Results of Capacity Testing 18650s for 1,600+ Cycles

[diegobg]

@Diegobg - Nice work! The 2 Amp discharge is putting more stress on the cells, so it looks like they are degrading faster, about 9% loss in capacity, right? Are the cells heating up at all?

Thanks!It is right. For now, the Liitokala NCR18650B cell has lost 10% or so degradation in 300 cycles.On the temperature, perhaps they heat up a little more than in the Opus test at 1Amp, but not much more. I think that in Opus the cells get very hot from the heat generated by Opus itself.

Now I am doing the same test to the cells of Varicore (another Chinese cell). These have degraded very quickly at first, but after 60 cycles they seem to stabilize. I leave you link:https://electrotinker.com/en/analysis/degradation-ncr18650b-varicore/

 

[Generic]

I skipped the update at 1300 cycles, but here I am at 1400! Here's the graph:





And where we are at SoH:





I also set out to see if degradation is linear or exponential. The CJ and THLD were clearly exponential. The ASO and LG cells, I'm not sure. Here is a chart showing the rate of change. For example, the data point at 200 is the % increase in degradation from Cycle 201-205 over the degradation in Cycle 101-105, and so on. The LG looks fairly linear, while the ASO seems to be going exponential.

 

[Generic]

First off, I made a mistake. I write down every result on a notepad, and then do data entry into excel when I hit a milestone and am ready to report an update. Well, for some reason, I wrote Cycle 1528 twice (...1527, 1528, 1528, 1529...). What I decided to do is average the results of the 2 cycles and use that number as the entry for Cycle 1528. To make up the test cycle, I will skip over Cycle 1666, assuming I get to that point, because I don't like that number :angel: .

Second, I kind of discovered something that I wanted to run by the community. I never really paid attention to the THLD cell's "INR18650" designation, as I understand that INR/ICR/IMR/NCR don't actually signify chemistry and are just naming conventions, but it got me thinking. If the THLD cell is labeled as an INR cell, and it actually is an INR (Lithium Manganese Nickel) vs. being an ICR (Lithium Cobalt Oxide), then the results of my degradation testing don't signify quality, but rather chemistry. It's not that the CJ and THLD cells are Chinese junk, but rather, they are low-cobalt-content INR cells that typically have a shorter life than their cobalt-rich ICR cousins. That also got me thinking again. If the CJ and THLD cells were truly INR cells, why did their PTC protection kick in to limit their short-circuit discharge to just 2 amps and why did the cells still heat up at that low of a discharge rate? My theory is that, since much of the cost of INR cells comes from having thicker copper and aluminum collectors, the Chinese manufacturers used thinner current collectors, while also using the low-cobalt INR chemistry. Thatwould explain how they got their cost down. So high discharge chemistry without the ability to do high discharge. What do you all think about that theory?

 

[Korishan]

That's very interesting.

Another thing to bring to attention is that Cobalt is also hard to come by. So having lower cobalt content also makes sense. Congo is where the worlds largest Cobalt mine is.

Congo is the leading cobalt producing country, accounting for more than 60% of the worlds cobalt production of 140 million metric tonnes in 2018. Cobalt is mostly mined as a byproduct of either copper or nickel. China is the leading consumer of cobalt, driven by the demand from its rechargeable battery industry.

And then Russia at only 250k metric tonnes.

So if there's any dispute between Congo and China, China can't easily switch resources.

I also didn't realize that Cobalt was a byproduct of copper/nickel mining. So that'd be another resource China would be after from them.

---

Side note: I appreciate you keeping your updates on the first posting and helping to keep the thread clean. Good work Generic. You are definitely not a generic user, but very professional imho. Maybe we change your username to Professional instead

 

[OffGridInTheCity]

Continued thanks for doing this.

My oldest battery (14s120p) has just passed 756 cycles at an average 47% DOD with no noticeable degradation - e.g. the packs don't require any balance to keep up withe very newest packs (<100 cycles). I'm hoping that at 50% DOD I'll get at leasttwice the numbers you're getting.... but the chemistry idea is interesting.

Note: The battery above is (aprox)50%LG-MF1-18650s and 50% Samsung-ICR18650-22P... that were 'new' - e.g. that massBatteryHookup ebike cell purchase a couple of years ago+ ebay 36v ebike 'new' packs.

 

[Generic]

Korishan - As far as I know, cobalt is the single most expensive part of an 18650. Which is why I posted the theory above, to get generic cells cheaper, they have to lower the cobalt content. It's not like these things are handmade and they save on labor. Also, it's not me trying to be professional, it's just my frustration with some threads I read that are like 15 pages long and you have to read every post to figure out what is going on, so I did it for people like myself who might click on this thread and have all the information they need just from the first post.

OffGrid - I think you will get A LOT of cycles. Your cells run cooler, with lower demand, and with a smaller depth of discharge than what I'm doing in this test. I honestly thought I'd be done after 100 cycles and here I am 18+months later still going. BTW, not to get too off-topic, but weren't those 22P's found to be fake later on? That would be interesting!

 

[OffGridInTheCity]

Korishan - As far as I know, cobalt is the single most expensive part of an 18650. Which is why I posted the theory above, to get generic cells cheaper, they have to lower the cobalt content. It's not like these things are handmade and they save on labor. Also, it's not me trying to be professional, it's just my frustration with some threads I read that are like 15 pages long and you have to read every post to figure out what is going on, so I did it for people like myself who might click on this thread and have all the information they need just from the first post.

OffGrid - I think you will get A LOT of cycles. Your cells run cooler, with lower demand, and with a smaller depth of discharge than what I'm doing in this test. I honestly thought I'd be done after 100 cycles and here I am 18+months later still going. BTW, not to get too off-topic, but weren't those 22P's found to be fake later on? That would be interesting!

Interesting... Here's a pic of the mix of cells in my 1st battery (discussed above). The 22Ps tested 100% at the time. It was in the beginning for me, when I believed that ebay 'generic' purchases couldn't be 'all bad'. Looking closely this morning, the 22Ps are30% of the battery rather than50% (just to correct the record). The date codes worked - https://batterybro.com/pages/18650-date-code-lookup-tool=> Nov 3, 2015 (for example)lending legitimacy in my mind.




Here's an example of one of the ebay purchases for these 22Ps.... I was trying for 10a, MF1 cells thinking other ebike packs must be similar - and indeed the first fewhadMF1s, but then the rest started coming with22Ps. My purchasing was ill-informed (based on what I know now)so I just got lucky in performance but totally overpaid at $2.25/cell.

 

[slipkn67]

This is very useful for the community! I never thought a 10 years old used 18650 can do 1000+ full cycles at 1A with almost no capacity loss. Did other brands make such high quality cells 10 years ago?
Would also be interested to see similar tests for other popular old brand cells like:
Samsung 22F, Samsung 26F (26C, 26D)
Sanyo UR18650FM and others
some old Panasonic and Sony

Thank your for the work you've done for the community.

 

[humbleONE]

Wow I read the first page of comments and I think it took me towards an hour, but reading six pages is out for now lol
Great to see someone with the mental stamina to do so many tests over sooooo long, how's your brain doing now?

For my powerwall project I just finished charging and testing just over 1000 cells almost with a single tester (4 slot) and it took me probably about 4 months 24hrs/day...
I'll make another post elsewhere on this but after doing it for so long when finished I really felt "lost" not doing the "habit" any more lol it literally took over my life for the time during this idiotic lockdown crap in Victoria's literal "police State".
BTW I live alone.... so a small powerwall will do me!

I was surprised at the start of the posts that LG was being canned as I'd read, heard and believed they were pretty good cells, in fact among the best, hence why they are used in "not so cheap" vacuum cleaners.... good to read near the bottom of the first page they almost outperformed everything else for performance and longevity. Go LG! (LG 2200 mAh cells are the largest concentration of cells I own for my project!)

Cheers all

Oh sh*t I joined here in July so it was nearer 6 months not 4! WOW

 

[enki]

I recreated the data from the newest graph (1585 cycles) for other curious souls using the excellent Web Plot Digitizer tool (https://apps.automeris.io/wpd/). Project attached, note that it is *not* a zipfile but a .tar file. To open it or load into Web Plot Digitizer you need to remove the .zip extension.

 

[Generic]

First off, I want to apologize to everyone for not posting sooner. I meant to log in to post so many times, and I didn't. Today I'm making an effort to post this, so here it goes.

Originally I was going to quit at 2,130 cycles when the ASO cell kicked the bucket, but I kept going because I was still testing other cells for my powerwall, and it didn't take any extra effort to continue cycling, but I did stop taking pictures of every cycle. I got to 2,425 Cycles and finally quit. As you can see, the LG cell was still alive and kicking, even after 2,425 cycles. Would it have made it to 3,000 cycles? I guess someone else will have to find out.





At the 2,130 cycle mark, I retested all the cells at each Opus discharge rate.



Crazy to see that the ASO still had more than half of it's original capacity at the 200mA rate.

One question I wasn't able to answer was whether cells degrade linearly or exponentially. While the INR cells both fell off a cliff, the ASO cell looked exponential and the LG looked more linear.



Some fun facts: (1) The LG cell held as much energy as 1.6 Liters of Petrol, or 2/5 of a Gallon of Gas. (2) Between the 4 cells, I burned off almost 37 kilowatts of electricity through one Opus.

Now that the test is over, one thing I want to mention was how I picked my cells. The LG cell was picked because it was the highest capacity 2200mAh rated genuine cell I had. The CJ and THLD cells each had deep spot-welds at the bottom that I initially binned as having a hole in them, but since they didn't leak, I decided to use them for this test. They held up. Lastly the ASO cell. I binned that cell because it was rusting where the top cap meets the separator between the positive and negative cans. It did rust more during the test, but never leaked. So a "rusty" cell made it multiple years and thousands of cycles. I'll go grab my popcorn. Lastly, I never posted the voltages of the cells when I found them. The voltages of the cells were: CJ - 2.64V; ASO - 1.82V; THLD - 3.51V; LG - 2.55V. I think the ASO did fine for being a sub 2 Volt cell.

I hope I provided some useful research for the community. Thank you to everyone who commented and supported me during this test.

 

[OffGridInTheCity]

Thank you for doing the tests / sharing the info! I (for one) am in this for the loooong haul and the data is fascinating.

Do you think the cells broke down in a dangerous way (get hot or formed evidence of fire risk or ??) as they reached the end-of-life?

 

[Generic]

I came back just in time to see your first pack's 1000th cycle, congrats! The LG cell was warmer than the others for the last 1,000 cycles or so, but nothing extreme. Out of the 4, the LG had the highest IR (probably, I have no means to test that) and for the longest time had the biggest voltage swings. The last few cycles I ran, when I did a Charge-Test, the Opus would go up to 4.28V and then immediately down to something like 3.46V, but as the internal resistance heated up the cell, the voltage would come up and the cell would discharge all the way down to 2.8V. It's been over a month since I quit this test, and the LG cell is sitting at 4.08V. I'll check on it in a few months to see if is a self-discharger. The other cells died gracefully and none of them ended up self-discharging or becoming heaters. I mean "died" in that I couldn't easily get them to degrade any more. The cells are still technically useable, just don't have the capacity to make it worth it.

I just noticed my Excel spreadsheet for 2301 and 2401 vs 1 is wrong, I forgot to change the formula when I copied and pasted. Oh well.

Also, I can't edit the title or the first post because I'm over the 20,000 character limit. That sucks.

 

[Edgar]

Generic: Thank you for sharing your experience with these "generic" batteries, I have several ASOs and had doubts about using them in a power wall. I think they will serve well. Greetings and thanks again!

 

[ferazop]

It is interesting to me that of the four cells 2 are behaving similarly to each other while the other 2 are similar to each other. techzpod download mobdro

 

[paddy72]

I just came across this post and i have to say thank you for this fantastic and long lasting job of testing!

My first question is: is the test data typical and significant for these cell brands / chemistry or is it maybe just statistical spread of the individual cells tested? To proof this you would have to test at least 4 cells of the same type and brand ;-) Better 10 of different production dates..
Or is it more likely that the testdata is really typical for the type/brand of cell? I am really not sure.

Anyway the test is very interesting and it also reviels that NMC chemistry can lead to 2000+ cycles (at least at 50% capacity left) or 1400 - 1500 cycles at 80..75% capacity. On the other hand you can get the same chemistry with just 500 or 600 cycles. So what are the factors that are most important here for cycle life?
And how much deviation is there for the same cell brand/type? The deviation is the most important factor in cell pack degradation and failure!

 

[OffGridInTheCity]

There are A LOT of variables but I'm heavily invested in a large 18650 powerwall - now in my 4th year of operation. There are so many variations but I'm hoping that low stress will yield LifePo4 cycle ranges.

My oldest battery has 1,513 cycles (as of yesterday) with the following general characteristics of operation....
- 3.5v and 4.0v operating range
- average peak charge of 3.84v and low of 3.5v.
- average 37.1% DOD
- low stress <150ma/cell (laptop level cells)

and no obvious degradation so far - no balance needed to keep up with other packs etc.

I'm beginning to think that when decline occurs it may be rapid rather than gradual. I'm not really worried about achieving 2,000 cycles (any more) but if it goes to 3,000+ cycles... which will take another 4 years.... then this experiment will get more interesting.

 

[paddy72]

There are A LOT of variables but I'm heavily invested in a large 18650 powerwall - now in my 4th year of operation. There are so many variations but I'm hoping that low stress will yield LifePo4 cycle ranges.

Did you buy new cells for your pack? If not, how do you know the exact cycles from recyled cells?

I guess with this 'soft' working practice you can achieve the 3000 cycles with NMC cells. Senec, a big german home battery producer claims 12,000 cycles for their modules, but they are build of NMC and NCA cells, afaik. Marketing at its best Some of them burnt down already, probably due to software / BMS errors but the cells also seem to be a problem.

 

[OffGridInTheCity]

Did you buy new cells for your pack? If not, how do you know the exact cycles from recyled cells?

Good point - When I say 1,513 cycles I'm talking about my battery bank.

The specific battery referenced above with 1,513 cycles was sold as 'unused' ebike packs from BatteryHookup and Ebay - and the cells tested 100% capacity. However, all of my cells are (currently) 9-14years'ish old by date code, including the 'new' cells with 1,513 cycles.

There are a lot of variables w/respect to cells and cycles. We take for granted that lithium-ion is way short and lifepo4 is way long but there is some data on lithium-ion in cars over the last 10yrs and some data that low stress = longer life. Lifepo4 adoption is more recent and no reason to think they have short life cycle wise.

However, when talking 10,000 cycles (e.g. 27years) of anything - there's so many other potential issues than chemistry/cycles it's hard to use as the only key data point. But 3,500cycles = 10yrs seems more likely to be relevant and lithium-ion may very well last this long under low stress - so it's an interesting topic to me.