My testing methodology for new 18650 cells, please criticize

Hello all,

New to this forum and new to the DIY battery universe. I try and do my research but don’t have hands on experience yet. I need to test 18650 cells before using them, I am therefore trying to come up with a good 18650 testing methodology for the packs to come.


I am building a few 3s6p packs made of protected 18650 cells ( product ref. “LiitoKala 18650 3400mAh battery 3.7V Li-ion Rechargebale battery PCB Protected For NCR18650B 18650 3400
“ : ”404 page ). These packs will be used to power a device using at max 3W and most of the time 2W (not much power) but for a long time (like 90 hours or so).

The reason I am testing the cells is to avoid ruining the pack with one bad cell and to arrange them efficiently using a calculator like for example.

The tools I used for the testing:

Capacity testing : one LiitoKala lii-500
Resistance testing: one SM8124

What I did/my results :
I am starting by testing a batch of 20 18650-cells, I labeled each one with a number to keep track.
I did several capacity tests using the LiitoKala lii-500 (in vertical position with a small external fan cooling the charger because it can get hot at 1000mA ). I also did resistance tests using the SM8124 because the resistance test of lii-500 doesn’t seem accurate (fluctuates too much).
Results raw data for the capacity test:
Results raw data for the resistance test:

Interpreting the resistance results:

Not much to say, the tester seems fine, cells have an average 55mOhm Resistance which seems ok for protected cells.

Interpreting the capacity results with a graph:

The capacity test fluctuation is too inconsistent (in some case 8% deviation like for cell 19) so something isn’t right.

Criticizing my results/methodology:

Not sure how the PCB affects the results.
The resistance test using the SM8124 seems consistent.
The capacity test fluctuation is too inconsistent (even 8% for cell 19) but maybe 1000mA is too much for parallel testing on lii-500 and not all tests are under 1000mA.

What I plan to do:

I ordered 4 more lii-500 a few days ago when I saw it took so long to test, so I will be able to test my 20 batteries at the same time and reduce to 500mA test so it won’t heat up and cool passively. Then I will do the same 3 or 4 tests and compare data.

What do you guys think ? Any help is appreciated.

Regarding capacity testing on the Lii-500 Engineer.

When doing NOR testing, selected current = 1000mA or 700mA, the discharge current is just 500mA.
When selected current in NOR test is 500mA or 300mA, the discharge current is 250mA.

I think capacity test for a 3400mAh battery are usually 0.2C (ie. 20% of 3400mAh or 0.68A discharge current). 500mA discharge current should be close enough. using a 250mA discharge current probably won’t be that truthful… You are using the NOR test instead of the Fast test, right?

refer to HKJ’s review of the Lii-500 Engineer:

Thank you for your response. Yes I only did NOR tests, as you can see in the link to my results.csv . The issue with 4 slots 1000mA NOR test (so 500mA discharge) is that it heats up, that’s why my reasoning was to do 250mA discharge (so 500mA NOR test). Why wouldn’t a 250mA discharge current be “truthful” ?

Btw for my use case, if I suppose my 2W constant load application, it makes (2,000/3.7/18) around 30mA constant load for each cell (in theory).

I saw the HKJ’s review of the Lii-500 Engineer, and don’t quite understand everything, but it’s strange his resistance tests are much more cohesive than mine (maybe because PCB?).

So what do you suggest ?

I’ve had 3 Lii-500. On all of mine the resistance test was worthless. Frankly, I wouldn’t believe it. You have a superior resistance tester anyway.

I’m surprised at the extreme spread of capacity on new Panasonic cells, unless they are fake. There have been some problems with fake cells at AliEx. There’s a thread somewhere on that issue.
I have seen cells that have been sitting around a good while need a few cycles to ‘wake up’ for want of a better term.

FWIW if you are building/soldering a pack together you don’t need protection. That will be provided by the external protection circuit you will add, or by a balance tap-unless you have some kind of holder you are able to put cells in and out and still get the pack configuration you want. I haven’t seen anything of that configuration that would work. Maybe in the E-bike realm?

Just built a 3S4P pack with balance tap for a friend. Largest pack I’ve ever built. Original pack/cells came out of a dive light. Simply would not perform, it was pathetic. Pulled the pack apart and tested all cells. Every one was OK. Likely the PCB board went bad, not the first one I’ve seen that way. Getting so I don’t trust those things. :person_facepalming: Got my early lithium experience with RC lipo cells, so am accustomed to balance taps.

Soldered them all back together, this time with the balance tap > works fine. :+1: But, I did have to give him one of my hobby chargers to charge it now. :innocent:

1. Do not get cells from ANY OF the Liitokala stores:

2. Get cells from QueenBattery instead:

3. Get a constant current tester. What you have for an internal resistance meter is excellent already, but the testers aren’t.

I’ve found the Li-500 capacity (NOR) test to be fairly consistent. It might give a difference of around 2% from one test to another, but not anything like the 8% you describe. Perhaps your unit isn’t working properly.

However, I usually do my testing at the 250mA drain rate (running the NOR test at 300mA or 500mA). As you mentioned, it gets too hot when testing 4 cells at the 500mA drain (700mA or 1000mA setting). Perhaps the 250mA rate is more accurate.

As others have mentioned, ignore the Li-500 internal resistance values. They’re really inaccurate.

Thank you for your responses,

The “wake up” you are talking about seems very relevant, the first test I did on my 4 “first” (labelles 1 2 3 4) cells had very bad (around 2800mAh) results for cell 1 and 2. I didn’t include the first test (one NOR test for cell 1,2,3,4) in the csv because I thought I had made a mistake in the test somehow since it was so far off.

My reasoning behind the protected cells was that if one cell goes bad, at least it will shut itself down before catching fire or other catastrophic event since I want to be able to use those batteries for like 100 cycles over 3 years. I will be using a BMS (not hobby charging, too inconvenient on the long run since I will be charging like 15 packs at the same time after each use) for each pack but I thought adding the individual protection included in each cell was safer. On a side note I made the acquisition of a great spotwelder (kweld) which is working great on all the tests I did. I will therefore not damage the cells with hot solder heating up the cells too much during assembly.

I’m sorry to hear about your bad experience with BMS, I heard quite a lot about the no-bms “trend”
(for example How to build a safe and better lithium battery WITHOUT a BMS. - YouTube) and they make a good poitn but usability is a big factor, my batteries being enclosed in a box with a xt60 female plug to charge from the outside.

@BlueSwordM :

1 and 2. Haven’t read it all yet but this seems a little scary. I will keep looking into this.
3. I already ordered this CC tester ( ) two weeks ago , should arrive soon, but I was thinking of using it on the assembled battery and not each cell since it would take so long to test each cell individually, or so I thought. In case this CC tester is any good, what 18650 cell “holder” would you recommend ?

@WalkIntoTheLight :

Yes perhaps my unit is bad, I have 4 extra Lii-500 I ordered last week that should arrive in the next weeks, I will be able to test if it was my unit that was bad or simply the model. And also test at 250 and not 500. It is also in theory possible that the PCB in each cell is so unprecise it cuts the power prematurely sometimes (wild guess though).

Can’t comment on the reliability of the ‘double’ protection methodology. Something about it doesn’t feel right though.

What the defective BMS seems to do is take unbalanced cells, and starts corralling them in at the top and bottom. They don’t “balance”, though you can get more expensive ones that do apparently. What they do it prevent any single cell from getting over or undercharged. Since you always have some level of imbalance it seems the swings from charged to discharged get smaller and tighter with each cycle until the pack effectively has no top or bottom end.
I don’t know if this is a characteristic, a defective BMS, or just a lousy design on a cheaper pack. The experience I’ve had has been people bringing me by packs that were not working right and wanting someone with at least a bit more expertise than they had (essentially none), to help them out. I don’t have tools to check BMS, nor the knowledge to do so. So, I take the pack apart and test the individual cells. So far, in my small sample, the individual cells have always been good, all of them, when the pack was performing badly.

Good luck on a big project.

If somebody says criticize me, the first thing that comes to mind is, Your ugly. :innocent:

@BlueSwordM : Getting back on your first point. After reading the thread it is indeed horrifying. I “dissasembled” one of the cells ( I ordered 40 at the time, I didn’t use one of the 20 I already “tested” but one of the others, so I can test the first 20 again with the other lii-500 that are expected to arrive soon).
Nothing is written on the cell, here is a video of the top of the cell I case the “form factor” can reveal a bad cell : dissasembled18650protected .

Yeah, you got stolen out of your money. Real NCR18650Bs have 3 legs on the positive pole.

Ask for a half refund. They will probably not accept a full refund, but a half will probably pass.

@BlueSwordM : ok seems like it may very well be the case, I’m taking this aspect of this thread on your “Liitokala Aliexpress Stores Battery Fraud: Massive fraud on 18650s in the Littokala Aliexpress stores! ” since it is off topic with the test methodology itself.

That being said, having bad (or simply good but not Real NCR18650Bs) cells doesn’t explain the disparity in test results. Maybe it is the PCB, or the tester that are bad. I have more lii-500 coming and one CC tester ( ) that should arrive soon so I will be able to do more tests.

Yes your capacity testing spreads out too much crevettedragon. This can of course be caused by slot variation, but is also cut-off influenced by inconsistent rail contact resistance. To fix the latter you should insert and wiggle the cells in the rails a bit to seat 'em properly, then power up the Lii-500 and check all IR readings. They should be within a narrow marging, e.g. ±10%, or as close as possible (cell internal resistance variation counts too).



Ok I am doing two consecutive 500mA NOR tests with the lii-500 with the cells well seated, on the same slot, and without touching them between two tests.
Will keep you posted with the results.

In the meantime, I am considering buying a EBC-A20, but need some research to confirm and also look for compatible cables (seems to be aviation connectors) for xt60(to test the assembled battery) and compatible cables+battery holder for testing individual cells.

@Barkuti Following up on your suggestion. I did 3 nor test on the same cells in the same slots ( I actually didn‘t even touch the cell to not make it move in the slightest between tests ). Took some time at 500mAh but here are the results:

So measured capacity stays roughly the same across the 3 tests (max deviation being 2.5). But as you can see the resistance test is nonsense: not only is the value absurd compared to what is measured with the much more trustworthy SM8124 but it also goes from simple to double with the cell staying at exactly the same place (I even unplugged the sector side after full charge before a new NOR so it stays still between tests) so I’m not sure how one could be expected to have the IR be withing a 10 narrow margin unfortunately.

I will test cell 2 at 1A using the CC load tester I ordered before which just arrived.

Getting frustrated so I also ordered the EBC-A20.

Will keep you updated

Told you, Lii-500 IR is useless. Maybe lucky individuals got a unit that worked but I had 3 and none worked.

For what it's worth: LiitoKala/Colaier Lii-500 inconsistent IR measurements FIXED

I take great care to minimize variation and even then it occurs to some extent (check here for example), this is normal. Contact resistances are normally high by themselves even when the surfaces are clean and tightly coupled to, let alone if they're not.

The order of magnitude of a cell's internal resistance is close to that of a contact, expecting to measure it this way is rather tall order.

Cheers ^:)

While it may be consistent, I find that my Lii-500 underestimates Li-Ion capacity during NOR test by about 5-6. That’s just by comparing my results when using 500 mA discharge rate to values tested by HKJ for various cells. HKJ may be using slightly lower cutoff voltage, but that in itself couldn’t explain the 5-6 difference I am seeing.

For example, using Samsung 35E cells, I get about 3175 mAh, while HKJ got 3379 mAh.

Does anyone else’s Lii-500 behave the same way?

Can anyone who owns both Lii-500 and some other analyzing charger (such as Opus 3100) comment on differences they’re seeing when performing discharge capacity tests on the same cell?

What do you both mean by consistent?

First of all let me say the Lii-500 scores a mAh when it is 100% done. If it were to measure cells with less than a full mAh of capacity, it will never score a single mAh point. This causes all capacity measurements to err down by half a mAh. To solve this let me say the capacity counter should update half a lapse before. Whether this is common in analyzing charger capacity tests or independent reviewer tests I do not really now, and in any case it does not make it or them exact. It is good to know in any case, and henceforth my capacity measurements will show 0.5 additional mAh.

First of all I will say that as early adopter I ordered a couple Lii-40A cells in early 2018. That 3rd of March I got 3865.5 and 3898.5mAh for them in my Lii-500, slots 1 and 4 as you may understand. It was a tight fit, both cells ended up slightly arse dented. This is a different slot and cell 3882mAh mean, while Henrik got 3960mAh.

Since every slot in a Lii-500 operates independently, all of them display measuring differences. I made a small test to get slot deviation info by testing the capacity of a couple old, laptop pulled Samsung ICR18650-26C cells. Since cell capacity theoretically degrades with each cycle, to cancel this effect I tested each cell two times per slot in a pendulum swing motion arrangement, 1 - 2 - 3 - 4 - 4 - 3 - 2 - 1 and 4 - 3 - 2 - 1 - 1 - 2 - 3 - 4. Between parenthesis is the slot and cell measured DC IR, in mΩ:

Cell E.-

Cycle 1 slot 1 2173.5 (55), slot 2 2210.5 (64), slot 3 2162.5 (72), slot 4 2187.5 (56)

Cycle 2 slot 4 2190.5 (49), slot 3 2181.5 (62), slot 2 2182.5 (63), slot 1 2129.5 (67)

Cell E slot 1 mean 2151.5mAh

Cell E slot 2 mean 2196.5mAh

Cell E slot 3 mean 2172mAh

Cell E slot 4 mean 2189mAh

Cell F.-

Cycle 1 slot 4 2155.5 (55), slot 3 2130.5 (55), slot 2 2146.5 (68), slot 1 2086.5 (55)

Cycle 2 slot 1 2089.5mAh (53), slot 2 2144.5 (67), slot 3 2127.5 (63), slot 4 2113.5 (62)

Cell F slot 1 mean 2088mAh

Cell F slot 2 mean 2145.5mAh

Cell F slot 3 mean 2129mAh

Cell F slot 4 mean 2134.5mAh

Slot 1 average = (2151.5 + 2088) ÷ 2 = 2119.75

Slot 2 average = (2196.5 + 2145.5) ÷ 2 = 2171

Slot 3 average = (2172 + 2129) ÷ 2 = 2150.5

Slot 4 average = (2189 + 2134.5) ÷ 2 = 2161.75

Soo many variables LoL! O:)


Wed, 05/22/2019 - 19:35

Well, I did say,

So, it seems your unit is about as accurate as mine. About, +/- 2%.

In addition, I don’t think it’s necessarily giving the correct absolute value, but if it reads a bit high all the time or low all the time, that allows me to compare cells to one-another, or in the future.

IMO, that’s good enough, considering the Li-500 is a budget charger/analyzer. I’d expect more accurate and consistent readings in an expensive analyzer.