Cycle life tests of High Power density cylindrical cells

[Pajda]

"Standard Power Rate" test (1C continuous discharge)

note: the 1C discharge test is performed at ca 90% DoD (4.15V-3.00V)

Test eqiupment:
- 2pcs ZKETECH EBC-X0510
- BF-2A four-wire cell fixture (10A version with Nickel plated contacts)
- tested in room with relatively stable temperature 23 ±2 °C

Remarks and comments:
- someone considers these testers a child's toy
- the most problematic is tester accuracy. It is 0.2 % ±0.01 A. The stability is relatively good.
- do not much look at the absolute values, particularly at low currents in nominal capacity tests, trends are here important results.
- will try to add more data as it will come
- each 50th cycle is inserted datasheet nominal capacity test with 10s DCIR test (see dots in figure)
- minimal charging CV phase cut-off current is set to 100 mA due to the ZKETECH testers limitation
____
"Table" with values measured on precision battery tester (not ZKETECH). Those test were performed in temperature chamber under 25°C, TC is using forced air flow to maintan consistent temp. This 25°C air flow acts as cooling force on the cell surface, so the measured temperatures under high load will be affected by this.

added Samsung 30Q6 to the chart (X0510 tester), sample source: ---
added Sony 18650VTC6 to the chart (X0510 tester), sample source: NKON
added Samsung INR21700-40T3 to the chart (A20 tester), , sample source: NKON
added Samsung INR21700-50S(1) to the chart (A20 tester), sample source: NKON
added Molicel INR21700-P42A to the chart (X0510 tester), sample source: NKON
added Molicel INR21700-P45B to the chart (X0510 tester), sample source: NKON
added Molicel INR18650-HJ2 to the chart (X0510 tester), sample source: Vapcell
added Tenpower INR21700-40TG to the chart (X0510 tester), sample source: NKON
added Samsung INR21700-40T4 to the chart (A20 tester), , sample source: Vapcell

 

[Pajda]

"High Power Rate" test (5C continuous discharge)
migrated from older thread Li-ion cells cycle ageing

note: the 5C discharge test is performed at ca 100% DoD (4.20V-2.50V)

Test eqiupment:
- 2pcs ZKETECH EBC-A40L
- 3pcs ZKETECH EBC-A20
- 3pcs BF-2A four-wire cell fixture (30A version with "huge" golt plated contacts)
- tested in room with relatively stable temperature 23 ±2 °C

Remarks and comments:
- someone considers these testers a child's toy
- the most problematic is tester accuracy. It is 0.2 % ±0.01 A. The stability is relatively good.
- do not much look at the absolute values, particularly at low currents in nominal capacity tests, trends are here important results.
- will try to add more data as it will come
- each 50th cycle is inserted datasheet nominal capacity test with 10s DCIR test (see dots in figure)
- 100 % DoD should be with asterix because:
a) HG2 and VTC6 datasheets recommends to go down to 2V when cycled at 5C
b) minimal charging CV phase cut-off current is set to 100 mA due to the EBC-A40L limitation (HG2 should go to 50 mA cut-off)

* There was ca 4 weeks pause in testing after 200 cycles 30Q and after 150 cycles VTC6
** Changed VTC6 tester from A40L to A20 model after 720 cycles
*** Missed data for 50S, 40T3 and P42A caused by the harddrive failure
____
added "Table" with values measured on precision battery tester (not ZKETECH). Those test were performed in temperature chamber under 25°C, TC is using forced air flow to maintan consistent temp. This 25°C air flow acts as cooling force on the cell surface, so the measured temperatures under high load will be affected by this.

added Samsung 30Q6 to the chart (A20 tester), sample source: ---
added Molicel P42A to the chart (A20 tester), sample source: NKON
added Samsung 50S to the chart (A40L tester), sample source: VAPCELL
added Samsung 40T3 to the chart (A20 tester), sample source: NKON
added Sony US21700VTC6A to the chart (A20 tester), sample source: NKON
added Samsung 25R8 to the chart (A20 tester), sample source: NKON
added BAK N18650CNP-25 to the chart (A20 tester), sample source: NKON
added Sony US18650VTC5A to the chart (A20 tester), sample source: NKON
added Molicel INR21700-P45B to the chart (X0510 tester), sample source: NKON
added LG INR18650-HJ2 to the chart (X0510 tester), sample source: Vapcell
added LG INR21700-H40B to the chart (X0510 tester), sample source: Vapcell
added Tenpower INR21700-40TG to the chart (X0510 tester), sample source: NKON
added Molicel INR18650-P28B to the chart (X0510 tester), sample source: Vapcell
added Molicel INR18650-P30B to the chart (X0510 tester), sample source: NKON

 

[Pajda]

"High Power Rate" test (5C continuous discharge)
note: the 5C discharge test is performed at ca 100% DoD (4.20V-2.50V)

18650 size
2500+ mAh samples comparison



3000+ mAh samples comparison



21700 size
4000+ mAh samples comparison



4500+ mAh samples comparison

 

[vanturion]

^Good stuff

since this post got a little long, took it to a separate topic so I'm not polluting your thread

 

[DogDipstick]

Can you measure AC IR? You use a four terminal bridge for the fixture,no?Does the ZKE not do 1000 hz AC?

How do you measure this " 10s DCIR"?

Interesting to see the DCIR climb significantly with the cycles. Can you plot temp vs young cycles vs old cycles? I can cycles my big pouch cells, 1x, 10x, 100x or infinity.. but... they would take a very long time to accrue 200-400 cycles. Wow. I would love to see a pic of the device testing system.. I always thought the ZKE machine does a " AC IR" test, not a DCIR"....
@ 2.5 hours per cycle, I can do 9.6 cycles per day, ( running full time) and that means about 21 days of constant cycling. For ~200 cycle tests on the spreadsheet.

I can do 32s, 40A , four channels, 6000w~... for my AC 1000hz machine, but, only 40A, that is usually like 1C, max, or 0.3C..max.. 5C is hard for me to do and I can only do a DCIR test on any power cell I have ( 15, 25, 40, 60Ah) cell, if I want to discharge at greater than 1-2C max. But you know what Ihave, .

Very interesting, the controlled repeatability.

 

[vanturion]

The EB Tester software can only test for some manner of DCIR (can input current over what appears to be 1s), but it's not precise at all. The manual even states this and that it's basically only good for "distinguishing a new battery from an old one". When I first started messing with the tester, I tried to manually calculate DCIR 10s, but it was no good for this. Not enough precision, which is why Pajda uses a separate tester to measure these values. Basically the EBC-A20 is not good for instantaneous measurements, but it is just fine for characterizing energy and capacity trends, for cheap too (or at least it used to be).

I'm not sure if you're asking Pajda or me, but I didn't have a way to characterize temperature without taking a manual measurement. Looking at Pajda's tables, compared to 20A, 10A is barely getting warm at the end of a shortened cycle anyway so it wasn't worth measuring from my POV.

I can see how characterizing DCIR performance might be useful for roughly estimating the lifespan of used cells quickly without running a capacity test, is that why you're interested in these values?

 

[Pajda]

I finally found a sample of the cycled VTC6 and added this to the graphs and table. This VTC6 sample lay in a box at room temperature for about 2 years after the end of the 1000 cycle of Standard power rate test and after then the final test for the graphs and table was performed.

 

[SniperGaulois]

Awesome work!!! What an impressive battery of tests. That's one of the best technical research posts I've seen on ES and compared to other 18650 graphs available... I'd summarize the result as: The LG has quite a lot of resistance and high temperatures, but it's still safe after 1000 cycles, so it gives you 20-40% more bang for the buck than the others? Is that a fair statement? NASA JPL has a well equipped lithium test lab, and they published a comparison of the 3 best 18650's in 2018, and LG also came out as their top choice... So, I'm running on MJ1 at the moment... It's a clever move if you want a used EV, look for all of them that use LG's

 

[Pajda]

Small update after a while

BAK N18650CNP-25 and SAMSUNG INR18650-25R8
- both cells finished 1000 cycles at 5C @ 100% DoD. From this synthetic test BAK perform a little better, it gives more energy and in particular the increase of internal DCIR is significantly slower. On the other hand, BAK capacity/energy starts to decrease significantly after 950 cycles and so I assume, that it will be completely dead after ca 1100 cycles. Unfortunately I will not have access to the temperature chamber for next few weeks, so the final tests to complete the graphs and tables will have to wait.

In any case, this leads to a situation where two 20A channels are now free and I can't think of another explicitly interesting competitor in the 5C category. One option is to focus on the 1C competition at 90% DoD, where can i try to find samples of Molicel P42A and Samsung 40T3 in my shelf.

 

[thundercamel]

Thanks for your work Pajda!

This is just my voice, but high capacity cells at 1C are what applies to me.
Originally bought LG MJ1 which vendors stopped selling (I believe due to vapers suing after improper usage?)
Bought Samsung 50E until they became hard to find
Considered Lishen LR2170SD
Bought some sort of Panasonic 21700 3.7v 4.8Ah 7A cells at a great price

Edit: just read the thread title. My opinion might not apply here :lol:

 

[Pajda]

Small update after a while

SAMSUNG INR21700-40T3
- I recently started a test at 1C 90% DoD. To maintain consistency, the results will be placed in the chart for the HP 1C category, though the results can be directly comparable to HE cells.

Molicel INR21700-P42A
- will start this week

 

[eMark]

Would be interesting to see a graph at 250 cycles and at 500 cycles for comparison to this ...

How would you describe what's going on with the erratic green line of VTC6? Is it also possible it too should've stopped at 700 or 750 cycles instead of 1000 like that of 30Q141 and 30Q6 that stopped at 700(Q6) and 750(141) ?

Was under the impression that 'Q6' was an improvement over '141' and would've survived for 750 cycles and '141' for only 700 cycles. Opps, it was Q6T that was supposedly an improvement over '141' and not 'Q6' ... right?

How do you go about pre-testing these six cells as Grade A and not borderline GradeA/GradeB ? Probably next to impossible to determne even during your cycle life tests of these six high energy cells ?

 

[thundercamel]

I would guess that the VTC6 green line shows performance recovering once the cell gets warm during discharge.

 

[eMark]

I would guess that the VTC6 green line shows performance recovering once the cell gets warm during discharge.

Would've liked to see all 10 cell graph lines (as in Pajda's other two 5C graphs) instead of just 6 cells. Especially to compare the graph line of 25R8 and BAK CNP25 to the two 30Q and the cycle life difference (e.g. 25R at least 600 cycles ?) ... https://endless-sphere.com/forums/download/file.php?id=308392

Of 6 popular e-Power tool packs (still use 18650 cells) 5 use 25R cells (including Milwaukee) with only RIDGID using VTC6 ... https://www.youtube.com/watch?v=Plhxnrba2D8 ...
e-Power tools changing over to 21700 cells (more sustained Bang for the $/£).

 

[Stealth_Chopper]

"How would you describe what's going on with the erratic green line of VTC6?"
'Looser'

I would guess that the VTC6 green line shows performance recovering once the cell gets warm during discharge.

Would've liked to see all 10 cell graph lines (as in Pajda's other two 5C graphs) instead of just 6 cells. Especially to compare the graph line of 25R8 and BAK CNP25 to the two 30Q and the cycle life difference (e.g. 25R at least 600 cycles ?) ... https://endless-sphere.com/forums/download/file.php?id=308392

Of 6 popular e-Power tool packs (still use 18650 cells) 5 use 25R cells (including Milwaukee) with only RIDGID using VTC6 ... https://www.youtube.com/watch?v=Plhxnrba2D8 ...
e-Power tools changing over to 21700 cells (more sustained Bang for the $/£).

 

[eMark]

How would you describe what's going on with the erratic green line of VTC6?

'Looser'

The reason i mentioned six popular e-Power tools that use 1865s is that only the RIDGID e-Power battery that was examined used VTC6 3Ah cells (superior quality). The elite professionals that can afford to purchase RIDGID e-Power tools also expect long lasting performance. The other five e-Power tool manufacturers (including Milwaukee) could only afford to outfit their 1865 packs with the less expensive 25R 2.5Ah cells.

I finally found a sample of the cycled VTC6 and added this to the graphs and table. This VTC6 sample lay in a box at room temperature for about 2 years after the end of the 1000 cycle of Standard power rate test and after then the final test for the graphs and table was performed.

Question for SC: Why would Pajda include that specific VTC6 cell with the intent to degrade it as a " 'looser' " ? Your scenario is hard to fathom as Pajda's intent was not to portray VTC6 as a 'looser'... rather quite the opposite.

 

[Pajda]

mall update after a while

added full data for 25R8 and CNP25, unfortunatelly I cannot find a 21700VTC6 Sony 1000cycle sample

 

[Pajda]

small update after a while

added more graphs representing also 3C discharge rate for NEW and after 1000 CYCLE (both voltage drops + temp. rise)

 

[Pajda]

small update

added SONY US18650VTC5A to the chart

 

[Pajda]

I managed to get samples of new cell models for tests from the high power category thanks to the support of NKON and Vapcell. :thumb:
Very soon I will start adding data for:

LG INR18650 HJ2 (a HG2 successor)
Molicel INR21700 P45B

 

[vanturion]

Excellent, will be great to see how the newest iterations of high power cells stack up. Thanks for keeping the updates going!

 

[Pajda]

small update:

Unfortunately, the Samsung 50S sample decided to die suddenly after 952 cycles in the 1C test. The cell did not finnish the CV charging phase, because the current flowing through the cell did not drop below 200 mA (the set cut-off was to 100 mA). After disconnecting the cell start losing the voltage to zero. It was not the first time when I saw this behavior. So the chart will not continue. At least in near future the Moli P42A and Samsung 40T3 will finnish the 1000 cycles.

The second unfortunate thing is that I melted down two BF-2A fixtures when cycling the new Moli P45B at 5C. So it will take some time to solve this problem, otherwise the results first 100 cycles were already available.

The positive are preliminary results of new LG HJ2 (HG2 successor). Both DCIR and particularly cycle life looks very good so far.

 

[riba2233]

Thank you for doing such a huge favor for the community!

 

[Pajda]

Added results after 1000 cycles for 40T3 and P42A at 1C discharge rate, updated graphs and table

 

[Sajeel]

Thank you for sharing your awesome work!

How many sample cells are you using for testing each cell?

I wouldn't mind donating and send some Samsung INR-21700 50G your way, if you'd be willing to add them to your list of tested cells. They boast about 80 % capacity after 1000 cycles. Might be interresting, no? PM me if you find it interresting .