48V LiFePo4 battery 16S2P or 2P16S

MisterSandals said:

The OP of this thread was asking about 2P16S vs 16S2P. I also am wondering what your thoughts are on your battery setup. How did you match your cell pairs? Any balancing issues?

Click to expand...

Ahhh, I read to many threads, they get all jumbled together.
I'm quite happy with the setup.
I didn't really match them, I used 8 of them in a smaller build for under a year... I bought 9 more and later bought 16 more, all 230ah EVE from Docan. When I built the 48v packs I top-balanced all of them to 3.55v then 3.645v in groups of 8. The only matching would be the last 16 ordered went in the same pack.
I'm fairly happy with the balance, no runners until they get above 3.59 then a few climb faster, not a big deal to me as I rarely run the packs that high.
I don't at all regret going with 2 separate packs, gives great redundancy. I made sure all the wiring was darn close to the same length and both packs behave pretty much the same.

I ordered some rs485 cables for the BMSs, should be here next week. I'm hoping to connect them to solar assistant, the inverter estimate is way off.

hello i'm neuw in this forum and French so bee indulgent with my accent!
well i have the same reflexion j'have ordering 32 cells for my home (off Grid) and wath is the best configuration 16S 2P or 2P16s ....

Every cell should have its own BMS connection, so 16s2p in my opinion. Otherwise you have no visibility to a cell failure or struggle.

MisterSandals said:

I'll be the dissenting voice. I like 2P pairing and have had great luck with my 2P4S battery.

With a 2P16S setup, you have the opportunity (aka a bunch of work) to create equal pairs of cells based on capacities.
So if you have 32 cells of varying capacities, you can make equivalent pairs by pairing the highest capacity with the lowest capacity, second highest capacity with second lowest...

Click to expand...

I'm sorry, but this is NOT a good idea. Just to be clear, a 2P16S battery is 16 pairs of cells in series. Each cell in a pair is connected to the other cell, both positive and negative.

The BMS in this case can only read the voltage of each pair. Because they are in parallel, it is reading the average voltage of the two. If a pair of cells are not the same capacity, the voltage of the lesser will drop more quickly as it approaches empty. This means that the average voltage of the pair can be above the minimum acceptable voltage while one of the cells in the pair is below it. Given that depletion past 0% SOC will damage a cell, you will quickly damage up to half of your cells.

The sole job of the BMS is to keep all the cells from going over or under voltage (as well as managing over voltage and temperature excursions on a pack level). The BMS is designed to stop when any individual cell reaches min or max voltage. Because of this, you need to match the cells in any parallel set (2P, 3P, 4P, whatever) so that these 2, 3 or 4 cells act in lockstep and act as if they were one cell so the BMS can monitor them. If your matched pairs of cells do not match the other matched pairs, this is not a problem, as the BMS will limit the battery to the capacity of the smallest pair. This can work perfectly well especially if the BMS supports active or passive cell balancing or you occasionally top balance them. The balancing will simply keep the top (full) of each pair in sync with the other pairs.

TLDR: Closely match the capacity of each cell in any pair of cells. The pairs do not have to match other pairs. If you have matched pairs of cells a 2p16s battery will work well for many years.

Ogre said:

Because they are in parallel, it is reading the average voltage of the two. If a pair of cells are not the same capacity, the voltage of the lesser will drop more quickly as it approaches empty.

Click to expand...

That is not exactly correct. By definition, anything in parallel will have the same voltage.. That is just physics. It is true that if they are not the same capacity there will be currents between the two as the weaker cell is charged by the stronger cell.

Ogre said:

you need to match the cells in any parallel set (2P, 3P, 4P, whatever) so that these 2, 3 or 4 cells act in lockstep and act as if they were one cell so the BMS can monitor them.

Click to expand...

The BMS monitors them as one cell regardless if they are matched with cells of exactly the same capacity, They appear to the BMS as one cell regardless of differences in capacity.

@Ogre It's already been demonstrated that LFP batteries don't have to match in capacity to play quite nicely in parallel. Our host Will has done experiments with a 5Ah pack in parallel with a 100+Ah pack, and shown that they both seem to discharge at about the same rate as measured in SoC. Do you have evidence this is different for cell matching in a mPnS setup?

Regarding BMS, I looked seriously at contactor based, and ended up going with the JKBMS and 1P strings. Active balance means if you're lazy you can throw batteries together straight out of the box and quickly get them balanced. Yes, you might lose a small amount of potential capacity compared to matching cells into more equal strings, but I expect you could get most of the same effect by running a few discharge cycles after balance is complete and watching which cells hit low knee first.

For those saying they want cell monitoring on each cell of a parallel pair, why aren’t you concerned with intra cell monitoring of the parallel segments of a large capacity cell?

I fail to see the difference between two 100ah cells in parallel, and a single 200ah cell. (If the external connection is bugging you then take a look at how many extra connections you are introducing by using 2 BMS.)

@TorC my experience is primarily with NiMh cells. NiMh cells have the ability to be driven negative (into cell reversal), which destroys them, without risk of thermal excursion (because they radiate the heat out nicely). While I see your and @Ampster points about the voltage of the smaller (lower capacity) cell being propped up by the larger one, You are still at minimum driving the first cell down to absolute zero capacity which will age the smaller cell and degrade it more quickly. A quick google search brought me to this: Management of imbalances in parallel-connected lithium-ion battery packs which I unfortunately can't get access to because of the paywall.
The battery pack I'm about to build will use matched EVE 310 cells in a 2P16S configuration (48V 31Kwh). Matched because I want it to last 10+ years in an RV setting and it will be stressed due to external temperature swings and intermittent charging (alternator charging @6000watts).
I've also been working with a colleague in Hong Kong to develop quality pre-packaged 12V packs (similar to SOK or Ampere time) for RV use. The preproduction samples are 4P4S 100Ah CATL cells, which gives a 12V 400Ah pack in a convenient form factor. So I'm not against paralleling cells, I'm just against paralleling non-matched cells.

Ogre said:

NiMh cells have the ability to be driven negative (into cell reversal), which destroys them, without risk of thermal excursion (because they radiate the heat out nicely).

Click to expand...

Oh, and in case you want to know how this happened, back in the day Honda (and early Toyota) hybrids had packs made from 120-240 cells in series, but the BMS only monitored them in strings of six cells. As the cells aged, the internal resistance of individual cells would vary, and the vehicles did not do any sort of top or bottom balancing. Eventually you'd have a string of six cells where one cell was near 0% SOC and the others were near full. Because the total voltage was higher, the BMS thought the SOC was somewhere in the middle. Upon discharge, the empty cell would become totally depleted and the rest of them would continue to flow through it and the empty cell would start charging in reverse.

These are defective cell strings we removed while servicing these packs.

We had the rest in the basement because we were concerned about the weight causing structural damage to the building. Each of those sticks weighs 2.4 lbs and is 7.2V 5.5Ah

Ogre said:

, You are still at minimum driving the first cell down to absolute zero capacity which will age the smaller cell and degrade it more quickly.

Click to expand...

How could a cell be at xero capacity and be the same voltage as it's buddy cell? Look at the discharge curves and tell me how that could happen unless both cells are in the lower knee of the discharge curve?
I am not arguing that one's goal should not try to build a pack from matched cells. I am only saying that if you do not have the luxury of doing that then you need to operate that pack more conservatively. I take a risk management approach and never take my pack below 20 % and only charge to 3.45 volts per cell. i also have low voltage minimums set on my BMS as the final defense to stop discharge.

Ogre said:

Oh, and in case you want to know how this happened, back in the day Honda (and early Toyota) hybrids had packs made from 120-240 cells in series, but the BMS only monitored them in strings of six cells.

Click to expand...

That is clearly a bad design. Maybe that is why Honda and Toyota went in the direction of Hydrogen and lost years of momentum. Another bad decision on there part.

Cells directly connected in parallel should be matched in capacity and overpotential current profiles (terminal voltage slump vs cell current) so they share load/charge current equally.

A battery assembly manufacturer can do this with vendor lot controls and sorting checks. A DIY'er buying a few cells is not likely to be able to ensure this.

What a DIY'er buys as 'matched' cells usually have conductive Rs similar, charged to same level, and maybe a capacity test. They are not normally tested for matching overpotential vs cell current. The 1kHz Rs (R_ohmic) measurement is only a small piece of the total cell impedance. R_ionic, the cause of most of cell voltage slump with cell current, dominates effective cell impedance. This takes a moderate DC load for a few minutes to check.

It will cost you a BMS for every series string but separate is better way to go. You do have to be careful of one BMS cutting out turning over all the charge or load current to remaining strings.

Ampster said:

How could a cell be at xero capacity and be the same voltage as it's buddy cell? Look at the discharge curves and tell me how that could happen unless both cells are in the lower knee of the discharge curve?
I am not arguing that one's goal should not try to build a pack from matched cells. I am only saying that if you do not have the luxury of doing that then you need to operate that pack more conservatively. I take a risk management approach and never take my pack below 20 % and only charge to 3.45 volts per cell. i also have low voltage minimums set on my BMS as the final defense to stop discharge.

Click to expand...

I'm talking about when the smaller cell has reached terminus (below the knee) and the other has not. Obviously if you are not going below 20%, you are not going to stress it as much, BUT you will still throw the cells out of balance very quickly. For example, let's try an extreme: 100Ah and 20Ah. What happens when you pull 60Ah out of them? What happens when you charge them back up? It's not proportional to the size of the cells. Your top balance will be lost quite quickly. Will it work? Yes, for a while. But the smaller cell will fail much more quickly than it would have.

Ampster said:

That is clearly a bad design. Maybe that is why Honda and Toyota went in the direction of Hydrogen and lost years of momentum. Another bad decision on there part.

Click to expand...

Up until 2012 at least they both still used 7.2V NiMH although Toyota used a six-cell prismatic design instead of cylindrical. Ford Escape Hybrids used the same cylindrical sticks just packaged slightly differently. Mind you this is pretty much every hybrid vehicle made worldwide from 1997 - 2012. After that, lithium became available. The primary reason for this is the cost of the the housing. If the cells ("D" cells) had to be housed independently of each other, the housings would have been much more complex, heavy and expensive. Auto manufacturers shoot for a lifespan of 8 years, so this made sense for them. Most battery packs would last that long, and they'd simply replace the ones that didn't under warranty. We however want longer life than that and we ourselves are the warranty department. BTW those batteries failed much more frequently in the American southwest and in Alaska. The heat and cold made a mess of them. That said, the Toyota packs survive much longer than the Honda ones simply because Honda uses 20-80% SOC and Toyota uses 40-70ish %. The Honda packs fail more because they are used more, but the Hondas get better mileage because they use more EV power vs gasoline.

Ogre said:

I'm talking about when the smaller cell has reached terminus (below the knee) and the other has not

Click to expand...

Explain to me and other readers the physics of how that could happen when cells are in parallel and their voltage is exactly the same?

Ogre said:

Mind you this is pretty much every hybrid vehicle made worldwide from 1997 - 2012. After that, lithium became available.

Click to expand...

Lithium was available long before 2012. I had cell phones in 2000 that had Lithium batteries. The Tesla Roadster used Lithium. Notebook computers used Lithium. Power tool batteries were transitioning to Lithium. I still believe Honda and Toyota were slow to adapt. I can add Ford to that also. I understand hindsight is 2020 vision. Ford was able to pivot and came out with the F159. Honda and Toyota are both losing market share because they were slow to react.

@Ogre
I agree with the concept that matched cells are better. I agree with @RCinFLA that DIYers do not have the equipment to match cells like the the OEMs do. I also believe that one can have a cost effective pack by using Grade B cells and using them conservatively. The laws of physics when applied to a risk management strategy can help a DIYer find that happy medium. I do not believe perfect should be the enemy of good.

RCinFLA said:

It will cost you a BMS for every series string but separate is better way to go. You do have to be careful of one BMS cutting out turning over all the charge or load current to remaining strings.

Click to expand...

That's why I'm going with a CANBus enabled BMS (Rec Q BMS). It can tell the auxiliary alternator how much power it wants, can tell the Victron Multiplus to stop charging, etc. It can control the big items to keep from tripping the contactor in most cases. I am twinning the (brand new) cells though because the cost of a second BMS and then the BMS master unit would add another $1000 to the cost of the system.

Ampster said:

Lithium was available long before 2012. I had cell phones in 2000 that had Lithium batteries. The Tesla Roadster used Lithium. Notebook computers used Lithium. Power tool batteries were transitioning to Lithium. I still believe Honda and Toyota were slow to adapt. I can add Ford to that also. I understand hindsight is 2020 vision. Ford was able to pivot and came out with the F159. Honda and Toyota are both losing market share because they were slow to react.

Click to expand...

Auto manufacturers hadn't really figured out how to use lithium technology economically. Tesla Roadsters were one of the first (if not the first) automotive applications of lithium batteries and that was 2008. The Prius plug-in came out in 2012-2013 but that was a 4.4Kwh pack that weighed 80Kg. (I believe this was the first major manufacturer use of lithium chemistry.) The differences between automotive and consumer use comes down to discharge rates and safety. That Toyota could travel 15 miles at up to 62 mph which means it is designed for a 4C discharge rate as the battery is depleted in 15 minutes by the 60Kw motor. Secondly, automotive lithium packs are designed to vent in a controlled manner and to specific areas. This is so the flames from the vents don't set things on fire. These are not concerns in laptops and power tools. They SHOULD have been concerns in the "hoverboard" market where the batteries were overdriven and therefore had a tendency to catch fire.