I have a number of salvaged 18650s and 21700s (Tesla Mod 3). I know the chemistries are a bit different and the cell capacities are different (I assume ATM 5ah for the 21700s and 2.6ah for the 18650s). I'm thinking of building a 48V 14s battery using strings of each. For instance 4x10 21700s in parallel for ~200ah and 4x19 18650s in parallel for ~198ah or 4x20 for ~ 208ah. It seems I should be able to use these packs in series as they have the same voltage and close to the same capacity. I think a 14s BMS should not have a problem managing such a setup. Is there a reason to not do this? By extension, it seems I should be able to increase the bank's capacity by adding 4x5 21700 packs in parallel to the 21700 strings and 4x10 18650 packs in parallel with the 18650 strings. Of course each string would get a ~100ah increase at the same time. Thoughts?
Is there any reason I cannot mix strings of 18650 Lions cells with 21700 cells?
- Thread starter rgbtxus
- Start date
Having things in series with the potential for different capacity/resistance is asking for trouble.
If you built a XP14S 18650 pack and paralleled it with a XP14S 21700 pack, I don't see a big problem, and they could be different capacities.
100% of cells should be tested for both IR and capacity and all parallel cell groups within a string should have the same capacity, and all cells should be fused.
If after the above testing, you meticulously matched everything up such that all parallel groups have nearly identical capacity, then I'd consider mixing them, but if it's just to avoid a second BMS, I don't know that I'd make that choice.
If you built a XP14S 18650 pack and paralleled it with a XP14S 21700 pack, I don't see a big problem, and they could be different capacities.
100% of cells should be tested for both IR and capacity and all parallel cell groups within a string should have the same capacity, and all cells should be fused.
If after the above testing, you meticulously matched everything up such that all parallel groups have nearly identical capacity, then I'd consider mixing them, but if it's just to avoid a second BMS, I don't know that I'd make that choice.
Thanks for the input. I have capacity tested all cells and measured their resistance. So far I have just made 2 24V 7s10p packs with 18650s for portable use either in series for 48V or parallel for 24V. All cells are individually fused and each pack has its own BMS. I am now thinking about a power wall. The ones I have seen have used have n packs in series and a single BMS, e.g., 14s40p. Generally the maker has taken the trouble to balance the capacity of each pack. It seems that matching impedance is often not done. Since I already have a sizable investment in 18650s and 2170s I am looking for confirmation (or refutation) of my scheme to mix them provided each "S" group is of similar capacity, e.g., 5s of 18650s and 9s of 21700 managed by one BMS where each "S" is engineered to have close to the same ah capacity. I reason that this should be no different than having the BMS manage cell groups that are the same. But, a possible fly in the ointment is that the resistance of the 21700s cells is significantly lower than that of the 18650s, like half. In a world with a BMS that could handle significant balance currents this might not be an issue, but I suspect the 21700 groups might be driven to cutoff long before the 18650 groups resulting in poor utilization of the power in the 18650 groups and undue stress on the 21700 groups. I would appreciate anyone who could enlighten me on this issue. Thanks.
Going to assume that when you say 7S10P you mean 10P7S. You have 10 parallel cells with 7 of those groups in series.
IR is a big factor. In a parallel group, they will take the brunt of the current. The higher IR cells will experience less of the current and may have added heating.
Since the parallel group is in... parallel, there's nothing to manage with a BMS. Parallel means ALL of the cells are at the same or nearly the same voltage even under load unless there is substantial resistance variation in the interconnects. The BMS isn't even seeing the current, the sensing leads are seeing OCV since the sensing leads don't pass the current.
The imbalanced flow in the group with the larger lower IR cells could conceivably trip fuses if using wires.
Each of your parallel groups should have the same capacity and the same IR. One the next level, the individual components should have the same capacity and same IR. That's where you're slathering the fly with that ointment.
IR is a big factor. In a parallel group, they will take the brunt of the current. The higher IR cells will experience less of the current and may have added heating.
Since the parallel group is in... parallel, there's nothing to manage with a BMS. Parallel means ALL of the cells are at the same or nearly the same voltage even under load unless there is substantial resistance variation in the interconnects. The BMS isn't even seeing the current, the sensing leads are seeing OCV since the sensing leads don't pass the current.
The imbalanced flow in the group with the larger lower IR cells could conceivably trip fuses if using wires.
Each of your parallel groups should have the same capacity and the same IR. One the next level, the individual components should have the same capacity and same IR. That's where you're slathering the fly with that ointment.
Sorry, new to this battery construction game and I may not be explaining clearly. Yes, I meant 7 groups in parallel in my example, i.e., a "24V" system. Each group will be either 18650s or 21700s. Each group will have close to the same ah capacity, like 200ah +/- 8. Each group will have cells with fairly matched resistance. The 21700 groups look to be about 12+/-2 mOhm. The 18650 groups would have a larger and wider spread, like 49 +/-4 mOhm. If I were to build such a system how would it behave under load? Am I correct in assuming the 12mOhm groups would take a much larger portion of the load? If so is there a way to quantify it? Finally if this scheme is deemed unworkable or unwise is there a sensible way to combine these two cell types in one system. For instance if I make say 7 groups of 21700s and 7 groups of 18650 each sized for approximately the same capacity and then combine them pairwise into larger groups of 1 21700 group in parallel with 1 18650 group and then put these pairs in series would that be workable? In this case each of the larger groups would have approximately the same capacity and the same resistance. Within each larger group there would be 2 subgroups of approximately the same capacity but very different resistance.
Remember that resistances in parallel decrease. 12 21700 in parallel would only have a 1mΩ resistance for the parallel group.
I much prefer the idea that you are at least keeping cell groups the same, i.e., all 18650 or all 21700.
Essentially, if you can match capacity and IR of each cell group, then you will minimize problems even if you are using different cell types; however, this may be challenging.
For example:
12X 21700 would have 60Ah capacity and 1mΩ total resistance
24X 18650 would have 60Ah capacity and 2mΩ total resistance
1mΩ isn't a big different unless you consider that it's 2X the resistance of the other cell group. That means the same current passing through both cells will have different effects. 18650 will sag more under load, and raise more under charge. The net effects will be less usable capacity, less charge retained from a given current, and more heat.
A balancing BMS will see the 18650 rise during charge and actually bleed off capacity from the higher voltage cell even though that cell is receiving less charge, so when charging terminates, the 18650 will settle to a lower voltage than the 21700.
ALL of these effects and their severity depend on how you're using them. If you're only subjecting every cell to an average of 0.1 or 0.2C, it's probably not a concern, but if you're pushing them to 1C or higher, then those effects will get very noticeable.
I much prefer the idea that you are at least keeping cell groups the same, i.e., all 18650 or all 21700.
Essentially, if you can match capacity and IR of each cell group, then you will minimize problems even if you are using different cell types; however, this may be challenging.
For example:
12X 21700 would have 60Ah capacity and 1mΩ total resistance
24X 18650 would have 60Ah capacity and 2mΩ total resistance
1mΩ isn't a big different unless you consider that it's 2X the resistance of the other cell group. That means the same current passing through both cells will have different effects. 18650 will sag more under load, and raise more under charge. The net effects will be less usable capacity, less charge retained from a given current, and more heat.
A balancing BMS will see the 18650 rise during charge and actually bleed off capacity from the higher voltage cell even though that cell is receiving less charge, so when charging terminates, the 18650 will settle to a lower voltage than the 21700.
ALL of these effects and their severity depend on how you're using them. If you're only subjecting every cell to an average of 0.1 or 0.2C, it's probably not a concern, but if you're pushing them to 1C or higher, then those effects will get very noticeable.
Thank you very much! This certainly helps bring these issues into focus for me. I think worst case I would be at .35C as currently envisioned with 80 18650s in a parallel pack. If I double the cells in a parallel group then <0.2C would be comfortable.