diy solar

diy solar

BMS with multiple battery modules - How many and how to connect.

Crashcourse

New Member
Joined
Nov 1, 2020
Messages
31
Hello folks! First timer here.
Just dabbling into Solar and thinking of building my own battery modules for a 24V (possibly future 48V) system.
I currently have six "Series 31" Deep Cycle Marine 12V batteries wired in 2s3p to the inverter, charged by a 60amp MPPT Charge Controller and eight 100W panels wired 2s4p.
My idea is to use 3000mah 3.7V 18650 cells, 30 cells in parallel in each pack X 7 packs for my 24V 4000/8000W Giandel Inverter. I will likely add more 30X7 packs in the future.
My question is, actually 2 questions, is this feasible and how many BMS units will I need and how would I connect each individual pack to the BMSs so I protect my batteries?
Thank you.
 
I haven't built anything with 18650 cells, but I had looked into it a bit, and this guy Jehu Garcia has a lot of videos building banks and powerwalls like, this using small cells. It's worth having a look at some of his videos... He even refers to nice kits for battery holders, BMS, sources for cells, etc...

 
Thanks but I have watched many of his videos, he uses mostly premade modules so he doesn't really address my question. How many BMSs do I need in a multple battery pack config?
 
Thank you, do you mean one BMS lead to each battery pack or one to each of the 30 cells in each pack? This is where I'm confused. Sorry for my ignorance.
 
Number of BMS's depends on your battery configuration. If you parallel a bunch of cells in to each serial string (like say 10p7s), you only need a BMS for every 7 series cell group (on a 24v pack), so that's one 7s BMS. In those cases they usually would put a fuse on each cell in parallel in case one shorted out it's protected (it can break off the p group).

Or you can have fewer unmonitored cells and use more BMSs (like say 7s10p), you have 9 more BMS's than the other configuration and much better monitoring on cell level. If you do use a bunch of 7s single cell individual packs (with a 7s BMS for each pack), and in some of Jehu's videos he has shown how in using those smaller BMS's they connect a parallel ribbon cable across all the BMS's in the pack stack so they can communicate on the same CANbus for easier management...
 
Number of BMS's depends on your battery configuration. If you parallel a bunch of cells in to each serial string (like say 10p7s), you only need a BMS for every 7 series cell group (on a 24v pack), so that's one 7s BMS. In those cases they usually would put a fuse on each cell in parallel in case one shorted out it's protected (it can break off the p group).

Or you can have fewer unmonitored cells and use more BMSs (like say 7s10p), you have 9 more BMS's than the other configuration and much better monitoring on cell level. If you do use a bunch of 7s single cell individual packs (with a 7s BMS for each pack), and in some of Jehu's videos he has shown how in using those smaller BMS's they connect a parallel ribbon cable across all the BMS's in the pack stack so they can communicate on the same CANbus for easier management...
Thank you so much, I understand much better now. So if I have 30 x 3000mah cells in each pack I should probably get a BMS that can handle around 100 amps, correct?
Also am I risking damage to the cells using a single 7s 24v BMS?
Thanks again, very much appreciate the help.
 
Thank you so much, I understand much better now. So if I have 30 x 3000mah cells in each pack I should probably get a BMS that can handle around 100 amps, correct?
Also am I risking damage to the cells using a single 7s 24v BMS?
Thanks again, very much appreciate the help.

Just to clarify, I'm assuming you're meaning 30p7s, which is 210 cells, or by 'pack', are you meaning a whole battery 'module' ?


Me personally, I am not necessarily a huge fan of wiring a bunch of 'p' strings together because you can't monitor on individual cell level, if you get a bad cell within a 'p' group (not a blown fuse, but just a struggling cell), it will pull down the volts on that entire group, making troubleshooting a lot harder (especially if they're all spot-welded together), because you have to pull apart the 'p' array and check each battery cell manually to find the bad one. If I was building a small-cell pack, I might rather do more individual cell monitoring (but then I guess there are more BMS's involved which could potentially fail, so go pick your poison).

Now don't get me wrong, even Tesla does a lot of cell paralleling, it just makes troubleshooting the paralleled cells harder, where Tesla tends to just replace entire vehicle battery packs when they have measurable failures... I think they even build the packs with enough calculated extra capacity that if a cell-fuse pops here or there and they lose some of the 7,104 cells, then you may not even notice, except your range goes down. Of course, Tesla, they have 74 cells on a 'p' string (on the 85kW/h bank), so they don't really notice much if one cell is dropping performance, they just lost one ant in the ant hill here or there...

Example:
"
The Tesla 85kWh battery pack contains 7,104 18650 cells in 16 modules wired in series; each module contains 6 groups of 74 cells wired in parallel. The 6 groups are then wired in series within the module.

So, the 74 cells in parallel provide the current handling; then, you have 6 groups in series, in 16 modules in series, which provides the 400VDC for the motor (16*6*4.2V per group of 74 cells), with a peak current handling of 4.87A*74, or 360 amps.
"

So if using one BMS, you need big high-amp BMS, if doing lots of BMS, they are small, and don't require a lot of current handling each one. You can also do cascading BMS, where you have lots of small BMS, and maybe you have 4 main modules, where you put in the larger 4s BMS on that tier which handles a lot more current than the smaller BMS operating on the lower tier.

So if I did build a home bank with large 'p' groups, I would use those snap in battery cradles, and have them mounted on rows of hinged boards off a wall, so I could easily swing the boards around to open up my cell accessibility, and unplug cells quickly, and have a box of spare cells onsite, so swapping them was a snap. I'm all about building serviceability into any solution.

The other thing to factor in as well is which battery chemistry you will use. These small cells can come in LiFePO4 which are 3.2v and also come in the high-energy density automotive cobalt chemistry ones which are 3.6v so that influences how many series cells you want 8s vs 7s (for a 24v bank), and thus which BMS's you might be looking for.
 
Last edited:
Just to clarify, I'm assuming you're meaning 30p7s, which is 210 cells, or by 'pack', are you meaning a whole battery 'module' ?


Me personally, I am not necessarily a huge fan of wiring a bunch of 'p' strings together because you can't monitor on individual cell level, if you get a bad cell within a 'p' group (not a blown fuse, but just a struggling cell), it will pull down the volts on that entire group, making troubleshooting a lot harder (especially if they're all spot-welded together), because you have to pull apart the 'p' array and check each battery cell manually to find the bad one. If I was building a small-cell pack, I might rather do more individual cell monitoring (but then I guess there are more BMS's involved which could potentially fail, so go pick your poison).

Now don't get me wrong, even Tesla does a lot of cell paralleling, it just makes troubleshooting the paralleled cells harder, where Tesla tends to just replace entire vehicle battery packs when they have measurable failures... I think they even build the packs with enough calculated extra capacity that if a cell-fuse pops here or there and they lose some of the 7,104 cells, then you may not even notice, except your range goes down. Of course, Tesla, they have 74 cells on a 'p' string (on the 85kW/h bank), so they don't really notice much if one cell is dropping performance, they just lost one ant in the ant hill here or there...

Example:
"
The Tesla 85kWh battery pack contains 7,104 18650 cells in 16 modules wired in series; each module contains 6 groups of 74 cells wired in parallel. The 6 groups are then wired in series within the module.

So, the 74 cells in parallel provide the current handling; then, you have 6 groups in series, in 16 modules in series, which provides the 400VDC for the motor (16*6*4.2V per group of 74 cells), with a peak current handling of 4.87A*74, or 360 amps.
"

So if using one BMS, you need big high-amp BMS, if doing lots of BMS, they are small, and don't require a lot of current handling each one. You can also do cascading BMS, where you have lots of small BMS, and maybe you have 4 main modules, where you put in the larger 4s BMS on that tier which handles a lot more current than the smaller BMS operating on the lower tier.

So if I did build a home bank with large 'p' groups, I would use those snap in battery cradles, and have them mounted on rows of hinged boards off a wall, so I could easily swing the boards around to open up my cell accessibility, and unplug cells quickly, and have a box of spare cells onsite, so swapping them was a snap. I'm all about building serviceability into any solution.

The other thing to factor in as well is which battery chemistry you will use. These small cells can come in LiFePO4 which are 3.2v and also come in the high-energy density automotive cobalt chemistry ones which are 3.6v so that influences how many series cells you want 8s vs 7s (for a 24v bank), and thus which BMS's you might be looking for.
Thank you, yes I meant each 30p pack, 3.6V 3000mah 18650s X 7S for 24V. If my math is correct a single 7S 100a 24V BMS would work for each 30P7S module I make.
And thank you yet again for the detailed explanation. :)
 
From my experience it would be better to keep it at one single bms system.
If one of many bms will fail, the rest will follow shortly, auwa that was a 10kv spark in the wallet.
Lets say every bms can handle 200a and you have a surge of 205a 3 seconds, the weakest one will give up.
All the remaining power will be drawn to the bms's that are still alive, this can be a cascade: done in 10 minutes or ten days....
It is nice to build a system with several "add ons", but will be more expensive in the future, cost wise. Even if everything(bms, fuses) will survive...
It looks nice on yt channels but practice on the long run or bigger is a different story.

I would suggest to buy a 123smart bms or a batrium. oke the initial purchage is very expensive but very cheap in the long run.
They are both the safest systems also, is your house really worth that 100$ saving in a short period(starting up period)?

If you are already thinking in about to go to 48v do it right from the start, the materials to start with a far more cheaper and better available.
Just think about cables and fuses..

Secondhand 18650 like all secondhand lithium chemistry need to be tested at there soh! Use only 80% and above soh.
And do a cell level fusing 5a per cell...

My experience:I build a 106kwh 48v with second handed 18650 li ion cells power wall......
Oh boy....once you get the hang of it
#1 Take all the cells you have and put them in a 48v configuration.
#2 Make them work with inverter bms and pannels.
#3 collect and test cells for a year or so
#4 make packs for 48v system ans simply add them to the existing packs(keep an eye on the v diff before connecting!)
The most important step:
#5 repeat from #3

J.C. is a great teacher but all the stuff he does is suited for small systems, less then 2kw and not with the intention to use at full or big scale or heavy loads.
The vids from Will Prouse are for medium use, but a capacity is capacity no matter what cell you use.
ps check all your packs every year on fuses and test soh....

Have fun....
 
From my experience it would be better to keep it at one single bms system.
If one of many bms will fail, the rest will follow shortly, auwa that was a 10kv spark in the wallet.
Lets say every bms can handle 200a and you have a surge of 205a 3 seconds, the weakest one will give up.
All the remaining power will be drawn to the bms's that are still alive, this can be a cascade: done in 10 minutes or ten days....
It is nice to build a system with several "add ons", but will be more expensive in the future, cost wise. Even if everything(bms, fuses) will survive...
It looks nice on yt channels but practice on the long run or bigger is a different story.

I would suggest to buy a 123smart bms or a batrium. oke the initial purchage is very expensive but very cheap in the long run.
They are both the safest systems also, is your house really worth that 100$ saving in a short period(starting up period)?

If you are already thinking in about to go to 48v do it right from the start, the materials to start with a far more cheaper and better available.
Just think about cables and fuses..

Secondhand 18650 like all secondhand lithium chemistry need to be tested at there soh! Use only 80% and above soh.
And do a cell level fusing 5a per cell...

My experience:I build a 106kwh 48v with second handed 18650 li ion cells power wall......
Oh boy....once you get the hang of it
#1 Take all the cells you have and put them in a 48v configuration.
#2 Make them work with inverter bms and pannels.
#3 collect and test cells for a year or so
#4 make packs for 48v system ans simply add them to the existing packs(keep an eye on the v diff before connecting!)
The most important step:
#5 repeat from #3

J.C. is a great teacher but all the stuff he does is suited for small systems, less then 2kw and not with the intention to use at full or big scale or heavy loads.
The vids from Will Prouse are for medium use, but a capacity is capacity no matter what cell you use.
ps check all your packs every year on fuses and test soh....

Have fun....
Thanks 100Kwh, wow, do you have a video or blog on the build? A link to some pictures maybe? I'd love to see it.

I'm just dipping a toe ATM. Small system of 8 x 100W panels (4p2S). 6 X Series 31 Deep Cycle Batteries (3p2s), 60A MPPT Charge Controller and a 24V 4000/8000W inverter.

I do plan on using new 18650 cells, have a couple hundred already but I *may* ditch the 18650 idea in favour of going with 3.2V LiFeP04 Prismatic cells, thoughts? Again, thanks for the info on BMS's, I understand the same prinicipals would apply.
 
Very expensive new 18650.
I sold them all(on a other forum), the risks where a tad to big for me, regarding my end goal
The effort in making, getting the cells, harvest testing ect
Price wise(second handed) incl cell fuses used electra(for testing&soldering) wires is lifepo4 far cheaper and safer.
20191109_161735.jpg
Found a old picture
"just" 2800 cells good for 48v 560a ~25kw
And this 4 times plus enough cells to make a other string of 280a....
If a single cell is ~1 euro and need 1400 to get 280a, with a soh greater than 95%, max cycle live 750...
Lifepo4 1350 euro for 48v 280a, soh 100%, cycle life 4000(-6000)

My thoughts: finish and not ditch your li ion, complete them to a 48v pack and invest in lifepo.
Cheap bms (incl a cut off!!!!!)with balancer on the li ions and a 123 or batrium on the lifepo, invest all your savings in lifepo and it will go automatically.
In the Netherlands the gas is the most expensive, in my case we used 1500e a year...we use(d) almost 600 euro on petrol, get my drift?
It is funny btw...i still must buy solar panels

Best
 
@100kwh-hunter Why would anyone buy up hundreds of these 18650 cells when it takes so many to get the V / Ah needed? I see this as a butt ton of work for marginal output. this doesn't even consider the connecting and containment.
 
Some people dont have so much money and want to start small/can not start big....
Some people do it for fun or as hobby..
Some people dont even know lifepo or other chemistry....
Some people(like me) build up a network to get e bike bats and laptop batts for nearly free.

In a year i accumulated ~8000-9000 cells for next to nothing, then i started to buy because the process was s o s l o w .
A lot of drop outs 2k to 3000 after testing....
In the resent year lifepo is becoming cheaper than bought & tested second hand cells.

Indeed it is a heck of a lot work......and a h***ish of a job to do in your spare time.
first drive to your contacts every week.
second harvest all the cells out there prisons.
inspect and test them.....(with the cells data specs)
In one of my first endevours i could test 10 cells a day, after building my own tester i could test 100 cells a day, full testing ir and soh.
My tester was build to find drop outs, heaters, slow chargers, low soh, under voltige, self dischargers ect
Match all the cells into packs
solder solder solder solder ect and on every cell a glass! fuse.
Inspect every year your packs and test the total capacity of one pack, repair or add when needed....
Yes a lot of work and not even the safest chemistry...
But a very good learning curve is involved in all the aspects of building a ess and the variety on chemistry.

I would do it again only for the learning curve.
 
yea, i didn't think of it that way... if one really wants to start, as someone on this board asked, I think i will start with these cells incase i screw it up. Its cheaper way to go and get your feet wet. HISTORY: back when I started making wine, i would drive to my regulars steak houses and restaurants to get my bottle collection started. I got lots, some diff sizes and shapes. Finally i had enough to stop driving for free bottles, and found a good supplier for new. I dont drive anymore. I hear you, great way to get started for cheap.
 
HISTORY: my mother drinks the same wine as my brother and his wife, yes we bought that wine also if we ran out.....
Is not much to start with, after several year the basement holds over 200 litres of wine that is aging from our own grapes.
Yes we started 15 years ago with buying a lot of apples pears grapes, yes we planted a 20-25 inch tall grape plant wondering if it will ever give a single grape.....15 years later..... 90 liters every year...
Its in our nature to get started cheap, it can be for a lot of reasons, especially when young, when you are getting older you laugh about it, all that wasted time energy and effort.....wasted..... heck no, learning curve!
 
Number of BMS's depends on your battery configuration. If you parallel a bunch of cells in to each serial string (like say 10p7s), you only need a BMS for every 7 series cell group (on a 24v pack), so that's one 7s BMS. In those cases they usually would put a fuse on each cell in parallel in case one shorted out it's protected (it can break off the p group).

Or you can have fewer unmonitored cells and use more BMSs (like say 7s10p), you have 9 more BMS's than the other configuration and much better monitoring on cell level. If you do use a bunch of 7s single cell individual packs (with a 7s BMS for each pack), and in some of Jehu's videos he has shown how in using those smaller BMS's they connect a parallel ribbon cable across all the BMS's in the pack stack so they can communicate on the same CANbus for easier management...
Sorry to bother you but, I have 3 tesla battery modules would it be better for 1 bms or 3 because I have 2 daly bms but don't like the balancing of them and I can get a jk bms with enough strings to do all 3. Can I do that?
 
Sorry to bother you but, I have 3 tesla battery modules would it be better for 1 bms or 3 because I have 2 daly bms but don't like the balancing of them and I can get a jk bms with enough strings to do all 3. Can I do that?

No bother at all.

Do you happen to be aware of the different lithium battery chemistries? I always tell people this whom are wanting to run automotive battery modules (li-ion, cobalt-based chemistries, etc), vs LFP (lithium-iron-phosphate), inside residential locations...

Cobalt-based lithium car batteries do have a higher energy density for their smaller size, but can also run risk of going into 'thermal runaway' if not closely monitored / protected. Car batteries like Tesla normally have water cooling built-in as well, which home users typically do not keep implemented in their home installation.

Just to be aware, that if these ever hit thermal runaway conditions inside your house, it could basically burn your house to the ground, since these fires are very difficult to extinguish. Better in my opinion to use an LFP battery type for inside house, or if using Tesla packs, then build a standalone structure for them to be in, detached from house. Because at least with a cellphone, you could throw it out the window if it catches fire.

Not to freak anyone out, but just something to be aware of.

More info about thermal runaway probabilities on the different battery chemistries can be compared here:


So, moving on to your actual question... Are your Tesla modules, complete Model S modules? From what I understand, they already have a slave BMS's built into the modules (requires a master BMS), and they could function if they were connected via CAN bus serial cable to a master BMS that knows how to talk Tesla comms protocol..

Such a device is available via EVTV:

Now it is not for me to decide if these are a cost-effective way for you to go or not, but just putting it out there.

This would be the simplest way to go with Tesla modules in my eyes. But I have never seen a Tesla battery module in person before (nor would I consider to use one for home use), so I cannot speak for the best way to try and mod it for non-automotive use. It may have to be disassembled down to the cells level to be able to wire in a different configuration using aftermarket BMSs on them. Not all packs give you easy access to the balance wires. Also the Tesla voltages may be way off for use with a 48v inverter, so you may need to reconfigure all the cells on the whole pack (how many in series groups).

Me personally, I have a 1p-Per-BMS policy, as I rather use large Ah LFP cells, and do a separate BMS for each 16s pack I build, but that's just me (like the magic recipe I like to roll with), plenty of people here don't think like me on this topic.

Hope this helps for something...
 
Last edited:
Back
Top