Alternative BMS concept, modular Banks

Hello to all,
My diy powerwall description is located here.
https://secondlifestorage.com/t-Grid-Tie-modular-and-flexible


Defined as something modular and flexible I chose a different approach in organizing the cells.
The usual Pattern I see here is putting a load of cells together to an 1sxp block together, say a 1s80p. Then a couple, say 7, of these 1s80p blocks are connected in serial to a 7s80p. (Shown are 4 Blocks)





My approach described in above link is to block a number of cells, say 48, together as a 6s8p block.
(The Picture shows 3s2p Blocks).





This block has to be assumed for the moment ats being properly protected and balanced.
Building several of these I can put them in parallel, on the level of the positive and negative connector.
(Thats the status of the picture)

Having a modular system now. I can pull out any of the blocks at (nearly) any time, even under load, and the others will take up the load and continue to supply. By this I have easy manageable block sizes (in weight and in size), which can be pulled out and serviced quite easily, while the bulk of them are still in operation. I can start with a smaller block count and add more of them later, giving the chance going on-grid quite early and upgrading it step by step.


So far, so good. Now, how to solve the above mentioned protection and balancing of each single block? If I try to supply any sort of state-of-the-art balancing I have to provide that for each block, which is surely neither a simple nor a cheap solution.


Leaving all considerations aside which i have done to come ti this, pls think about the following.

• I put a cheapo BMS without balancing to eack 6s8p Block. This BMS will check cell-exact the voltages and cut the supply out (for this block) on cell under- or overvoltage.
• I connect all cell connection of same xs Level to a Central point, attaching a resistor of (say) 2 Ohms. This central point is in the main surveillance station of the whole pack, it is not near the cells.
• From this CP i cant connect to any cell level of all packs simultaneously, but only over the 2 Ohms resistors.
• This CP provides a different type of balancing. Instead of surging an Pack in the usual style, or a cell level in my modular struckture, i provied to supply that level with loading current.

What do I get now ?

• All cells of same level are connected via the resistors. So they level out their voltages to the same level, assumed to the same SOC. (like all cell in an xp array)
• All Modules (my structure) have a cheapo BMS to safety protect their cells.
• All levels are connected to the Main control to being balanced from there, over the resistors. There is no surging BMS near the batteries, there is no heat near the batteries, there is no surging at all, no additional losses.
• Balancing is done by charging a cell level, instad of discharging. That means it can and will be done any time i like, not (only) at top voltage level to prevent overcharging.
• The connections between the modules run over resistors, which limit the currents, so i can use cheapo JST XH connectors, qualified for 3 Amps. Means I can (theoretically) charge quite more, because the current is distributed (supported by the resistors) over the blocks.
• There are 2 possible reasons why I have to balance: The blocks are out of same SOC by history, or they contain bad cells surging power for their losses. The latter would cause the first as well. So, when I charge a Level/Block I target the possible reason for that directly. Otherwise I had to surge all other blocks for the same amount of energy . and dissipate that.
• When loading targets the questionable blocks/levels directly, a logfile about the balanced levels give significant info about seviceable levels.
• when cells in one of my module heavily go out oft sync, the Module will be cut-off by the BMS. Surveillling the current of each block will show cuts-offs and point to Errors immediately. Pulling the block out gives an 6s8p array to be investigated (where only one p-row ist in question, regarding Balance protocols), which would give little problems.

So far, so good.
Question now from me, what do you specialist think of this approach?
Any comments on safety, on efficiency, advantages and disadvantages ?
All comments welcome.

I have many issues with the way things are done by powerwallers , but one thing I believe is correct is to have large banks of cells in parallel.

Once you get a few hundred wired into one 3.7 V unit , and a number of these in series to make your wall , there is no need for balancing , a voltmeter on each unit will show that capacity has ended up virtually the same in each parallel block , due to random selection of cells, and the voltage of all the packs stay very close ...

You say you can now easily remove parts for maintenance ...what maintenance ??? there should be none ... and how would you know if it was needed???

Putting resistors in is the last thing I would do ...they're wasting power and heating things up.

I have a similar approach. I use a monitor not a BMS for now as my wall is only 4.6Kwh in size. I have small packs of 8 cells (1s8p) that are mounted to a back plane to make up 7s and multiple of these packs in parallel. I make the 8p packs in sets of 7 that as as close (tested before and after assembly) so when I install them in my wall the entire wall stays in balance.

This allows me to remove one packs or more and the wall stay running.

Now I only run my batteries from 4.1V to 3.6V and this negates any low or high cell issues as they seem to stay withing 10-30mV separated. I only notice any imbalance when the pack gets over the 4.1 per cell or below the 3.3V per cell.

When I do see one cell start to become too imbalanced I use a resistor Bank (3 Ohm 10W resistors for a 1A ish draw) to pull down the highest pack. This has only been needed twice as cell 7 was getting to 4.15 when the charging stopped. So I put 4 of these resistors on that one cell and brought it down to 4.0v (when removed the load to rested back at 4.11V) but the last time was about 3 months ago.

ozz93666 said:

I have many issues with the way things are done by powerwallers , but one thing I believe is correct is to have large banks of cells in parallel.

Once you get a few hundred wired into one 3.7 V unit , and a number of these in series to make your wall , there is no need for balancing , a voltmeter on each unit will show that capacity has ended up virtually the same in each parallel block , due to random selection of cells, and the voltage of all the packs stay very close ...

You say you can now easily remove parts for maintenance ...what maintenance ??? there should be none ... and how would you know if it was needed???

Putting resistors in is the last thing I would do ...they're wasting power and heating things up.

Click to expand...

Oz, i didnt try to consider things "right" or "wrong". I am sure that the large banks structure has its features and advantages, otherwise it wouldnt be used so commonly.
I just try to make things different because it doesnt have some feature which I consider important for myself. I do not expect that everyone agrees.

Thanks for your input. Ill make my mind up regarding the losses of the resistors, which might show up dynamically during loading and unkoading. This is a loss which seem not to be present in thight connected banks, yet it might be it is hidden in the currents betweem cells which show different inner resistance. Very interesting point.

---

jdeadman said:

I have a similar approach. I use a monitor not a BMS for now as my wall is only 4.6Kwh in size. I have small packs of 8 cells (1s8p) that are mounted to a back plane to make up 7s and multiple of these packs in parallel. I make the 8p packs in sets of 7 that as as close (tested before and after assembly) so when I install them in my wall the entire wall stays in balance.

This allows me to remove one packs or more and the wall stay running.

Now I only run my batteries from 4.1V to 3.6V and this negates any low or high cell issues as they seem to stay withing 10-30mV separated. I only notice any imbalance when the pack gets over the 4.1 per cell or below the 3.3V per cell.

When I do see one cell start to become too imbalanced I use a resistor Bank (3 Ohm 10W resistors for a 1A ish draw) to pull down the highest pack. This has only been needed twice as cell 7 was getting to 4.15 when the charging stopped. So I put 4 of these resistors on that one cell and brought it down to 4.0v (when removed the load to rested back at 4.11V) but the last time was about 3 months ago.

Click to expand...

Very interesting. You are not so far away from my "modular" intentions. I already saw many times the advice not to use the full voltage swing, while people use different values for different reasons.
Can you tell what imbalances occur? Is it one cell being to high, or can it considered to be one too low ? I guest you use a fixed voltage to load, is that correct ? can you imagine to load the lowest pack instead of charging the highest?

I have found some of my cell fuses to have broken due to flex in the bus bar. So basically my fault causing a decrease in capacity in that cell. So it would discharge quicker than the rest and then charge up faster. So I have just recently taken half of the wall own to test. Found my mistakes and 1 bad cell that I missed was a self discharging cell and replaced that one cell.

It's been great now that i re-installed that half and I have taken the other half off and I'm testing those packs for broken fuses and bad cells

I have a build in progress with 2kWh modules that are 11s2p (LTO cells, 23.V nominal) and these are connected together via DB15 connectors through0.27Ohm resistors and 3A Axial fuses. The reason for the low resistance permanently connected is the overall energy transfer on a 24x7 basis for packs that are "relatively" balanced the continuous flow should not be over 1A. The issue you may see on smaller packs is with high loads the packs may see larger voltage differentials appear under load and this will cause larger transfers to occur.









See my post in my LTO Powerwall for more info.

The downside for me is that I had to solder up 225 wires just for the pack connections and still have another 300 or so to go for the full setup and testing equipment, which I should have some images of in a couple of days.

The way you describe what you want to do is pretty much the way I'm going to set mine up as.

Basically you are create "System Voltage Packs". Each pack runs at the system operating voltage. This allows for easy maintenance of any of the packs without disrupting the whole system.

With this setup, it is needed to have a bms control for each pack. But it needs to be very smart and adaptive, and can be synced with the other packs. Each pack will have it's own balancing board so the boards wouldn't need to have to handle 5A per pack per series in the pack. It could easily be 1A, so they could be small and efficient. Also easier to keep the packs in balance that way as well.

I'm currently working on this type of bms. I've ran into a few issues (mainly shorting out, programming errors, or bad contacts on the breadboard) that has set me back a little bit. I have new parts in the mail on the way here and continuing to work on the pcb design.

There are a few other designs I have in mind that I would like to implement in the system as I expand it's functionality. It takes time to build these things

"It takes time to build these things" - so very, very true.

I have tried out cross connecting 2kWh packs (what would be adding in a pack after it has been out of service) with only 100mV difference for 11s2p (i.e. 26.1V connecting to 26.2V, they are LTO cells) and get 10A flowing. This made me realise I need to add in two volt meters to each pack that are visible when connecting/switching as a 3V difference would cause slight issues to say the least. From the test cell data a difference of 100mV per cell is 40A per cell for discharge, so 3V over 11s would end up with well over 80A flowing from the pack.

One volt meter for the cells and the other for the voltage on the other side of the breaker/switch - in your face readings and not via the computer AND then double check with a multimeter each time. I'm getting paranoid as the batteries have no "off" switch. Will have some more test readings soon and some pictures...

The DB15 connector is good for 5A per pin and I think it's 250V or 500V rated, but realistically at 5A I think it would melt the internal plastic... the cables sodlered to it are 0.75mm2 (24/0.2 stranded). Cheap, electrically proven since the 80's and can be locked in place with two screws.

Hi completelycharged, hi Korishan, i see now that my idea of moduling and cross-connectiong is not so outrageous (and bad) as i was afraid of.

Very interesting infos, i will comment on them in detail, sorry at moment very busy.

---

completelycharged said:

I have a build in progress with 2kWh modules that are 11s2p (LTO cells, 23.V nominal) and these are connected together via DB15 connectors through0.27Ohm resistors and 3A Axial fuses. The reason for the low resistance permanently connected is the overall energy transfer on a 24x7 basis for packs that are "relatively" balanced the continuous flow should not be over 1A. The issue you may see on smaller packs is with high loads the packs may see larger voltage differentials appear under load and this will cause larger transfers to occur.

Click to expand...

Wonderful info. Your module size ist just 10 times of mine, and the reasons (and problems) of resistor in the crossconnections show up even more.
Sorry, i do not have at moment a circuitry CAD system running, so i refer to your plans.

I do not have fuses. My resistors are sitting directly at the pack. Thats saves some solder joints, but that is not the point. I think of being the resistors my fuses as well, since the will burn on a dangerous overload anyway. Since burned resistors or fuses are hardly detectable (by measuerment), i would prefer that these never burn at all, or at least only in safety situations.

For the size of the resistors, there seem to be taken in account different situations, normal use, plugging a relatively unbalanced pack to the live one and the situation with a single bad cell.

I personally like this approach. I am in a situation where I have reasonable quantities of different types of cells. 18650, 3000mAh pouches from newer laptops, 10,000 to 15,000 mAh pouch cells from eBikes, etc.

I am settling on the conclusion that I will need to build system-voltage packs of a single cell type, and then connect them all in parallel. This mightalso allowfor a system of mixed chemistry, so long as the chosen system voltage range is acceptable - not part of my plans at the moment however.

I have also been thinking of building fully self contained SV packs and housing them in these... They are cheap, strong, kinda fire-proof and stackable. Dimensions are32" x 11.25" x 6" which I figure provides enough volume for 3-4 kWh of cells and some BMS etc circuitry.

Overload = Burn = Fire <> Safety

For me, fuse everything. A blown fuse is far better than the smell of burning insualtion. The time is worth it.

I have played around (in the 1980's) setting fire to 10ft lengths of house wiring twin and earth with a few hundered amps and it moves, swels, smokes then shoots flames everywhere within about 2 seconds. The length of the wire istelf reduced the amps to a few hundered because of it's resistance and my transformer was rated around 7.5V at 750A I think, weighed around 20kg. This is not something I would never like to occur in a battery pack and axial glass fuses are excellent value.

Does not matter what voltage you connect at as long as the overall voltage differential is low. Connect the main power bars first, let the balancing occur and then connect up the BMS lines when the main power imbalance has reduced to less than 1 Amp.

Balancing needs to be done in relation to the "energy" in the packs and not the voltages, but few BMS units do this as it requires monitoring of a few cycles without balancing to evaluate the pack behaviour in terms of self discharge, charge efficiency and discharge efficiency.

I think I am getting confused....


We are talking about balancing betweensystem voltage packs - correct?

As these packs are in parallel with each other, they balance by design. Each pack will always be at the same voltage, and as load is applied to the system, the energy supplied by each pack will be in proportion to its total capacity. One pack will not discharge quicker than any other, regardless of capacity. If this DID happen, then there'd be packs in parallel at different voltages, which is pretty much impossible unless the resistance between them is high.

As far as I can see, the only balancing required at system-voltage level will be when adding a new pack to an existing system. Even then once it is at the same voltage as the rest, there will be no current flow between packs. No potential difference = no current flow. Until such time as the voltage of the new pack and the rest of the system equalize, there is the potential (ha-ha) for a LOT of current flow.

I believe that this is where resistors or some other 'alignment' mechanism is required to bleed from the high and add to the low while maintaining a safe current level.

Consider a 12V system of 2 packs. One pack is a 10Ah lead-acid 6s1p arrangement, and the other is a 200Ah 6s1p. If these packs are both at the same voltage then they can be connected in parallel with no current flow, and together, provide a 12V 210Ah 'system'. It is the aligning of the voltages that could get ugly if mishandled.

Of course, my example packs are just 12V lead acid batteries, but if the packs are 18650 based, in a 14s config, then we have a 48 volt system. So long as each pack maintains its integrity, whatever is happening on the inside is unrelated to what happens on the outside.

Am I getting this correct?

If I AM correct, then it has just dawned on me that a nominal 48V system could be built using different chemistries by simply ensuring that packs are switched in and out of the system once their voltage limits are reached and this can all be handled by MOSFETs.

It's been a long week - I may be smoking something crack, but if nothing else, I am learning something....

Grumplestiltskin: Balancing system voltages packs, yes. However, there are two different methods. 1 method is connecting absolute Pos/Neg in parallel with another system voltage pack. This is the one you describe. The other method is not only absolute Pos/Neg, but also interconnect all the cells that are in series with the pack adjacent to it. So there would be X+1 number of connections per series.
If the packs are 7s, then you would have Parallel connections:
- Cell 1 | Cell 2 | Cell 3 | Cell 4 | Cell 5 | Cell 6 | Cell 7 +

So, Negative would be connected in parallel, as would the connection between Cell 1 and Cell 2 be connected to other pack. This would basically create a double wide pack. Electrically speaking, it'd go from 7s10p to 7s20p, for an example. The former setup would be 2x7s10p, electrically speaking.

I hope I didn't muddy the waters up with that

Korishan said:

The other method is not only absolute Pos/Neg, but also interconnect all the cells that are in series with the pack adjacent to it. So there would be X+1 number of connections per series.
If the packs are 7s, then you would have Parallel connections:
- Cell 1 | Cell 2 | Cell 3 | Cell 4 | Cell 5 | Cell 6 | Cell 7 +

Click to expand...

Ahhhh.....I understand now. That would of course not apply to different chemistries, but I got it now...

A question then - assuming the chemistries are the same and the number of cells in series are the same, then isn't it a little redundant? I mean, the internal balancing of each pack SHOULD have every cell the same, and if the absolute voltages are the same then it follows the individual cells are the same.

I guess it would be possible to monitor the current flow in these parallel connections, which in a perfect world would be zero. If it weren't, then that would be a sign of a weak cell group or a BMS issue in the pack that is getting out of balance. I also see where the resistors come in - these parallel connections are not intended to carry system-level current.

My concern would be that if a pack goes rogue, with all these interconnections it is that much more complex to isolate it from the rest. Puling a DB25 plug is no big deal, but managing it electronically is another issue. The BMS in the bad pack should be able to isolate it at the system level, but then you'd need something similar for each parallel connection.

Thank you all (esp. Korishan) for stepping through this with me. I don't mean to hijack the thread, but I think these are questions that are directly related.

Grumplestiltskin said:

Ahhhh.....I understand now. That would of course not apply to different chemistries, but I got it now...

A question then - assuming the chemistries are the same and the number of cells in series are the same, then isn't it a little redundant? I mean, the internal balancing of each pack SHOULD have every cell the same, and if the absolute voltages are the same then it follows the individual cells are the same.

I guess it would be possible to monitor the current flow in these parallel connections, which in a perfect world would be zero. If it weren't, then that would be a sign of a weak cell group or a BMS issue in the pack that is getting out of balance. I also see where the resistors come in - these parallel connections are not intended to carry system-level current.

My concern would be that if a pack goes rogue, with all these interconnections it is that much more complex to isolate it from the rest. Puling a DB25 plug is no big deal, but managing it electronically is another issue. The BMS in the bad pack should be able to isolate it at the system level, but then you'd need something similar for each parallel connection.

Thank you all (esp. Korishan) for stepping through this with me. I don't mean to hijack the thread, but I think these are questions that are directly related.

Click to expand...

Correct, use this method for only identical chemistry and voltage ranges.
It could be redundant, however, it adds a layer of balancing to the mix. Plus, it allows you to narrow down how many bms balancers you need. You could basically have 7 balancers (for a 7s setup) and no matter how many extra packs you add to the parallel banks, those balancers take care of them. Of course, there is a limit to how much the boards can handle.

The resistors are really for taking care of any surges that might occur while connecting up in parallel. Some connections may make contact before others, and this probably helps. Also, like you said, system current is limited through these.
But you would never want to connect a pack to the system that is widely out of range with the others. Say, 1V or more at the parallel level, not the full pack level.

There are other safety measures you could employ to narrow down a rogue cell. You could put small current sensors between each pack on the balance leads. That way you could see which direction the current is flowing. On those sensors, you would essentially want to have 0.00A current flow, except during balancing which all of them should show current flowing in the same direction, and approximately the same value (or would it be cumulative? from pack 1 to 2, 5mA, pack 2 to 3 5mA + 5mA, pack 3 to 4 10mA + 5mA, etc, ec) Hrmm.

I'm sure someone else will chime in and add some more info, or at least make mention.

I'm balancing the packs by connecting them together with 4 x 8AWG cables -for the main power -and the secondary connections for cell balancing via DB15 connectors and the 3A fused 0.27Ohm resistors. This makes for a lot of wires, 14 out of each pack (2 +ve, 2- ve and 10 cell lines) and an approach that for my use and application makes sense because I will be using the packs at home and away quite frequently. The cross connection at the cell level also allows for the option of just one BMS.

If you use different chemistries the pull from each pack will be very different and you may have issues with cross balancing flow after a large load has been switched off. Recovery voltage for each chemistry is different and so is the internal reisistance. Plus each cell chemistry has a different voltage storage range. I would not mix chemistries on a pack unless they are isolated or as a feed in/out buffer, i.e. your lead acid is only used if the main pack level drops to your reserve level and only then would that cell discharge/flow into the main pack.

Lead Acid == 250 cycles
Lithium == 1000 cycles

Mixing these two just means you destroy the lead acid early.

Hello, sorry for my silence. No time. I will for sure contribute details of my construction.

Hi all, new here, so forgive any stupid questions etc.

Did anyone get anywhere with this concept? I am aiming to convert an apc surt10000xli ups to lithium. The 4 battery cartridges run 96v nominal lead acid with 4Ah batteries. Conversion to lithium willbe 27s and I think I can cram 5p into the cartridges. Problem, the cartridges are long and thin so I need to arrange 27s1p blocks nose to tail within the cartridge. Does anyone know how I would sort a bms to deal with this setup, and an easy way to do the parallel connection bit?

Thanks

Oooo, interesting scenario. How do you plan on storing them? Are they going to be visible, or inside a tube of some sort?
If the tube, then I think you could break the tube in sections, and have a connector at each section that runs a wire to each cell. Maybe every 3 cells as that would 9 sections. Then you can tie all 5 tubes together at those junction ports effectively making it a 27s5p.
If making a straight run for each string, then it's a little more difficult as they would be a bit more flimsy.

In regards to the concept above, I don't know so I can't comment on that part.

Post a new thread with question about the UPS setup and include pictures. That way you'll get more interest as everyone might not look at this particular thread (sad, I know, but true )