Mixing 150 Panasonic NCR21700 with 50 Samsung INR21700

[braincore]

I'm constructing a 20S-10P battery pack for an e-scooter. I have:

- 150 cells of Panasonic NCR21700 5000mAh
- 50 cells of Samsung INR21700-50E 5000mAh

Both sets of cells are supposedly those used in Tesla Model 3s and Ys though I assume each of their battery packs use homogeneous cells.

Should I make each parallel cell group heterogeneous (mix some Samsung cells in), keep each cell group homogeneous (make a couple of the parallel groups entirely Samsung), or avoid this idea all together and try to buy more Panasonic/NCR cells?

 

[braincore]

What I can gather from the datasheets is that the characteristics (voltage curve, cycle life, ...) between the batteries are different enough that it doesn't make sense to mix them in any way.

Would appreciate any other opinions, but it looks like I should stick with only using the Panasonic cells and build out a 20S-7P pack rather than a 20S-10P.

I picked up these 150 Panasonic cells locally and they appear to be the same as the ones advertised here (7 cells in 24V Battery pack Model 3 Y 2170 lithium ion 5000mAh NCR2170A XT60 | eBay). I've seen reference to these cells in other posts (via search). Does anyone have any first-hand experience to report?

 

[docw009]

I know the 50E's are 9.8A cells, and I believe the 21700A's are 15A, If true, I agree with you about not mixing them.

As for the ebay packs. who knows if they are really new, or when they were made.

 

[Chalo]

What I can gather from the datasheets is that the characteristics (voltage curve, cycle life, ...) between the batteries are different enough that it doesn't make sense to mix them in any way.

While I haven't seen their data, voltage curves aren't relevant to what you're asking.

Imagine that you have two vessels for liquid, say a bucket and a long necked bottle the same height as the bucket. As you dispense them separately, the percent of volume remaining versus percent of water level remaining will be quite different between the two vessels. But if you join them by a tube at the bottom, they will not only both maintain the same level as they're dispensed, but they'll both be full and empty at the same time. It's the same when you put different cells in parallel, as long as their full and empty voltages are the same.

If the cells you want to use have radically different cycle lives, you can make parallel groups of different cells separable with plugs, so when the shorter-lived cells wear out, you can detach them from the longer-lived cells. Because parallel connections only carry small balancing currents, those connections don't need to have low resistance.

It's never as ideal to mix cells in a pack as it is to use identical cells, but it's not that big a deal. It's not nearly as serious a disadvantage as you're already accepting by using cells that are much too small for the job you are trying to do, which will cause you to make large parallel groups instead of having a 1P pack of appropriately sized cells.

 

[amberwolf]

If you make separate groups of different cells and place those in series, they're going to perform differnetly from each other, and the BMS (if any) will react to this if any of them go out of limits because of it. The battery will also not remain in balance like it should.

If you instead parallel each type of cell with the other in proportion for each group, so that there are the same number of a particular cell in each group, then each group should be the same as every other (assuming all the cells are new and identical, perfectly matched to each other within that model of cell).

I don't think you have enough cells to build a 20s pack this way, as whatever you use has to be divisible by 20. So you'd need either 40 or 60 samsungs, and either 160 or 140 panasonics, to build a 20s10p pack. (or other numbers divisible by 20)

Regarding performance of any specific cell, I recommend Pajda's and others cell testing threads here, whcih cover various brands and models, and the lgyte-info.dk testing site which has info on quite a few.

 

[braincore]

Imagine that you have two vessels for liquid, say a bucket and a long necked bottle the same height as the bucket. As you dispense them separately, the percent of volume remaining versus percent of water level remaining will be quite different between the two vessels. But if you join them by a tube at the bottom, they will not only both maintain the same level as they're dispensed, but they'll both be full and empty at the same time. It's the same when you put different cells in parallel, as long as their full and empty voltages are the same.

Got it. I understand they'll drop voltage together as they're utilized. I was more concerned with having to use conservative bounds on the max/min voltage (due to differences between the two) for maximizing cycle life which further cuts into range. Likely an over optimization.

If the cells you want to use have radically different cycle lives, you can make parallel groups of different cells separable with plugs, so when the shorter-lived cells wear out, you can detach them from the longer-lived cells. Because parallel connections only carry small balancing currents, those connections don't need to have low resistance.

Thanks for the idea.

It's never as ideal to mix cells in a pack as it is to use identical cells, but it's not that big a deal. It's not nearly as serious a disadvantage as you're already accepting by using cells that are much too small for the job you are trying to do, which will cause you to make large parallel groups instead of having a 1P pack of appropriately sized cells.

Out of curiosity, what cell would you recommend for the job?

 

[Chalo]

I don't think you have enough cells to build a 20s pack this way, as whatever you use has to be divisible by 20.

He can make a 20S 9P pack with 7 of one and 2 of the other in each parallel group. The remainder can go into vape pens and flashlights, which is what they're actually good for anyway.

 

[braincore]

If you make separate groups of different cells and place those in series, they're going to perform differnetly from each other, and the BMS (if any) will react to this if any of them go out of limits because of it. The battery will also not remain in balance like it should.

Agreed. Between Chalo's advice and yours, it's clear that the only viable option is really to keep each cell group in series identical even if they're composed with a mix of cells.

I don't think you have enough cells to build a 20s pack this way, as whatever you use has to be divisible by 20. So you'd need either 40 or 60 samsungs, and either 160 or 140 panasonics, to build a 20s10p pack. (or other numbers divisible by 20)

Should have clarified that I'm able to acquire more Samsung cells.

Regarding performance of any specific cell, I recommend Pajda's and others cell testing threads here, whcih cover various brands and models, and the lgyte-info.dk testing site which has info on quite a few.

I'll take a look. Thanks!

 

[Chalo]

Out of curiosity, what cell would you recommend for the job?

Well if you want 50Ah, I would recommend a 50Ah cell. Panasonic makes a 51Ah cell in the PHEV2 format that would be a great choice.