Li-ion cells cycle ageing

[docware]

I think that most significant issue with this particular test is a ZKETECH EBC-X0510 accuracy (stability).
Second issue is temperature stability.

Percentage values used only to help differ the lines at 100% SOC.
Yes, I know that temperature variations and ZKETECH are two obstacles to accurate measurement. I declare 25 ± 1°C but trying to have better range, something like 25,3 ± 0,5°C. Of course, your 0,1°C range is ideal.
Because charging is CC/CV, I believe that final voltage 4,2 V and cut-off current 0,1 A accuracy/repeatability is relevant more than 1 A charging current real value. More significant is temperature because it affects ability of Li-ion cells to absorb delivered charge. And of course temperature is influencing measuring during discharging. Temperature variation is probably mostly issue for high capacity cells (M36, 35E, GA) while high power cells (HG, VTC6, 30Q) probably suffer typical high capacity decay at the first cycles.

One would assume more variety of results if ZKETECH effect is really significant. Or the deviation is stable. ZKETECH and temperature bias will be less significant in each another measurement.

As we can expect gradual slowing down of the capacity decay, it is not unrealistic to be somewhere between 95 – 98 % after 8 months (240 days).

Pajda, how is average monthly capacity fade 2 – 4 mAh consistent with M36 95,4 % capacity after 8 months (i.e. 156 mAh fade) ?

 

[Pajda]

I think if we have more data we can evaluate the tester accuracy more precisely.

Pajda, how is average monthly capacity fade 2 – 4 mAh consistent with M36 95,4 % capacity after 8 months (i.e. 156 mAh fade) ?

Yo caught me with my pants down, thanks The answer is simple "generalization" error which i fight against. There are cells with significantly faster degradation (capacity fade) than others, also there is a huge influence of cycle life (inserted capacity and DCIR tests) for some cells which has yet to be removed from the data. This leads to M36 (100% SoC) 29 mAh loss in 1th month and only 8 mAh loss in the 8th month.

For comparison:
M36 losses 156 mAh(100% SoC) and 143 mAh (50% SoC) after 8 months
29E7 losses 69 mAh (100% SoC) and 30 mAh (50% SoC) after 8 months.
* This data include 8 nominal capacity and 10s and IEC DCIR tests.

 

[docware]

.... also there is a huge influence of cycle life (inserted capacity and DCIR tests) for some cells which has yet to be removed from the data.

You probably want to clean aging data with help of data collected from cycling ?

 

[Pajda]

Exactly. The closest data set I already have for M36 is from cycling 0.5C-0.5C at 100% DoD where in the first 8 cycles is ca 51 mAh capacity loss.

 

[Pajda]

Recently I have started another calendar life test to check how the cold temerature storage affect the modern cell life. I just put 8 cell samples into the freezer in refrigerator. The storage temperature oscilates between -15 °C to -25 °C so I calculate with -18 °C average. Test cycles will be again at monthly basis in TC set to 25 °C.

.. added another 50 cycles of 30Q.

 

[docware]

What I found out from 30Q and VTC6 cycling data is that the quick capacity loss at the beginning happened mainly during the first 8 cycles. DOD for both are slightly over 70%.




However these are high power cells. For high capacity cells like M36 or 35E I see different picture, where initial capacity drop isn´t so fast. Typical are positive and negative spikes, probably thanks to the temperature variations. DOD slightly above 60 % for both cells.




-25 °C isn´t it much ? Specifications are usually down to -20 °C.

 

[Pajda]

-25 °C isn´t it much ? Specifications are usually down to -20 °C.

Yes it it but it is air temperature which oscilates due to the cheap kitchen freezer regulator hysteresis. I suppose that the temperature of the cells in the box will be relatively stable and definitelly higher than -20°C.

It is common problem with battery testing, where you are always short of some equipment. I cannot occupy precission TC for such a long time. The same problem is with hot temp. storage test.

 

[Pajda]

For those who are awaiting another data for "5C cycle life torture test", I had to recently break the test for at least 14 days due to the usage of room where the test is performed.

Also small teaser from cold storage calendar life test

- Samsung 29E7 after one month storage at 100 % SoC in -18 °C average temperature after heating to 25 °C can discharge 99.3% of its nominal capacity (including calendar ageing). The calendar ageing effect itself is 99.7 % (with included one full nominal capacity test and two DCIR tests).

 

[docware]

New update, Samsung 29E7 and LG M50T added.

 

[docware]

30Q selfdischarge has been mentioned several times in this threat as well as in others.

After cycle testing, it was obvious that remaining aprox 90 % capacity of 30Q cell could be used in some not demanding application. So flashlight was my choice. Charging the cell to 3,8 - 3,9 V only. However, I still had notion that I have to charge this flashlight too often. Decided to check selfdischarge, charged cell at 1A to 4,2 V and let the cell sitting in the fixture with logging voltmeter connected. From the picture bellow is obvious that selfdischarge is extremely high. After 32 hours the cell was at 50% SOC ! During charging it was even not possible to achieve 100 mA cut-off, I had to end charging at 120 mA. From all information we have now is obvious, that probability of Samsung 30Q selfdischarging after some utilization is very high. I´m afraid this cell is no way.
Currently SONY VTC6 is resting in the measuring fixture for comparison.

 

[docware]

Here is selfdischarge comparison of 30Q and VTC6.



I made some selfdischarge voltage drop measurements on the cells in calendar aging. Obviously at 100% SOC new 30Q has markedly higher selfdischarge than others.

 

[docware]

Selfdischarge update.
Another two cells from cycle aging test checked (Samsung 29E7, Sanyo GA), both seems to be OK.
Then new cells Samsung 30Q No4 and Samsung 40T No3 checked, both have only a few full charge/discharge cycles. It is obvious that even new 30Q No4 (orange chart curve) has slightly worse selfdischarge from the very beginning. 40T seems to be OK.

 

[brone]

30Q selfdischarge has been mentioned several times in this threat as well as in others.

After cycle testing, it was obvious that remaining aprox 90 % capacity of 30Q cell could be used in some not demanding application. So flashlight was my choice. Charging the cell to 3,8 - 3,9 V only. However, I still had notion that I have to charge this flashlight too often. Decided to check selfdischarge, charged cell at 1A to 4,2 V and let the cell sitting in the fixture with logging voltmeter connected. From the picture bellow is obvious that selfdischarge is extremely high. After 32 hours the cell was at 50% SOC ! During charging it was even not possible to achieve 100 mA cut-off, I had to end charging at 120 mA. From all information we have now is obvious, that probability of Samsung 30Q selfdischarging after some utilization is very high. I´m afraid this cell is no way.
Currently SONY VTC6 is resting in the measuring fixture for comparison.

Samsung 30Q No3 (after cycling test) selfdischarge.jpg

Wow, that's pretty awful. Got to check the SOC on my 30Q pack which I rarely use because the capacity dropped so fast on it

 

[Pajda]

This week I finally managed to continue the 5C torture test. So there is ca one month pause after 150 cycles of VTC6 and 200 cycles of 30Q. No significant recovery visible.

 

[docware]

Calendar aging test update. Seven cells passed 200 days milestone.

 

[Honk]

Samsung 29E6 looks really strong in these tests.
It makes me smile as I have a few hundreds of E6 in my freezer for future use.

 

[Pajda]

By the end of May 2021, the last competitor Sony VTC6 will reach 1000 cycles at 5C 100% DoD continuous test. This will free one of EBC-A20 testers for another test if you are still interested. I would like to stay with 5C discharge, as I consider this test practicaly unusable for any application it just look badass. 8)

You can choose any cell that should at least theoretically survive 5C continuous (up to 20A as it is limited by the A20 tester) I'll try to get him but it will take some time, I prefer Nkon or other EU shop with some history. Right now I have only Molicel P42A in the shelf from this category.

You may also have noticed that the new Samsung 30Q6 model has been added to the comparison. The assumption is that at ca 500th cycle it will fall to the HG2 level and somewhere between 550-650 cycles it will suddenly die. We'll see.

 

[ridethelightning]

strange about the 30q. i hears somewhere that it has fast capacity loss above a certain number(400) of cycles, but i had no idea of the self discharge phenomena.
I have several large 30q packs in storage and they seem to stay charged just fine( stored for a year now)

 

[vanturion]

Right now I have only Molicel P42A in the shelf from this category.

Perfect! If no one else feels strongly, I'd like to see that one tested at 15A. I ended up sending back the P28As I was going to use for my new packs in favor of waiting on P42As to come back in stock. Even seeing the trend established in the first 50 cycles would be super useful if the timing allows.

 

[Fataga]

Right now I have only Molicel P42A in the shelf from this category.

This one is super interesting for me also. Do it please!

 

[Pajda]

So P42A then... The sample is from 2K215 01 series, which mean production date 2/2020. If everything goes well, the test can be started this week.

The only issue which must be decided is if the test curent will be 15A or 20A. From my perspective it should be 20A (5C) to be fair to other competitors in the chart as P42A is a 4000mAh nominal capacity cell. And other reason is that I already finished 3C test which is close to 15A.

In any case, expect decent results, a significant decrease in capacity (approx. 15%) in the first 100 cycles and then a linear decrease to ca 70% of the nominal capacity at 1000th cycle. Molicel P42A significantly outperforms Samsung 40T1 version, but I expect an interesting duel between P42A and the new revision 40T3 from Samsung.

 

[vanturion]

Right on, thanks - happy with the data however it comes. Only suggested 15A as it would be easier to directly compare to the results from your 18650 3000mAh chart. I would've run my 25R test at 15A, but the cheap battery fixture I got is only good for 10A.

And other reason is that I already finished 3C test which is close to 15A.

Are you saying you already ran P42A @ 15A or were you just talking about your Page 20 chart?

In any case, expect decent results, a significant decrease in capacity (approx. 15%) in the first 100 cycles and then a linear decrease to ca 70% of the nominal capacity at 1000th cycle. Molicel P42A significantly outperforms Samsung 40T1 version, but I expect an interesting duel between P42A and the new revision 40T3 from Samsung.

From the P42A manufacturing spec sheet, it looks like they're claiming ~85% capacity after 500 cycles of 20A discharge, which looks like it's relatively on par with the impressive HG2 capacity results after 500 cycles. Eager to see whether this result can be independently verified, but if you've already cycled it at lower amperage and are only expecting 85% after 100 cycles, that is somewhat disappointing.

 

[Pajda]

For that P42A data comparison, the problem is what capacity value they chose as the default value of 100%. I related those percentages to their typical value of 4200 mAh. If their 100% should be minimal value of 4000 mAh, then my measurement almost matches theirs. This is common problem and so the best approach is to get own values and do not much look in the datasheet.

I think the absolute value of 15A would be very misleading fo comparison, as it is only 3.75C load for 4000mAh cell vs 5C for 3000mAh cells. The same isuue for 25R as it should be tested at 12.5A discharge (5C) for direct comparison in this chart.

 

[john61ct]

IMO the constant for current discharge should be a selected C rate, not in amps.

 

[vanturion]

For that P42A data comparison, the problem is what capacity value they chose as the default value of 100%. I related those percentages to their typical value of 4200 mAh. If their 100% should be minimal value of 4000 mAh, then my measurement almost matches theirs. This is common problem and so the best approach is to get own values and do not much look in the datasheet.

Ah, thanks for the clarification. Maybe not so great of a business practice from the buyer's perspective to overestimate their nominal capacity then.

I think the absolute value of 15A would be very misleading for comparison, as it is only 3.75C load for 4000mAh cell vs 5C for 3000mAh cells. The same issue for 25R as it should be tested at 12.5A discharge (5C) for direct comparison in this chart.

I guess I was thinking about all of this in terms of use on an ebike in application rather than agnostic testing consistency between various batteries with different capacities. For example, if I'm looking to compare performance at a specific current/cell based on design limits of wires, fuses, cooling capability in motor/controller, and/or whatever other parameters, then it makes some kind of sense to look at a range of cells tested at this particular amperage. Even with this logic though, ebikes in real world conditions are, of course, no where near operating at 100% all the time anyway so it's not like matching amps between different capacity cells in this kind of testing is resembling real world conditions. I see your point now for fairness of testing between cells of different capacities. In any case, it's easy to extrapolate or guess performance at current values above or below what was tested too when comparing test results to manufacturing spec results.