent cells.
in a series configuration , the bms will kill the whole series battery if one cell hits the LVC point . Leaving energy in the other cells that can’t be extracted

Click to expand...

I suppose there is some aspect of user preference involved here as well...

But in my experience, my example is in my RV system, having 6 batteries, each with their own BMS, if some power phenomenon occurs on one of them, I just take it out of the loop by turning off its breaker, and if it's on a Friday, I can just wait until Monday if I want, to take a look at it at my leisure and start troubleshooting on it then... Still having my other 5 batteries keeping me going through the weekend.

Then on Monday, I log into the BMS from Xiaoxiang app and see those 4 cells sitting there, each voltage indicating exactly one cell, read the error log message, take appropriate action, reset BMS, close the circuit breaker, and my wife didn't even know a 'power incident' had taken place at all, still on a Netflix marathon or something.

But that's just me, everyone may have their own reasons for why they want one way or another, not saying anyone else needs to do it in accordance with my preference.

Goboatingnow said:

Am I overlooking anything obvious here. ?

Click to expand...

No, on a moblile installation there are several advantages to having several smaller redundant packs.. My 3P16S configuration was driven by the fact that I already had an $800 automotive BMS three years ago and had more confidence in a contactor based solution for the 100 Amp draw I thought I would need. Today there are $150 contactor based BMSs that would have worked for me. My pack is also stationary and fits nicely in a six foot space in my garage below my Inverter and subpanels.

sunshine_eggo said:

.



True, but with separate BMS (context of my response), you still have an active battery.

Click to expand...

I agree, that this is the big advantage of a series first situation , ie you take the system offline battery by battery , leaving you a lower capacity bank but one that is still functioning

Ampster said:

No, on a moblile installation there are several advantages to having several smaller redundant packs.. My 3P16S configuration was driven by the fact that I already had an $800 automotive BMS three years ago and had more confidence in a contactor based solution for the 100 Amp draw I thought I would need. Today there are $150 contactor based BMSs that would have worked for me. My pack is also stationary and fits nicely in a six foot space in my garage below my Inverter and subpanels.

Click to expand...

yes a 3P16s , in efect gives you 3 x 16s batteries that can be shut down separately

Goboatingnow said:

yes a 3P16s , in efect gives you 3 x 16s batteries that can be shut down separately

Click to expand...

Unless I have it backwards my pack is one battery with one BMS. I do not have a mobile system and have a convenient spot below my inverter and subpanel to compress Two rows of 24 cells. I agree there is no best setup except the one driven by user preferences. My backup is the grid so I don't need the same redundancy as someone anchored off an island or in the middle of the ocean.

Have a look at the most common failure points in a system in order.
1: terminal connections
2: BMS (driven up by increased use of BMS using marginal components)
3: Inverter
4: Charge controller
5: LiFePO4 cells

Parallel cells first minimises the first two.

If you want redundancy build two independent systems, redundant batteries are pointless when your inverter fails.

The ability to use cheap FET based BMS is the only reason i can see that makes sense.

I'm in the redundancy and serviceability without down time group.

toms said:

Have a look at the most common failure points in a system in order.
1: terminal connections
2: BMS (driven up by increased use of BMS using marginal components)
3: Inverter
4: Charge controller
5: LiFePO4 cells

Parallel cells first minimises the first two.

If you want redundancy build two independent systems, redundant batteries are pointless when your inverter fails.

The ability to use cheap FET based BMS is the only reason i can see that makes sense.

Click to expand...

I know, you're exactly right... For my final house system, how about 3 batteries and 4 all-in-ones (2 of which I will likely not install into live system and put in a faraday container, along with 7 spare 16s BMSs).. I hope I've planned it all out the best way I think I can hehe...

Oh, and lots of spare terminal connectors (and heat shrinks)... And the hydraulic crimper I bought even came with a spare replacement seal kit...

Hey what can I say, my wife doesn't like power outages, just wants things to work, forget about the 5-nines here, she wants 100% uptime..

But I can't afford to build the house yet because I spent all my money on redundant power is the only problem now.

timselectric said:

I'm in the redundancy and serviceability without down time group.

Click to expand...

Same here. 4s2p is what I have. Not a huge system.

In an RV, space considerations can trump anything else. I have room to place separate 4s batteries, but finding room for one large 2p4s battery is difficult.

It’s ironic that one of the “advantages” of series first is cell redundancy, yet most of the reasons for having to use that redundancy are BMS or connection related, and series first multiplies that failure mode.

It will be interesting to see long term how the trends develop.

There are certainly some reasons to use this method - but reliability isn’t one of them.

Sailors (like you find on cruserforum.com) prefer to parallel first because they like one house "battery" not just a bank. Much of the reason is because you can spend the extra money for a REC BMS, that will then tie directly in with the Wakespeed alternator regulator and the Victron Cerbo GX, and charging from alternator or solar is controlled via feedback from the REC. I remember getting frustrated because some of the sailboat guys that I talked to about having multiple batteries (each with their own BMS) in parallel for a single house bank made no sense to them.

toms said:

It’s ironic that one of the “advantages” of series first is cell redundancy, yet most of the reasons for having to use that redundancy are BMS or connection related, and series first multiplies that failure mode.

It will be interesting to see long term how the trends develop.

There are certainly some reasons to use this method - but reliability isn’t one of them.

Click to expand...

If you have one BMS and it fails, you're completely down. If you have 10 BMS's and one fails, you are still running at 90%. And you can replace it at your convenience. Going on your theory that a BMS is a guaranteed failure point. It would make sense to have redundancy.

timselectric said:

If you have one BMS and it fails, you're completely down. If you have 10 BMS's and one fails, you are still running at 90%. And you can replace it at your convenience. Going on your theory that a BMS is a guaranteed failure point. It would make sense to have redundancy.

Click to expand...

If thats the logic, I choose 10P with a second BMS on the shelf and save 8x cost of BMS.

The logic is that if you need redundancy, you run two independent systems - most full time off grid systems i set up are like this after prospective owners have seen how my system has operated long term and after speaking with other long term off grid owners.

Having redundant batteries is no benefit if your inverter fails.

If redundancy is the main reason for series first - it is a poor reason.

As i said, generally outlay cost is the biggest driver of series first, and the jury is still out on how that translates to lifetime cost.

schmism said:

If thats the logic, I choose 10P with a second BMS on the shelf and save 8x cost of BMS.

Click to expand...

One thing I like also about having six 120a BMS's, is when I'm drawing 250a running the toaster oven and the AC at same time, or in the morning hours charging on the bank with a solid solar charge of 200a, the amps move across all six BMS's distributed evenly, and they never get even warm or ever running close to peaks, theoretically this should help them last longer.

With six BMS's in parallel, I could charge or discharge at 720a if I wanted to and be fine... I don't know of a consumer BMS on the market that can handle 720a continuous charge/discharge current (especially one that wouldn't have any loud fans in them)...

Samsonite801 said:

One thing I like also about having six 120a BMS's, is when I'm drawing 250a running the toaster oven and the AC at same time, or in the morning hours charging on the bank with a solid solar charge of 200a, the amps move across all six BMS's distributed evenly, and they never get even warm or ever running close to peaks, theoretically this should help them last longer.

With six BMS's in parallel, I could charge or discharge at 720a if I wanted to and be fine... I don't know of a consumer BMS on the market that can handle 720a continuous charge/discharge current (especially one that wouldn't have any loud fans in them)...

Click to expand...

You use a BMS that has an external method of disconnecting the load / charge circuits.

Once again, if you are using cheapy ebike BMS, then sure series first is the go. Doesn’t mean it’s going to be cheaper in the long run - and will certainly be less reliable.

A decade of watching people try to make the FET based BMS work long term has led me to this conclusion. Because it’s cheap there is no shortage of new people wanting to learn the same lessons.

It all really depends on how many cells you have and what voltage you need to configure them for.

If I am building a small 12V battery and I have 4 cells then I only need one BMS. If I have 16 cells then I could do 4P4S or 4S4P. The former requires 1 4S BMS, while the later requires four 4S BMSs, That's four times the cost for BMSs for the same pack of batteries. Now if I am building a larger 48V battery and I have 64 cells, then at a minimum I still only need one 4S BMS (16P4S). On the other hand if I put them in the (16S4P) configuration, I need four 16S BMSs. 16S BMS are more expensive than 4S BMSs, and I would need 4 times as many of them. Again if the ampacity of my cells are small, then 16P isn't so bad. If the ampacity of my cells are large, then I wouldn't be able to find a BMS or cable large enough to get the power out of a 16P configuration.

People building smaller batteries don't need or care about monitoring, redundancy and hot swap-ability, and it greatly increases their cost.
People building larger batteries need monitoring, redundancy and hot swap-ability, and it barely increases their cost.

Reality for most people lies in the middle somewhere and that is why the question keeps coming up.

Small = PS
Large = SP

Now the question is what is the purpose is of the BMS that you are building. You are doing it "so channel capacity isn't an issue" and "looking at this recently for a future boat install 300AH , 12V". Since your voltage is about as low as it gets (4S), and your ampacity isn't that large (300Ah LiFePO4 cells), your battery is pretty small. You really only need four 300Ah cells to produce a 12V 300Ah battery. That is one 4S BMS and nothing in parallel. So unless you really need the redundancy of SP, the PS model would be more cost effective for you and much less hassle. Its easier to put 4 cells into a 4S configuration than it is to put 12 smaller cells into a 4S3P configuration.

Even if you don't go for a 4S configuration, in your case the 3P4S configuration is still better because your cells are so small. You won't get any benefit of redundancy trying to swap out a 12V 100Ah battery from a pack of three of them. The remaining two batteries will be dead long before you get the replacement bolted in place.

schmism said:

If thats the logic, I choose 10P with a second BMS on the shelf and save 8x cost of BMS.

Click to expand...

It's not my logic.
I have redundancy built-in to every part of my system.
More paralleled panels than needed. More paralleled inverters than needed, and more paralleled batteries than needed.
Backups for the Backups.
If anything fails, I can isolate it from the rest of the system. And replace it at my leisure. Zero downtime is important to me.

Well, and I come from the troubleshooter background, was an ASE mechanic for a decent share of my life, troubleshooting every day for a living, later moved on to copy machine troubleshooting/repair, then wireless ISP installation and troubleshooting/repair, datacenter troubleshooting/repair in very complex clustered virtualization environments, with distributed storage architectures, etc..

Troubleshooting is in my blood. And I really enjoy the idea of making management and monitoring system health, something where you don't need to go grab your multimeter and get down on your hands and knees to go look and start disconnecting stuff anytime you have a problem, but rather as system where I can simply log into a bluetooth app, and know that every cell voltage is tied 1:1 to an actual cell. It really simplifies monitoring in my opinion.

Also, if I have a problem due to a failure on a Friday afternoon, I'd rather be able to wait until Monday to fix it, instead of 'red alert, the system is down' kind of fire to go put out right now.

Now, should everyone do it my way? No... I just prefer that strategy on my own stuff, it works for me in my life. And I still keep spare BMS's on the shelf, just like I keep spare hard drives for all my RAID arrays in my data storage units. Even spare power supplies, spare network adapters, spare RPi's, spare chassis fans, spare routers, spare cell phones, spare laptops, etc, etc...

In my industry as a life-long troubleshooter, I see nothing but breakage and failures day in and day out, I never get to hear the good news when it happens, only the bad news. Identifying failures and fixing them quickly and easily, is a top goal on my plate.

The only time I might consider doing P first is with the small cylindrical-cell packs, like how the Teslas are wired up... But then you have to fuse each cell as well. I would never build a pack in that way, just due to the PITA factor of assembling such a tedious system that limits health monitoring down to groups of cells (and then how would you even know if some fuses blew on any cells and you lost some capacity, until that cloudy Winter morning).

Remember that when you swap out a cell that is in parallel, the voltage needs to be near the same as the rest of the battery. Otherwise you are going to dump a massive amount of current from the new cell to all of the older cells. That could trip your BMS. So the idea that serial first is the best solution for fault tolerance is complicated by this fact. You can't just start flipping breakers on and off to spare cells.

In addition the general rule of battery design is that all cells are similar. That is why we match cells and don't mix different brands, chemistries or ages. By implementing a hot swappable design you are violating this concept. Even if you have spares that are of the same brand, model, and age, the fact that they are spares implies that they have a different usage history. You would need to periodically rotate your spares into the battery to ensure that all cells remain equal.

Again for smaller batteries, the best plan is to keep a complete spare battery around, not just a spare cell. For larger batteries that is not possible due to expense, weight, and installation complexities, so a hot swap model is required. I don't think most people meet this criteria.

If the goal is to always have a backup cell available if one fails in a parallel configuration, then the best design is to have the backup cell already in the battery and simply run off of all but one of them. That way they can be rotated by flipping a breaker off and letting the others handle the load. But if you are going to do that, then all of the cell may as well be in use. That eliminates the need to rotate them and lets you use the money you invested in all of the cells. That is the point of parallel connections, they are all backups all the time. Its already automatically a redundant design.

Ideally everything would be in parallel. Serial connections are what cause the need for a BMS. Because we don't have 3.2V inverters and inexpensive lightweight large-gauge wires, we need serial connections to push the voltage up. Once we do that we need a BMS. So the whole question is when to do that.

Small = PS
Large = SP

Again the voltages must be nearly the same when you introduce a new cell into the mix. This is even more problematic with larger cells since they can transfer higher currents. So to change out a bad cell requires that you wait until you can bring all cells back to full charge. If you don't want to wait, then you must have a full battery on standby and switch between them mutually exclusively.