7s BMS how do you manage the parallel?

friedpenguin

New member
Joined
Apr 1, 2019
Messages
27
Isn't the goal of a BMS to monitor every battery? And if I'm not mistaken the average 'made in China' BMS including the smart ones are one connection per cell. So if you're 7s2p to 7s10p or more how to you manage the parallel cells? Batrium is out of the question. If I'm spending that much money I'm not going to be recycling and just buying new LifePo4 which are out of the budget too. If this has been asked please point me to the right thread(s).

Project details: Off grid greenhouse. Already have seven 24v 10ah LifePo4 but trying to scale up to 200ah total storage over time and I have access to too many laptop batteries and an Opus C3400.
 
There are no BMSs, that I am aware of, that can monitor "each" cell that is connected in parallel. I'm sure there are proprietary system built, but not available to us.

The only way is to either have a bunch of cheap chinesiums to monitor each single series. But, even that would become really expensive really quickly.

This is really why we fuse each cell of recycled cells in a parallel pack. So when, if, a cell goes rogue, it gets knocked outta the connection loop so it doesn't kill the pack. However, that doesn't fix the slow leakers. Only monitoring over a period of time can we tell which pack(s) are exhibiting signs of a slow leaker. Then the pack must be checked.
 
By having cells in parallel in the pack, the cells tend to act together & the BMS still monitors them for voltage.
Because they are in parallel, the voltages are all equal within that parallel section.
The more cells in parallel, the more a BMS will have to work to maintain balance.
With care when matching capacity & weeding out slow leakers like Korishan mentioned, you can keep the battery balanced.
 
So it comes down to how many 'p' you're comfortable with and what your BMS can do? So a 24v 40A is probably good for 7s4p max? If you fuse each cell and one cell dies what happens to the other cells in that series? I'd like to be able to build a hot swap 7s setup and fuse each 7s1p so if one goes south have a way of notifying and replace just the 7s bank. That's outside the scope of the BMS probably but I could find another solution to monitor the voltage of each 7s that's hopefully affordable.
 
The hot swap thing is viable, but in practice it has its own challenges you have to match the pack voltages before introducing the hot swapped pack into the system. Any large delta will create a significant inrush of current. Larger the delta, larger the inrush. It could be significant enough to blow cell level fuses.

As far as the parallel cell count its all electrically connected, you cant monitor then individually in a pack just think of it as one cell, and in those instances one cell monitor per parallel set is more than sufficient.

If you wanted to actually monitor individual cells, that would need a pretty complicated setup, plus I fail to see the true benefit after looking at ROI. For cost of the BMS and control system, I would just buy new cells and take my chances with an extremely low failure rate.

The PCB Justin/Jehu did is basically your 7s1p board if you dont use the stacking headers or ribbon cable, but the entire point of that was to simplify the monitoring requirements by adding that into it.
 
friedpenguin said:
So it comes down to how many 'p' you're comfortable with and what your BMS can do? So a 24v 40A is probably good for 7s4p max? If you fuse each cell and one cell dies what happens to the other cells in that series? I'd like to be able to build a hot swap 7s setup and fuse each 7s1p so if one goes south have a way of notifying and replace just the 7s bank. That's outside the scope of the BMS probably but I could find another solution to monitor the voltage of each 7s that's hopefully affordable.

you're getting your P's and S's mixed up.

1/ if one cell dies, either the fuse blows, or it doesn't (wrong spec of fuse!). IFF the fuse blows, then NOTHING happens to the other cells in series (or parallel). BUT your pack is now swinging out of balance. each group of cells (thats 1s) has a capacity ... and unless the battery was built with too much capacity in that group of cells (unlikely) then things are going to go south ... sooner or later. IFF the fuse doesn't blow, the every other cell in parallel discharge to the failed cell ... then on the next discharge of the battery -- without a BMS -- these cells flip polarity and things get interesting (that'd be the bad type of interesting) real fast. Of course, ACTIVE balancing, or even super smart PASSIVE balancing (including Batrium, if I'm not mistaken) can compensate for this, but then check the logs ... you are losing power to a problem pack ... fix it or ... lose efficiency.

2/ making a battery out of strings of 1p packs is ... very very bad practice. ... like AveRageJoe said to his dad ... 'please don't do that'.

a couple of other things ... cell vs battery -- they are different. words have meanings. suggest you take more notice of what you are saying (genuine/sincere advice here, not picking on you). I know it takes time to learn, but you seem to already have a good grasp of relevant stuff here, so ... best to use the right words, and speak with clarity.

as for individual cell monitoring in a battery ... its a LOT of wires to monitor each cell. Im sorta working on an project to do such a thing, but it would ONLY EVER be used for 'breaking in a battery' (which I do BEFORE battery construction! I balance charge each prospective cell for the battery to a very precise voltage. they can come from different stock/caches or existing cartridge style batteries I have in production use .. I shuffle them, just like I stack 'em ... thus eliminating 99% of 'problem cells' before final battery construction).

To monitor each cell (pair, in my case!) in a real battery (5+ KWh) composed of little 18650's is ... non-trivial. how big (capacity) are you building? I use 72p for my big builds, that 36 pairs ... in a 13s config ... that'd be 468 channels for an ADC or discrete op-amps to process ... totally non-trivial. and thats not even a 'big' battery! just a string ... the clue I'll give you is this: use the cell-level fuses as shunts ... usingdifferential op-amp's, one could put one input on the busbar (easy!) and the other input on the cell's side of the fuse (lots of wires, and also ... it's generally really hard to actually wire anything to the cell side of the fuse, eh?! unless you use cartridge style, which I do often, and always do for 'breaking in' a battery before construction...).

I admire your enthusiasm friedpenguin. Welcome to the 'board (SLS). Thanks for sharing your Q's. ; ) hope this helped ...

-----------------------------------------------------------
EDIT: I use string/battery interchangeably, as they are identical, unless. ... (segway, another hint follows!)

you join each interconnect between groups of cells, between each string ... to be clear, what I talking about here is the idea of joining (connecting/shorting) the interconnects of each string, IE then the PACKS are paralleled not just the strings). then things get very exciting. there is a now an easy place to measure performance of each 'pack' in each string of the battery (monitor the current flows between this interconnect of interconnects). so in theory one could isolate a problem 72p (in my case) to investigate, as opposed to 'something in the string' or whatever.

the problem comes IMHO, from the reality that no-one wants to or can harvest every cell then make their battery. its incremental, kinda organic. it grows. so the question in my mind is how to cope/manage that. scalability has many many many aspects.
 
DCKiwi summed it up pretty good. The BMS would alert you (if it's a smart bms, not a cheap chinesium one) when a pack is going bad. It'll alert you that a pack is getting drained faster and charged faster (missing cell(s) in the pack resulting in a faster charge/discharge cycle).

I will disagree slightly on the monitoring every cell aspect. It is possible, but it is a lot more work, and a lot more tedious. I personally plan on this route, but, I also plan on having hot swappable cells. Mine won't be soldered/welded together, but firmly pressed in the holder. Then have a small pcb that monitors each cell. However, I'm still designing it and working through bugs with it.

For practicality purposes, and ones sanity, monitor only a pack at a time, not per cell. At least to start off with. Then just use scripts or database to show the discrepancies.
 
Trying to find the balance (no pun intended) of parallel cells vs BMS cost. I'd rather keep my packs reasonably small and easier to manage so 7s5p should average 10Ah @ 24v assuming ~2ah per 7s. At an average cost of $40 per 'battery' that seems reasonable. If I were buying new cells I'd go bigger. From there it's just a matter of figuring out how to tie into an esp8266 or Pi and how many batteries I can monitor per device. The Wemos D1 are a favorite board of mine with plenty of I/O so if it could monitor half a dozen packs and report those stats via mqtt I could set alerts if something goes wrong and take action on the appropriate pack. Besides, seeing some of the Grafana charts on here make me realize I need more widgets in my chart!
 
I apologize but really feel like are over thinking this while it does seem like an interesting project, it sounds like its more of a BMS project than a battery pack project.

I feel all this added stuff just adds unnecessary complexity and increases the potential for issues. Do you have an estimate of what you might end up sending on just BMS? There comes a point where when this thing starts to scale, other solutions make more sense (large packs in parallel with a single cell monitor per pack) similar to how most Batrium projects are built and implemented.

Going back to a more traditional built methodology

Cells in parallel dont need a separate BMS. They are all electrically connected, they will all read the same voltage. Large parallel packs are also more tolerant to balance issues vs smaller parallel counts.

Even if you have made separate 24v battery packs, all connected in parallel, they would ultimately all share the same voltage at the pack level.

Maybe I am still unclear on what it is you are trying to accomplish, but if you just need to power some stuff using a battery pack, there are easier ways to accomplish this.
 
Plan was to use the cheap 'smart' BMS and monitor them. By keeping the packs smaller I could add them to my system one at a time each with their own BMS over time. As it stands I have seven 24v LifePo4 batteries each with their own BMS. I'll never run a load greater than what half the batteries can push at once just so there's always reserve to keep the critical stuff running 24/7. As it is I have a blocking diode on the battery connected to the Pi so if the voltage on the main bank caused a LVD the Pi would keep running (which has only happened once so far). By having smaller packs I'd be looking to minimize downtime by just pulling a small pack if it was out of spec with the rest of the system. The BMS would be there to prevent overcharge and offer LVD for any pack that drained faster than the rest. All the packs would then equalize each other. Even the battery with the blocking diode will get equalized to any higher voltage than it presently was.
 
friedpenguin said:
Plan was to use the cheap 'smart' BMS and monitor them. By keeping the packs smaller I could add them to my system one at a time each with their own BMS over time. As it stands I have seven 24v LifePo4 batteries each with their own BMS. I'll never run a load greater than what half the batteries can push at once just so there's always reserve to keep the critical stuff running 24/7. As it is I have a blocking diode on the battery connected to the Pi so if the voltage on the main bank caused a LVD the Pi would keep running (which has only happened once so far). By having smaller packs I'd be looking to minimize downtime by just pulling a small pack if it was out of spec with the rest of the system. The BMS would be there to prevent overcharge and offer LVD for any pack that drained faster than the rest. All the packs would then equalize each other. Even the battery with the blocking diode will get equalized to any higher voltage than it presently was.

small/replaceable packs == 'cartridge' style. scales well. good method if you need to get battery online 'fast' without waiting for full load of tested cells and/or if you wish/need to reassign cells to another task etc.

however... I say better to build each cartridge well, than to think/plan to replace them. afterall, you're busy collecting more cells to expand/complete, not just maintain, right??!?

self-discharging cells are probably your biggest enemy. as I read it you are not stressing your strings/battery. so therefore my advice would be ... to concentrate on eliminating (removing) ANY and ALL cells which don't 'sing well' with the others. removing one single bad cell can help hundreds of other cells sing and dance in harmony ... truely. and harmony is KEY ... whatever the battery is ...

also, ditto what Crimp Daddy says (above). : )
 
friedpenguin said:
Plan was to use the cheap 'smart' BMS and monitor them. By keeping the packs smaller I could add them to my system one at a time each with their own BMS over time. As it stands I have seven 24v LifePo4 batteries each with their own BMS. I'll never run a load greater than what half the batteries can push at once just so there's always reserve to keep the critical stuff running 24/7. As it is I have a blocking diode on the battery connected to the Pi so if the voltage on the main bank caused a LVD the Pi would keep running (which has only happened once so far). By having smaller packs I'd be looking to minimize downtime by just pulling a small pack if it was out of spec with the rest of the system. The BMS would be there to prevent overcharge and offer LVD for any pack that drained faster than the rest. All the packs would then equalize each other. Even the battery with the blocking diode will get equalized to any higher voltage than it presently was.

Hi Penguin,

I think it sounds like a good plan. I am thinking along similar lines. I've been very slowly moving into the electronics and battery area and I like the idea of 'hot-swappable' packs. I think any system has it's pro's a cons ... I don't think the 'standard' 80+p packs that are then series'ed into 7s or 14s is a bad idea by any means but if you have issues, and you need to remove a pack to resolve an issue, unless you happen to keep a few 80+p packs on hand, not being used, then your system, essentially, has to go done while you fix the problem. Moving forward, I hope to be largely off-grid and so having my battery system completely off-line doesn't seem to work well for me. Combined with my slow progress, means by building smaller packs (7s18p, only 126 cells) means I can use the system as I build system.

I am keen to follow your progress and see how you go. I currently am working with the idea of a 7s18p design, which, depending upon average cell capacity is, more or less, 1kwh per pack. I think, for me, any smaller is, as others have suggested, maybe more work than it's worth ...but time will tell I guess. I'm not sure I'll monitor every 7s18p pack but I will definitely have balance leads on each packto allow it to happen if I move that way.
 
Back
Top