[OR-Carl]

Hello all! Like many others on here, I am contemplating splitting an OEM pack in half to get a 144v nominal pack with the same Kwh capacity as in the original vehicle.



My pack is 36kwh, consisting of 12x 26.6v 3.0kwh modules (7s126p) made up of 18650 cells. They were manufactured by Tesla for use in the Mercedes B200E.



Each module has a wiring harness with one pin that connects to the negative terminal, one pin for each succeeding cell group (so climbing in voltage relative to the negative: 3.77v, 7.54v, 11.31v etc up to full module voltage of 26.4v on the last pin). There are also 4 pins for the 2 included thermal probes.


I have ordered mating wiring harnesses and pins, and my plan is to connect each cell group (126 cells) in parallel through a 2A fuse.

The wiring between modules will be about 4 feet of 20 awg hookup wire, which can easily handle 2 amps of current. Once connected and allowed to balance, the 2 cell groups should behave like a single group now with 252 cells.



Clearly, that small gauge wire is fine to monitor the voltages, and even apply some balancing loads. What I am wondering is what happens when I demand the max current from my batteries? With an 88kw motor and a nearly flat pack at 135v, I would be looking at a current draw of about 650 amps. This is within the specs of the module, which I believe are rated at 400A each or 800A total with them paralleled. So if everything was perfectly balanced, I would not expect any current to try to pass over the fused link. But, if one of the 126 cell groups was weaker than its partner, at high current demand it would sag to a lower voltage. That voltage difference would drive a current, right? How much would one expect a cell to sag at peak load?



4 feet for 20AWG is about 40 mOhms, plus there will probably be some resistance in the plugs and through the fuse, so say 50 mili ohms total. For 2 amps to flow;



V=IR; V = 2 * .050; V = 0.1v


So if one cell group dropped 0.1v lower than its neighbor, 2amps would be driven accross that wire, right?


In and of itself this would not be a dissaster - the fuse would open, the wires would be safe, and the power would still flow through the main connections. However, with a blown fuse the BMS now only sees half the cells - and that already weak cell group that is in most need of monitoring is not being seen by the BMS. I had thought I could wire the BMS to both sides of the fuse, but that way the BMS just becomes a secondary fuse, and I am not sure I want to find out how many amps it is rated for



Option 1:

The 20 awg wire is rated for 11amps (as chassis wiring) so I could bump up to 10 amp fuses, and then this would only be an issue if one cell group dipped .5v below the other. I would then need to manually inspect the fuses periodically to make sure they had not blown.



Option 2:
Implement some sort of system to remotely monitor the fuses and alert me if one fails.



What do people think? I do not really like the thought of having a separate BMS and charger for each string. Then again the cost of an extra BMS and charger is cheaper than burning down my vehicle, and possibly whatever it is parked next to

 

[MattsAwesomeStuff]

Since the only thing that changes is whether you'd order another BMS or not...

I'd just try it and see.

I have a feeling it's fine. Or, just use heavier gauge wire.

0.1v difference on a single cell is significant. That can be like, 20% of the capacity of a pack.

0.5v different would be an immense difference.

On a hunch, 2 amp is probably fine, but I would up your fuses to larger ones, and, if it happens to burn them, no big deal on the cells very temporarily being out of balance. Unbalancing is something that happens gradually over months. If you'd be aware that fuse blew somehow, you're going to make it home.

 

[john61ct]

Spitballing

Datalogging the voltage vs amps drawn would show a weaker group's (much) higher sag level via a quick lookover after a trip.

A custom MCU circuit could trigger an alert live so you go into limp-home mode.

 

[kennybobby]

Please take a look at Wolftronix's threads about this subject. He has the knowledge, skills and abilities to build a parallel pack, and took the necessary precautions.

https://www.diyelectriccar.com/forums/showthread.php?t=181490

https://www.diyelectriccar.com/forums/showthread.php?t=186185

Please don't listen to anyone who says "Just try it and see";

Parallel is a dangerous configuration as you note in your title.

Your diagram has no contactor or means to separate the two parallel strings, and this is a fatal approach.

Range extender gone wrong: http://www.torquenews.com/3618/warning-not-to-try-tamper-nissan-leaf-battery-pack

Hot Laef:

 

[drother]

https://www.orionbms.com/manuals/pdf/parallel_strings.pdf

 

[MattsAwesomeStuff]

Don't listen to Matt, he has no clue or experience with this. "Just try it and see"; or try it and die.

For Christ's sake it's a battery with fused connections between cells, not a bomb. No one's going to die. At worst you might discover "Hmm, yep, I blew some of those fuses. More current was flowing than I anticipated. Good to know, I'll add a second independent BMS."

Don't listen to all the chicken little's who scream about how the sky is falling and how you need to hire an expert with everything.

...

You can absolutely, positively connect up two packs in parallel, provided that their voltage are matched when you first connect them.

After that, by virtue of being in parallel, they'll stay balanced thereafter.

You correctly identified that the only circumstances they might become unbalanced is in the most extreme possible discharge, and even then you seem to have lots of overhead for the fuses.

You're not going to suddenly have 0.5v difference across two cells in parallel, which is what you calculated would cause more than 10 amps to flow.

Your diagram has no contactor or means to separate the two parallel strings, and this is a fatal approach.

You know what else doesn't have a contactor or means to separate two parallel strings? Every battery with more than one cell in parallel. Teslas have DOZENS of cells in parallel with each other per module, no contactors, no one suddenly dies.

All you're doing is having thinner connectors in parallel than you'd have if they were connected with nickle strips right next to each other.

What we're not talking about:

- Blindly connecting cells with unknown voltages in parallel with each other.
- Blindly connecting whole parallel packs with no BMS or individual balancing.

Range extender gone wrong: http://www.torquenews.com/3618/warning-not-to-try-tamper-nissan-leaf-battery-pack

Bull... shit.

Your fearmongering is ridiculous.

I'll quote the article: "The video’s editor writes that the battery pack was penetrated to connect a range extender and proceeded to [...] allow water into the battery box"

Yeah that's totally the same relevant danger. Never connect battery packs in parallel, it's super dangerous. Let me give evidence by showing an example of someone who drilled holes into their battery box and let their battery box flood with water and so it shorted and later burned down. That's totally exactly the same thing as connecting a 10amp fuse between each balancing lead on 2 parallel packs, with a proper BMS balancing cells, and observing whether the fuses blow to find out if you need to spend money on a second BMS. YOU WILL ABSOLUTELY DIE AND IGNORE ANYONE WHO SUGGESTS OTHERWISE!

Never wear blue jeans. Here is an article about how many heroin overdosers were wearing blue jeans at the time - a significant portion!

How do you even write this BS with a straight face?

 

[OR-Carl]

Hey guys, thanks for the feedback.



Kennybobby, I looked over that thread by wolftronix, and I think this quote got to the heart of what you are driving at:

The main problem with paralleling large battery packs together, is when one cell fails shorted. You now have lots of available power from the other good packs, attempting to charge the (now lower voltage) pack with the short in it.

This can cause thermal runaway (fire).

This is the "eddy currents" mentioned in the pdf you posted, Drother, more on that in a bit. Now, wolftronix was talking about leaf modules, which are a little bit different than the tesla ones, in that they are using big pouch cells, and there are no fusible links anywhere in the modules, right? Arnt the modules just 2 series cells in an enclosure?



Correct me if I am wrong, but each 18650 cell in my tesla "cell-group" does have a fuse on it. While eddy currents could still flow, a single cell that completely shorts is going to blow its fuse, and simply turn a 126 cell-group into a 125 cell-group. These eddy currents would seem to be limited if the cell groups were connected through the BMS wires - because now any small irregularity is going to correct itself over those links. On the leaf, a dead short in a single cell seems like it would be a much bigger problem.





So I read through that PDF, that did sum up the problems pretty well.





Note: While most lithium batteries can be directly paralleled together, check with the cell manufacturer to ensure that the cells can be safely paralleled and to see if there are any specific requirements for the specific cells used. In some cases (such as with some 18650 style cells), cell manufacturers may require individual fuses or fusible link wire to prevent over current through a single cell in the event of a cell failure or an internal short within a cell. Consult with the cell manufacturer to determine if such a design is necessary.




When two or more strings are paralleled together, currents will flow between the strings. These currents form due to differences in the total pack voltage between strings. The amount of current that flows is determinedby the difference in total string voltages, resistance of each string, and the characteristics of the cells.



Because of the possibility of these currents, it is absolutely essential that each string MUST have a contactor, shunt trip breaker, or other automatic and redundant means of isolating the string from any other strings if a critical fault occurs. This is in addition to the standard over-current protection for the string. The designer must consider and ensure safe behavior in the event of a failure of any single component, including diodes, contactors, and BMS units.



I can definately see the utility of having a fuse or contactor to separate the strings to prevent eddy currents, particularly if you have large format cells making up the strings. But what I am also seeing is that each tesla module is really already a parallel-series string.



I would like to hear peoples thoughts on what a "critical fault" would be for a module of fused 18650s. (Apart from "the fuse fails to open" and somehow allows enough current through for the cell to catch on fire )


I definitely would plan on leaving a lot of headroom on my pack regardless of how I wired it, and that seems like it would at least reduce the danger at the expense of some kwhs. I did look at the documentation for the BMS I am thinking of using (dilithium from thunderstruck) and they do show that you can use it to monitor parallel strings. I do appreciate the warning, and the input on how to make it work safely. I suspect I am not the only one thinking of using parallel arrangements of tesla modules, so I hope this thread can serve to steer others in the right direction.



I want to reread the thread about the guy who burned his house down with Tesla modules, and give some thought to which of the proposed solutions in that PDF seem the most workable with my setup.



Would love to hear from anyone currently using Tesla packs - how are you monitoring your cells? Has anyone had problems with cells dying in the cell groups?



Matt, I didnt see your post as I was writing this, but, uh, maybe tone it down a bit. If I am not mistaken you have worked with small format cells before, right? Maybe speak from that experience, but maybe try and read other peoples comments more as constructive criticism than personal attacks. I am here to learn something, and much of what you wrote does not further that aim. If you want to post in my threads, please dont be so antagonistic.

 

[MattsAwesomeStuff]

maybe try and read other peoples comments more as constructive criticism than personal attacks

I disagree with people in a pleasant and constructive way all the time. There was nothing constructive about this. It was directly a personal attack, out of the blue.

"Don't listen to Matt, he has no clue or experience with this. "Just try it and see"; or try it and die."

Where's the non-personal constructive part?

He doesn't even defend his attack, just cites how someone did something completely unrelated on a flooded and rusted pack which then caught on fire.

I don't think he even read what you wrote originally, his comments seem to think you're just strapping two packs together with only the end points in parallel.

*shrugs*

I think you've put a lot of thought into this, particularly with your worst case scenario calculations. I'd go by what they're telling you, not unfounded paranoia.

 

[john61ct]

This could be a thread of great value to the community, if we can just focus on the content issues rather than on personalities.

Even those of us without a clue can contribute a lot through statements needing correction and clarifying questions.

No one is building life-threatening vehicles based of any one poster or thread.

One observation:

putting strings or packs in parallel has very little to do with paralleling **cells** at the bottom 1S group level.

Seems to me the thread topic is HowTo do the former safely.

 

[brian_]

You know what else doesn't have a contactor or means to separate two parallel strings? Every battery with more than one cell in parallel. Teslas have DOZENS of cells in parallel with each other per module, no contactors, no one suddenly dies.

While is true that common EV packs have two or three cells in parallel at each level, with no contactors or fuses to isolate them. On the other hand, Tesla's modules have a huge number of cells in a parallel, and realizing a likely problem they have a fine wire acting as a fuse for every individual cell connection. The Tesla design is not a good example for this argument.

 

[brian_]

... Now, wolftronix was talking about leaf modules, which are a little bit different than the tesla ones, in that they are using big pouch cells, and there are no fusible links anywhere in the modules, right? Arnt the modules just 2 series cells in an enclosure?

Close... a classic Leaf module (they have changed for the latest version) contains four pouch cells - they are paralleled in pairs, and those pairs are connected in series, for a 2S 2P configuration. There is an external terminal for the point between the pairs, so every cell group level is accessible to the BMS. Those BMS taps have a substantial termination, so paralleling these modules including at the BMS taps seems inherently more feasible than with modules having only a small-gauge wire for each tap point.

Of course the Wolftronix approach does not connect the modules in parallel at every module or every cell level, but others have done this.

Correct me if I am wrong, but each 18650 cell in my tesla "cell-group" does have a fuse on it.

That is correct.

 

[brian_]

One observation:

putting strings or packs in parallel has very little to do with paralleling **cells** at the bottom 1S group level.

Seems to me the thread topic is HowTo do the former safely.

The Wolftronix solution is putting strings in parallel (not at the cell level), but Carl is contemplating parallelling at the cell level via only the BMS wires. They are indeed different.

 

[brian_]

https://www.orionbms.com/manuals/pdf/parallel_strings.pdf

I've only skimmed this, but it looks like an excellent explanation of the issues and of the resulting design considerations. I'll note, though, that it considers only the two common cases:

1. parallel connections at the cell level which are substantial - able to handle full pack current and ensuring negligible differences between the voltages across parallel cells
2. strings which have no interconnection between other strings other than at the every ends; in other words, no cell-level connectivity at all.

Carl's proposed "thin" paralleling at the cell level, with substantial resistance and very limited current capacity in the parallel connections, is not considered at all in this document, as far I as I saw in my quick reading.

 

[OR-Carl]

Hey Brian, thanks as always, for sharing your knowledge on the topic. My purpose with this thread was indeed to explore any possible avenues to address the inherent risk associated with 2 strings that are in parallel only at the ends.

Carl's proposed "thin" paralleling at the cell level, with substantial resistance and very limited current capacity in the parallel connections, is not considered at all in this document, as far I as I saw in my quick reading.

In a perfect world, I would probably think about putting heavy gauge connections between each cell-group, effectively merging the two strings into one. With the space constraints within my box, I do not think that is going to happen. I am going to take a look at the possibility of putting in thicker wires - This might not be that difficult, as the modules in this pack are not encased in a plastic shell - they are just wrapped up in that flimsy electrical isolation plastic. I think carefully drilling (with a stop collar) and tapping a hole into the metal bus that holds the cells might be doable. Anyone know what it is made out of? Aluminum would be a piece of cake, stainless, maybe not so much.



I guess the question would then become, how much eddy current would cell level connections need to be able to handle? If I fuse those connections, exceeding my anticipated limit should not lead to disaster, I wouldnt think. I suppose it might cause a cascade, where one cell-group dropping out causes more stress on the adjacent cell-groups, and you wind up back with end-paralleled strings. As long as you had some way of knowing that happened before you went to charge it again, it would probably not be that big a deal.



One thought I had - what about a resistor in line with the fuse? Could be less than 1 ohm and like a watt of power dissipation. That way eddy currents are going to much prefer to go out the main terminals. If one side dipped lower or was driven higher, power would very slowly bleed over to its neighbor. Once they were equal though, the resistor would not be dissipating anything. With a single BMS, though, you would be monitoring one side of the resistor - so could the readings be a little screwy?



What do people think about that idea?

 

[kennybobby]

The inline resistor is not a good idea.

The idea of using sufficiently large wires (8 or 6 AWG) to link the two strings at each CELL level and then using 1 CMU channel for those 252 cells in the CELL actually sounds feasible.

In the tesla packs the CELLS are connected with ~1/8" aluminum plate, so you could likely drill and tap in order to bolt your links.

Definitely got to be careful what you touch, good luck to you.

 

[OR-Carl]

The idea of using sufficiently large wires (8 or 6 AWG) to link the two strings at each CELL level and then using 1 CMU channel for those 252 cells in the CELL actually sounds feasible.

I will have to check and see how much space there is between the modules - they stacked them together pretty tight.

In the tesla packs the CELLS are connected with ~1/8" aluminum plate, so you could likely drill and tap in order to bolt your links.

That would certainly be easy to work with.

Definitely got to be careful what you touch

At the module level I am only ever dealing with 26 volts, which is not too scary. On the other hand, I will say that I would have to be very careful about what the drill bit touches With some way of limiting the depth, like a shaft collar, or a sleeve over the bit, it shouldnt be too dangerous.


Thinking about this, though, bigger wires mean less resistance, so larger eddy currents would flow even at small voltage differences. Since each group of 126 cells is rated for 400A of output, is a larger wire really any better as long as it is below that max rating? I suppose it would depend a little bit on the resistance across the fuse.

 

[kennybobby]

i'm thinking no resistor and no fuse--you are making a link from one CELL plate to the corresponding CELL plate in order to create a 252-cell CELL for the BMS/CMU board. Any currents will need to flow thru the wire just as they do in the plates.

8AWG is 1/8" in diameter and has a 10 sec fusing current of 472 Amps, and 2.4 kA for 1 second.

But one risk and downside here is that the links are basically exposed and unprotected by a fuse in the event of a wreck that damages the pack or wiring. If a fuse is desired it would need to be rated for protection without nuisance blows, and one would be needed for each link (# of series CELLS). They will be big and bulky and take up space. If the modules are within an enclosure then the risk of no fuses might be mitigated.

 

[boekel]

Don't do it with 'one' bms: you cannot know if a fuse (or wire) has blown.


But...you're in luck: SimpBMS by Tom de Bree supports these modules with the standard BMS slaves on them...

https://www.diyelectriccar.com/forums/showthread.php/fs-tesla-vw-outlander-bms-master-198263.html

if you want safety: let every string have it's own contacter, when something is wront, contacters will open and strings are isolated.

 

[OR-Carl]

Hey Boekel, thanks for the input.

Don't do it with 'one' bms: you cannot know if a fuse (or wire) has blown.


if you want safety: let every string have it's own contacter, when something is wront, contacters will open and strings are isolated.

I agree that if a fuse has blown, the safety of the system can no longer be guaranteed. Due to the rather large number of fuses that would have to be monitored, and the desire to keep them inside the battery box, I am starting to think that having a BMS monitoring each string might be worth the added expense.



I am going to be using the dilithium BMS, and they show you can use that system to monitor the second string.

If I did ditch the cell interconnections, I would then put fuses and contactors on both strings.


As much as I would like to shave off $550, safety equipment is probably not the place to try and pinch pennies.

 

[boekel]

Why would you throw away a perfectly functioning distributed bms and use extra wiring (extra risk) and extra costs to replace it?