Samsung 50E capacity testSamsung 50E capacity test - 5000mAh or not?

Original article: https://www.thunderheartreviews.com/2019/01/samsung-50e-capacity-test-5000mah-or-not.html

Samsung INR21700-50E is a 21700 size cell which is being sold as 5000mAh one while in its official datasheet the capacity is mentioned as "Min. 4900mAh" and there is absolutely no "5000mAh" mention anywhere.


The battery was bought from my reliable supplier (Queen Battery) and tested with ZKETECH EBC-A20 and a self-made battery holder. It's a PC-connected battery tester supporting 4-wire measuring and discharging at up to 20A.


I've used version 3.0 of my battery holder based on 0.5mm thick pure copper terminals



I've followed all the prescriptions of the IEC61960-2003 standard concerning battery's capacity measurement. Before each discharging cycle each battery was charged at standard charge current mentioned in its datasheet to charge end voltage. Before each discharging or charging i've held a 1-1.5hrs pause. The environment temperature was 23.0-24.5°C. To be sure in results i've done each test minimum twice (usually 3-4 times).

Samsung INR21700-50E
The cell is marked as INR21700-50E SAMSUNG SDI 2I75 which means that the production date is 05 Jul 2018.


The main specs from Samsung 50E datasheet:
Capacity (min): 4900mAh
Nominal voltage: 3.6V
Standard charge current: 2.45A (0.5C)
Charge cut-off current: 98mA (0.02C)
Max charge current: 4.9A (1C)
Charge end voltage: 4.2V
Max. continuous discharge current: 9.8A (2C)
Discharge cut-off voltage: 2.5V
AC impedance at 1KHz: ≤35mΩ
Max weight: 69g

Measured DC IR at 4.9A in fully charged condition was 26.5±0.5mΩ

Measured weight of the tested cell was 68.8g


Capacity test results:

At 0.2C (0.98A) discharge 50E's capacity doesn't reach 5000mAh and is slightly higher than 4950mAh which makes me think that 50E must be classified as a 4900mAh cell rather than a 5000mAh one. All the curves look very nice and smooth. At 10A 50E's capacity is more than 4700mAh which is a good result for a 4900mAh cell.

50E is a slightly better cell compared to 48G but the difference is hardly noticeable but bear in mind that 48G's standard charge current is much lower than that of 50E - 1.44A vs 2.45A.

Here is the video version of this review:


Check out my YouTube channel for batteries, chargers and other stuff reviews.
I've launched my blog where you can find all my reviews in one place. Every new test/review will be first published on YouTube and in the blog. I'll be happy to see new subscribers, comments, suggestions and just your thoughts.

Hmmmm missing true four wire setup for accurate test results.
The terminal sens wires must separately connected to the copper plate.
They can't pass through the crocodile clips!
Are you aware what I'm talking about?

Honk said:

Hmmmm missing true four wire setup for accurate test results.
The terminal sens wires must separately connected to the copper plate."...

Click to expand...

Yes ,.TH has had this pointed out several times, and is working on a revised set up.!
For a true "4 wire" test, the voltage sense connections should be directly onto the cell terminals.

Good timing on this capacity test as I just received a 14s3p battery with them on board.

Hi thunderheart, wish you all well for this year. And wish many interesting cells like Samsung 50E to test. Hopefully you finish BF-2A modification early. Still don´t understant why you need to test more than 10 amps, we are not vaping here.

As DCIR 26,5 miliohm you measured seems to me little bit low value, could you please describe your DCIR measurement ?

Hi Docware! Thanks a lot! I wish you and all of ES members lots of money to build big packs and safe operation. Let the fire burn in your hearts not in batteries!))

I've finished testing the BF-2A without modification and it's a sad sad story... I'll publish the results soon. It definitely needs to be modified.

I test at >10A (when it's possible) because my aim is to test the cell thoroughly not only for using it in battery packs. And i'm sure high discharge rates are good to compare the characters of high drain cells.

DCIR measurement i do at 1C current on a fully charged cell after 1hr of rest. I repeat it 10 or more times to be sure in the result.

Thank you for the high drain test.

BF-2A sad story - you always have some surprise, I am curious
Unfortunately, it is not possible to get correct DCIR values with „EB Tester Software“ automatic „Resistance Test“. You need to have some external voltage logger, fast and precise enough to get acceptable results and do measurement „manually“.

I have new Samsung INR21700-50E from nkon, the production date same as you – 05 Jul 2018 . After second capacity measurement (1 A and 2 A) I have measured DCIR cca 29 – 30 miliohm at 4,17 V and cca 28-29 miliohm at 3,71 V. Measurement according IEC 61960-3 : 2017 , par 7.7.3. Just small modification, I have 0,2C discharge time in the range 10 – 12 s instead required 10 ± 0,1 s . But this time should not affect the result significantly.
Here is record of 3 measuring pulses


View attachment 1

And here is measurement of the voltage difference 0,2C 0,98 A versus 1C 4,9 A. From voltage delta 113 mV and current delta 3,92 A (4,9 – 0,98) we calculate DCIR 28,8 miliohm.



By the way, the cell wall temperature is 20 °C above ambient at 5 A discharge. This is not a high drain cell.

docware said:

BF-2A sad story - you always have some surprise, I am curious
Unfortunately, it is not possible to get correct DCIR values with „EB Tester Software“ automatic „Resistance Test“.

Click to expand...

Neither the capacity and energy are calculated correctly. As i told you, the same data collected by EB Tester Software calculated in Excel gives higher total values. I started to measure DCIR because ES members were asking for and i'm stuck with EB Tester Software to make test results comparable with older ones. If i start calculating in Excel it would make my future tests incomparable with those already done.

I appreciate that you involve DCIR in your measurement as it is important characteristic of the cell.
Just if you are going to measure according to IEC 61960-3 standard, we can compare apples to apples.

ZKETECH equipment is quite OK to measure capacity and energy with reasonable accuracy. But definitely is not appropriate to measure DCIR with ZKETECH automatic Resistance Test.
Again, you need to have external voltage logger and run ZKETECH manually to measure DCIR appropriately.

What about BF-2A sad story ?

Hi thunderheart, here are my results from Samsung 21700 - 50E measurement :
Testing conditions : ambient temperature during testing 24 – 25 °C, charge 2,45 A to 4,2 V, 0,1 A cut-off, rest at least 60 minutes.
Equipment : ZKETECH EBC-A10H, BF-2A fixture


View attachment 2


As my results are higher than yours, I investigated what can be the reason. First, temperature during my testing is little bit higher, but not significantly, my higher temps may add no more than 10 mAh.

I have checked the current measurement and found that EBC-A10H has some different results during long term testing.
So I should subtract cca 50 mAh from result at 1 A and cca 35 mAh at 2 A. Testing at 5 A seems to be OK. Results corrected for current measurement inaccuracy :
1A 4 983 mAh
2A 4 960 mAh
5A 4 952 mAh

You have nearly the same result at 1 A after calculating the temperature difference.
But there are significant differences at 2 and 5 amps. Which invokes question, have you done any current calibration before and during the testing ?

Frankly, my results after the correction are still little bit strange. The difference between the 1 and 5 amps is too small.
Pajda, could you please provide some output from your 50E capacity measurement to compare our results ?

I also measured warming of the cell surface during 5 A capacity discharge test. Used equipment : thermocouple K, wires diameter 0,127 mm (Omega) fixed to cell by Kapton 0,05 mm tape, thermometer Omega HH 520.

I have to correct my previous statement that ZKETECH equipment is OK to measure capacity and energy with reasonable accuracy.
The cooling fan of my equipment is switched-on according the manufacturer algorithm depending on the immediate power and internal temperature. At 1 and 2 A discharging mode is cooling fan usually off. I have found that it is a reason for deviations during measurement.

I suppose that because of that cooling algorithm, ZKETECH measuring shunt has little bit increased temperature and consequently also higher resistance and voltage drop. Current measurement evaluate this as a higher current and regulate it down, resulting to lower real current than setting. So that 1 A setting was giving in fact discharge current in the range 988 – 992 mA. At 2 A discharge the real current was in the range 1 982 – 1 988 mA depending on running time.
The instrument calculates capacity from the wrong values. Resulting effect is unreal higher measured capacity within lower current range.

At 5 A discharge the real current was 5 001 – 5 003 mA, which gives acceptable accuracy, better than 0,1 %.

I have made small modification of my ZKETECH EBC-A10H, which consists of adding DC/DC converter connected to 24 V internal power supply. Cooling fan is now supplied from this DC/DC converter and running continuously. At the same time lower voltage and lower fan speed allows to have less acoustic noise. At 1 A setting instrument now gives 998 -1 000 mA and discharge 2 A gives usually 1 997 - 1 998 mA .

I have repeated Samsung 50E capacity measurement after the modification, see the graph. In my opinion the results are now corresponding.





ZKETECH potential current measurement discrepance and missing true 4-wire measurent may be reason for the significantly different results of thunderheart measurement. Which cast a cloud over all his results, unfortunately. On the other hand, solution may be simple. Of course, the first step has to be current calibration during the whole measuring.

The need of true 4-wire measurement was repeated again and again in the past.

Another reason for different results could be accelerated aging of the cells thanks to his repeated high current testing.
By the way, there is BF-2A with brass current connectors for current 20A instead usual 10 A stainless steel connectors.

Great observation, test improvements and feedback.
Thumbs up!