Parallelling 2 Li-ion batteries with different internal resistance

Hi,

my 36V ebike battery (Bosch) is 6 years old and has covered 25000km over that period. It still functions perfectly but the range has obviously dropped a little bit since new and I guess the internal resistance of that battery is pretty high by now. I'd be interested to use a new 36V battery softpack with its own BMS in parallel with the Bosch battery to increase the range. This has already been done by many people on a german forum (www.pedelecforum.de) so I know this would work. There are different ways to achieve this but one way which is appealing for different reasons is to open the Bosch battery pack and connect the external battery between the Bosch BMS and the cells directly (without ideal diodes). Of course the two batteries would be connected together only after making sure they have the same voltage. The external battery would use the same cells (Samsung 29E) as the original one but would have a larger capacity (e.g. 20Ah vs 11Ah).

Now my question: since the original battery is pretty old and the extender pack would be brand new, I guess the internal resistance of the extender pack would be much lower than that of the original battery. So it is obvious that the extender pack would provide most of the current, which is perfectly fine. However what happens when the voltage in the two batteries tries to equilibrate if the internal resistance of the two packs is vastly different? In other words, is it safe to parallel two batteries with vastly different internal resistances e.g. a 6 years old pack with a brand new pack even if the batteries use the exact same cell type or is this unsafe to do?

Thanks for your insight!

EDIT: the question applies to both discharging and charging. The two batteries would be charged together too by connecting the original charger to the original battery. I don't really see a problem for charging as everything would be watched by the two BMS. Is that correct? However during discharging, if the extender pack has a slightly higher voltage at some point (quite unlikely I guess since it has a lower internal resistance), it would charge the original pack without going through the original BMS because it is connected between the BMS and the cells. On the other hand, if the Bosch pack has a higher voltage than the extender (e.g. after a load phase with most of the current being provided by the extender, it would charge the extender but would go through the extender's BMS so it should be safe...

Do the BMSs on the packs have separate charge and discharge ports? Or are they the same port?

If the former, then to charge them you need to disconnect the discharge ports (or use diodes to do the paralleling. To discharge them you need to disconnect the charge ports. Otherwise the BMSs cannot shut off current flow between the packs.

If the latter, then they can safely be paralleled and left that way for charge and discharge, as each pack's BMS will manage that.


If you're just literally paralleling the cells themselves, and letting the single BMS deal with it, then there's no issue there, either, as long as none of the older cells have leakage issues (which can drain them and any cells paralled to them). So as long as the original cells have been staying well-balanced, they don't likely have this issue.

amberwolf said:

to charge them you need to disconnect the discharge ports (or use diodes to do the paralleling. To discharge them you need to disconnect the charge ports. Otherwise the BMSs cannot shut off current flow between the packs.

Click to expand...

Sorry I don't follow. Surely the whole point of paralleling is to create in effect one larger bank out of the sub-packs.

Why would you want to shut off the current from flowing between packs?

Yes, both sides need to be at similar voltages before joining. . .

john61ct said:

Why would you want to shut off the current from flowing between packs?

Click to expand...

Because the BMS can't protect against overdischarge / overcharge / overcurrent unless the charge and discharge ports are the same.

If they are separate ports, then you either should have diodes on the ports to prevent current flow between them, or the charge ports should be disconnected from each other while discharging, and the discharge ports should be disconnected while charging.


If you don't do this, then let's say you have a pack that's got a weak cell (group), and you're charging the packs from a common charger, with separate BMSs that have separate charge and discharge ports.

The weak-cell pack's BMS shuts off it's charge port because that cell reaches full voltage (before the rest), but it can't shut off the discharge port as an input (its not designed to), so that port, still paralleled to the other pack that is still charging, lets current flow into the pack with the already-full cell.

So that full cell continues to charge (along with the rest of that pack's cells), and is now overcharging. If it's severely imbalanced compared to the other cells, and got full when the other cells are still really low in voltage, then it could overcharge by more than a volt or even several volts, which can create a fire hazard.


Similarly, same situation, but discharging. Say that low cell has less than half the capacity of the rest. Packs are discharging, and that cell reaches empty, so that pack's BMS shuts off it's discharge port to protect that cell from overdischarge.

But it can't shut off the charge port as an output (not designed to), so that port, still paralleled to the other pack that's still dischargeing, lets current flow out of that pack with the already empty cell.

So that cell continues to discharge, overdischarging, and at some point could actually reverse becuase of the current flow thru it, which agian can create a fire hazard.


(either situation *could* result in an immediate fire...how likely that is, depends on the circumstances, actual cell damage, and properties of that particular brand/model of cell and it's chemistry).

OK I guess the problem is that the BMS design is assuming only one battery.

To protect effectively when the string is paralleled, the separate-port type must be designed for that.

Seems like the common-port design has a big advantage there.

What are its disadvantages?

Do you know of links to resources that compare the two types in detail?

amberwolf said:

Do the BMSs on the packs have separate charge and discharge ports? Or are they the same port?

Click to expand...

I think the original battery (Bosch) has a common charge/discharge port because the positive and negative terminals are the same. However it uses an additional 5V connection while charging, probably to let the BMS know it is connected to the charger. So during discharge, the 5V connection is absent and the BMS might not be able to protect the cells from being overcharged by the extender pack. In fact, since the extender would be connected between the BMS and cells on the original pack, there would be NO protection against overcharging cells in the original pack from current flowing from the extender pack anyway? Is that correct? So I guess it's a no go situation??

Regarding the extender pack, all I know is that it has different charging and discharging cables. So I guess that means different charging ports. But in my setup, the extender would have to be charged using the discharge port since it must be paralleled with the original pack on that port. Then again I guess this means the BMS on the extender would not protect the cells from being overcharged by current flowing from the original pack...another no go situation?

Well I thought I may have issues with different internal resistances but it seems the problem is far greater than just that. Might be safer to use a battery switch and use each battery one at a time then. Not as good as a parallel setup but I don't want to risk anything if it's not 99.999% safe!

What about ditching the extender pack bms and parallelling all the balance connections between the old pack and new. Effectively making one big pack controlled by the Bosch bms, which is presumably of higher quality than the one on the extender pack. It would require modifying the Bosch battery (which is out of warranty anyway) and might be a bit messy with wiring.

district9prawn said:

What about ditching the extender pack bms and parallelling all the balance connections between the old pack and new.

Click to expand...

Well, not sure I have the skills to do that myself honestly... Also the Bosch BMS is tricky to deal with and I'm not sure it would accept to see a pack with a larger capacity. They've tried to lock their proprietary system as well as they could to force people buy new Bosch batteries. It might work but I'd have to have someone do it for me as really I don't feel confident doing it myself.

That last method is basically going to just make you a bigger battery, so there's no worries about how the BMS operates.


BUT: I wouldn't just add balance wires to the new pack and connect them to the old one. I'd physically parallel the cells to teh old ones with current-carrying connections, so that the new cells hold up the old ones under load directly, rather than just thru the main pos/neg connections.

This basically requires putting the new and old cells into the same physical container.

OK so the last method is probably the best then. However is there really no problem to parallel old and new cells with vastly different internal resistances in a same pack? No side effect to this setup as far as safety is concerned?

The alternative then would be to ditch the old cells and use the Bosch BMS to monitor new cells in a larger pack (basically just the extender). Maybe that would be the safest option afterall.

@amberwolf I read your blog on you house's fire back in 2013...scarry!!!

kilou said:

OK so the last method is probably the best then. However is there really no problem to parallel old and new cells with vastly different internal resistances in a same pack? No side effect to this setup as far as safety is concerned?

Click to expand...

There are always potential side effects. The ones in this case should be minimal, as long as none of the original cells are leaky (meaning they drain cells in paralle with them). If the pack is operating normaly but just has less range than it used to, that's unlikely, so should be safe to do.

The alternative then would be to ditch the old cells and use the Bosch BMS to monitor new cells in a larger pack (basically just the extender). Maybe that would be the safest option afterall.

Click to expand...

You might have to wire the new cells in parallel first, then disconnect hte old ones, because some of these OEM BMS systems will brick (shutdown permanently) if disconnected from the cells. I don't know how the Bosch BMS works, so it might or might not do this.

@amberwolf I read your blog on you house's fire back in 2013...scarry!!!

Click to expand...

Rather than scary (for me, anyway), it was devastating...losing the dogs still haunts me, and I still have nightmares about it, and them, though the dogs I have now (Yogi, Kirin, PeanutButter, and Jelly) and two others I had along the way after the fire (Tiny and Teddy) help quite a lot.

But I suppose it is a bit scary to realize you might leave for lunch or whatever) and come back to what I did....

@amberwolf yes sorry, the word "scary" was poorly chosen, my english vocabulary is quite limited (I'm from Switzerland). It must have indeed been devastating. I'm happy your current dogs are helping you getting better.

Regarding the battery stuff, I'm slowly getting convinced that paralleling the two batteries like you proposed i.e connecting each block in parallel (20 wires) and not just the two main terminals, is the best option. With this, the Bosch BMS would monitor both packs at the same time. I'm also pretty confident now I could even do it myself...after getting a refresh on electricity of course! And I know that it is ok to disconnect the Bosch BMS from the battery on my generation (1st generation Bosch). Some people have already done that without adverse events. Depowering the BMS is an issue only on the newer Bosch systems as far as I know (from 2014 on). Anyway I wouldn't even need to do that if paralleling the twi batteries.

However I'm still unsure about the safety of mixing old (6years, 25000km) cells with brand new cells who are able to deliver much higher current I guess... For instance, there is a clear warning about this here:

https://electronics.stackexchange.com/questions/74459/why-is-it-bad-to-mix-new-and-old-batteries

Sounds like it's not a wise thing to do. Is there any additional simple safety measures that could be used to mitigate potential problems. For example would it be advisable to add a resistance in the wire connecting each new and old blocks of cells in parallel in order to make the new cells look that they have the same internal resistance as the old ones? Of course that would make the pack less efficient but...

Or is this "bad to mix old with new cells" stuff only a problem when the old and new cells are connected in series but that it would be perfectly safe to do so in parallel??

Does your system REQUIRE the Bosch BMS in order to be functional?

You have an old, tired, formally 11 ah pack which now has a much reduced capacity. You plan to get a new, 20ah pack, and combine the two.

The original pack will likely supply about one-quarter the range of the new pack, which amount will decline further over a fairly short period of time.

At some point, the original pack becomes more dead weight than functional usability.

Combine that with the effort, hassle, possible danger, loss of warranty on expensive new pack due to modifications required to combine it with the old pack of negligible value, I submit that the minimal gains available are outweighed by the negatives.

Get another motor kit and use the old pack on that, have a backup grocery-getter.

Well I totally agree but that's the problem of proprietary ebike systems: the Bosch motor can only work with the Bosch BMS because the two communicate via CAN bus. You cannot use the bike without a Bosch BMS. So this is why I'm trying to figure out how to increase the range with this limitation in mind. Ditching the old cells is an option but it requires unsoldering a very thin ribbon cable with a high risk of damaging it. This is also a piece that is not provided as a spare part. That's why parallelling two batteries seems like the easiest option. However I won't do it if this is unsafe or even slightly risky.

The Bosch motor/battery is not a kit, it's a complete system. Even the bike frame is made on purpose. The whole thing works perfectly and is very reliable so I don't see myself ditching it to get an ebike kit and a new frame. My goal is to see if there is a reliable, simple and safe way to increase the range. If that's not possible, I'll get a new 11Ah Bosch battery that will last me another 6 years so that's more than fine too...just costs 500$.

This video illustrate the process of recelling the Bosch battery with new cells.

https://m.youtube.com/watch?v=PO0BQssvGEU&t=1s

At 3:20 you can see the removal of the ribbon cable that monitor cell voltage for the BMS. This is the problematic part for me. I wouldn't recell the battery as in the video because I don't have the skills and equipment. Rather I'd buy a new 20Ah pack without BMS but that already has the nickel plates soldered. So appart from connecting the battery positive and negative to the BMS, all I'd need to do is to connect that ribbon cable to the cells to have them monitored by the Bosch BMS but I'd need to solder individual wires that connect the end of the ribbon to each group of cells which is not so easy I guess... But if that could be done that would be it. Then obviously I should design a proper case for the battery but it looks doable.

Do you know if it's possible to convert a flat ribbon cable such as that in the video into separate wires?

Having dealt extensively with proprietary hardware in another industry, that is why I avoid it like the plague and heinous evil foisted upon mankind that it truly is.

Now, that is not just a ribbon cable, that is a Mylar ribbon cable. Fairly fragile and delicate, and IMO a royal PITA to deal with. SFAIK, it is at least theoretically possible to separate the ribbon into individual leads, but it would take a very steady hand and very sharp vision, neither of which I possess. What MIGHT be possible would be to substitute something like an old PC parallel drive cable, which would be dramatically easier to work with. The edge connector at the end would be an issue, but I have seen adapters or installations where such conversion has been made, may have been custom.

Did you notice, at the end of the video, the customer is no longer present, and there is no definite indicator that the battery is actually functional? Dude appears to know what he is doing, I am not claiming otherwise, just saying.........

kilou said:

However I'm still unsure about the safety of mixing old (6years, 25000km) cells with brand new cells who are able to deliver much higher current I guess... For instance, there is a clear warning about this here:

https://electronics.stackexchange.com/questions/74459/why-is-it-bad-to-mix-new-and-old-batteries

Click to expand...

AFAICS they're referring to series connections of them, like if you were to take the whole old pack and the whole new pack and put them in series (rather than parallel), or if you had a device that used several series D or C cells or something like that, changing just one of them can have the results that link discusses.

Paralleling the cells simply means that the low resistance cells provide proportionally more of the current under load, and during the lower-load portions of the ride all of the cells will keep each other balanced within their respective groups.

There's no current being pushed thru higher resistance cells by the low resistance cells in your situation.

Is there any additional simple safety measures that could be used to mitigate potential problems.

Click to expand...

Not really, other than verifying that the original cells appear to be relatively well balanced after a long ride when they're close to empty. If they are, then they're all basically the same as each other, and there should be no issues.

For example would it be advisable to add a resistance in the wire connecting each new and old blocks of cells in parallel in order to make the new cells look that they have the same internal resistance as the old ones? Of course that would make the pack less efficient but...

Click to expand...

No. You never want to add any resistance, it just makes more heat and more waste, and could potentially cause problems keeping stuff balanced.

@AngryBob: yes the video doesn't make it clear that the battery works at the end but I'm pretty confident it does. I've already read about other people having successfully recelled the generation 1 Bosch battery. I wouldn't risk it with a gen2 though but mine is gen1.

Regarding the mylar ribbon cable, I'll try to see if it's possible to cut it and attach another e.g PC keyboard connector at the end and solder thin wires from that connector. That could be an ootion I guess. Thanks for the suggestion.

amberwolf said:

Paralleling the cells simply means that the low resistance cells provide proportionally more of the current under load, and during the lower-load portions of the ride all of the cells will keep each other balanced within their respective groups.

Click to expand...

My (very limited) understanding is that, in a group of parallel cells being put under load, if new cells provide more current than older ones, then their voltage will drop slightly below that of old cells. During a low or no load phase, the voltage will balance, meaning that older cells will provide current to the new cells to balance them. That current will likely depend on the voltage difference between old and new cells, which in turns depends on the difference in their internal resistances I guess. Now wouldn't that current potentially be larger than what the old cells can sustain, making them heating away without the BMS having the ability to know about the intensity of such current?

Just to educate myself: What happens during charging in this setup? In a group of parallel cells with old and new cells, won't the newer cells with a lower resistance be charging faster than the old ones? So wouldn't they be overcharged while the older cells are still charging, with no means for the BMS to know about it since it doesn't monitor cells individually but rather monitors each group of parallel cells (comprising old and new cells) as a whole?

Thanks for all your replies so far! Very helpful and informative!

No. A group of cells in parallel all stay at the same voltage, whether getting charged or feeding a load.

Their **relative** SoC is the same, since they are the same chemistry, **but**

the newer / healthier (higher mAh capacity) cells are seeing greater energy in and out per Voltage-change increment.

AKA "working harder" "good supporting bad"

That is what is "naturally balancing" the group, once they've remained connected in parallel for some time, there is no significant current flowing between the cells.

The newer cells will be wearing out faster joined with the older ones, than they would if all were in the same condition.

kilou said:

Now wouldn't that current potentially be larger than what the old cells can sustain, making them heating away without the BMS having the ability to know about the intensity of such current?

Click to expand...

No. The difference will be millivolts, the resistances milliohms, the currents involved will be milliamps to less than an amp--if the cells can't handle that then they're so worn out they should be replaced completely anyway.

Just to educate myself: What happens during charging in this setup? In a group of parallel cells with old and new cells, won't the newer cells with a lower resistance be charging faster than the old ones? So wouldn't they be overcharged while the older cells are still charging, with no means for the BMS to know about it since it doesn't monitor cells individually but rather monitors each group of parallel cells (comprising old and new cells) as a whole?

Click to expand...

All the cells will remain at the same voltage, and the current will be distributed by the cell resistances, just as in any paralleled-cell pack.

Since the voltages remain the same (because they're paralleled by what amounts to zero resistance under these conditions), the BMS still does the same job it did before.

Since the new cells have more capacity than the old ones, and their resistance is lower, they'll absorb more of the current than the old ones, so the charge will basically still go like it would normally, except it will take longer because the new cells are taking much more current than the old ones, since they have a lot more capacity.

Keyboard cable would not have enough wires, a VGA cable would.

Looking at the amount of work required to parallel as you are suggesting, and considering the issues involved, might it not be more beneficial and not very much additional work to just recell the pack, and have a battery which would last many years, instead of slightly less work to achieve a much heavier pack which would need rebuilt again in a much shorter period of time?

You could also just buy the 36V 4.4aH (~150wH) scooter battery packs and make a new battery bank. Each pack has its own BMS and this design is modular. Build yourself a new bigger battery box capable of expansion and there you go. I use this in 72V mode and had one go bad on me. I simply replaced that one for $30-45 and all set to go. Especially for 36V speeds, these things are solid. 72V speeds not so much but acceptable for the price.

Hi,

I would like to use a second no brand battery in parallel with my bosch battery. But I would not connect it trough the BMS. I would just use a separate pack in parallel with an ideal diode in between. Would this still work with the current bosch line of motors?

I've already done it with a bafang motor, works great. I would do it with a bosch motor for a friend (won't ever buy bosch myself, I hate locked systems).

I've been lookin around on the net but I can't find the answer.

Thanks in advance!