High drain 21700 - Samsung 30T discharge test

[icherouveim]

Thank you for your contribution, I think you are doing great. Nothing is absolutely perfect in this world. We need guys like you to help us with our decisions and to watch technology progression.
Please keep doing the good work with other interesting cells like this

 

[thunderheart]

Thank you for your contribution, I think you are doing great. Nothing is absolutely perfect in this world. We need guys like you to help us with our decisions and to watch technology progression.
Please keep doing the good work with other interesting cells like this

ευχαριστώ, icherouveim Yeah, you're right, nothing is perfect, neither the batteries, nor our testing equipment, but the latter i try to improve from time to time (without getting fanatic about that)))). I'm gonna do some charger tests in the future though i don't know will it be interesting for ES people

 

[madin88]

Now tell me about what consistent 0.3V or 10% difference you're talking about? If my holder is a "bad" one with higher resistance then i'd get the same picture (higher HB4 results) but with a bit lower capacity and bigger voltage drop. BUT my results show that HB6 is better. Once again, if my holder's resistance would be higher, the only difference with lygte's result would be the lower capacity and bigger sag. Verdict: the difference in results is not caused by the holder.

When i compare the discharge graphs of HB4 + HB6 from lygte and yours than i can see following:

On your 20A graph both cells have around 2,9V after 1000mAh drained, while on the 20A graph from lygte the voltage is around 3,25V after 1000mAh

thats exactly 10% difference :wink:

The data i get is more than enough clean for almost any usage. I don't aim to achieve 0.0000000001% accuracy - it's useless in case of batteries.

The question is why there is 10% difference in voltage sag between your tests and others, and why the 10A and 15A curves of the 30T are exactly similar.
If i would do such tests, investing time and effort into this - which is a great thing and BIG THANK YOU for that! - i personally would not take it as "nothing is pefect" but rather like to find out the reason for that.

 

[thunderheart]

thats exactly 10% difference :wink:

Again the same song))) I repeat for the 3rd or 4th time - how do you explain that lygte's tests show that HB4 is better and mine shows that HB6 is better? 10% difference?))))))))))) Higher resistance?))))))))))

The question is why there is 10% difference in voltage sag between your tests and others, and why the 10A and 15A curves of the 30T are exactly similar.

I don't know. Ask lygte's tester, bearing in mind that even the cells from the same batch are not completely identical

If i would do such tests, investing time and effort into this - which is a great thing and BIG THANK YOU for that! - i personally would not take it as "nothing is pefect" but rather like to find out the reason for that.

I've updated my holder 4 times. What do you think why did i do that? If i had a lab and tons of money, i'd build/buy an almost ideal testing platform. If i was the president of Russia i'd... Well, i think i'm going too far))))))))) I've got what i've got and i try to make it better as much as i can without becoming a fanatic.

I appreciate criticism and bright ideas but i don't understand chasing 0.00001% accuracy which is absolutely useless when you test batteries. Try to test the same cell 10 times in exactly the same conditions and you'll never get 2 identical results. Even bigger will be the difference when you gonna test 2 cells and even bigger when they are from different batches.

 

[Hillhater]

Lygte's results are not the only ones showing that difference.."mooch's" results on ecig that i posted are also the same as lygte....approx 10% higher voltage across the various cells i checked !
Regarding the HB4/6 tests, may well highlight the significance of the problem you have.
Lygte's worst result (HB6) is constantly higher voltage than your best results, and consequently has a higher Wh capacity.....but not as high the the HB4 ! .where he recorded an extra 600mAh (12% !) capacity more than your result (@5A)....which is why they rate the HB 4 as the better cell.

 

[madin88]

I've updated my holder 4 times. What do you think why did i do that?

the problem is not the holder itself, the problem is that this battery tester measures the voltage over an alligator clip some cm away from the cell where up to 20A going through. i am not saying you do something wrong, what i am saying is having a common clip for A and V is not optimal.

I appreciate criticism and bright ideas but i don't understand chasing 0.00001% accuracy which is absolutely useless when you test batteries.

If you talking about Ah, yes you have right they don't change much, but as the tester measures about 5-10% lower voltage at 15-20A, the watt hours will be by same percentage off.

 

[thunderheart]

Lygte's results are not the only ones showing that difference.."mooch's" results on ecig that i posted are also the same as lygte....approx 10% higher voltage across the various cells i checked !
Regarding the HB4/6 tests, may well highlight the significance of the problem you have.
Lygte's worst result (HB6) is constantly higher voltage than your best results, and consequently has a higher Wh capacity.....but not as high the the HB4 ! .where he recorded an extra 600mAh (12% !) capacity more than your result (@5A)....which is why they rate the HB 4 as the better cell.

I don't have any problem, the problem is on your side)) I repeat for the 5th time: how can it be that HB6 appears to be better than HB4 in all my tests if my results are "constantly about 10% lower" than lygte's? I remind that lygte's tests show that HB4 is better than HB6.
If i had a problem with resistance/voltage sag i'd have the same curves as lygte just with lower capacity and energy.

the problem is not the holder itself, the problem is that this battery tester measures the voltage over an alligator clip some cm away from the cell where up to 20A going through.

The same thing you can see on lygte's holder where each terminal is placed in 1cm from the battery and they are not made of copper, if you look at their color. I'm tired of repeating the same thing over and over and over again... Please read previous comments.

If you talking about Ah, yes you have right they don't change much, but as the tester measures about 5-10% lower voltage at 15-20A, the watt hours will be by same percentage off.

The tester measures the voltage which the cell has so there is no "5-10% lower" problem. Just write somewhere with big letters that all the cells are not same and that voltage drop from my holder to the measuring point is so low that i can ignore it (thanks to pure copper). Once again, read my previous comments to not make me repeat, please.

 

[thunderheart]

P.S.
My UNI-T UT61E's reading with shorted probes:

And the resistance between the battery contact area and measuring point:

 

[larsb]

I think the key point here is the 10&15A charts being the same. It's physically impossible to get them that similar with 50% higher current in one test without something being off.

Could be:
Measurement or setup error?
Or
Two cells with a 50% difference in IR?
Or...

Anyway it needs to be retested, otherwise there's a possible error someplace that might affect all tests.

On another note:
What i miss most from the testing available at dampfakku/lygte/ecig is to have also a cycle life test. If a cell is 20% more expensive but lasts 40% more cycles etc....
Battery value cannot be judged without it.

Anyway, thanks for the effort, keep up the good work and find that root cause!

 

[liveforphysics]

The higher rate cells use inherently more reactive anode/cathode materials and/or higher surface area particle coating blends.

This means over time, they have increased nucleation sites with high reactivity, and have proportionately increased rates of calendar decay mechanisms.

What this means, is when comparing a couple cell types, if one of the cells was manufactured a year before the other, that cell would be expected to see a 10-25% impedance climb depending on storage voltage and temperature.

However, my complaint with the data wasn't with the HB(x) cell tests, but the 30T test data with the 10amp discharge curve fitting the 15amp curve means your data collection process has some non-trivial error source that should be addressed before sharing more incorrect datasets which others may be confused into believing represent good data.

I suggested the mechanically segregated voltage sense whisker probe touching a few mm away from the current source/sink connection, not because I'm fussy about 0.01% accuracy, but because I wasn't able to get even 1-2% repeatability from any fixture that didn't use separated voltage sense, but in this case the error margin is in the 10-15% range, but you gotta close in on that error margin by tightening up all the sources of measurement error you can find and correct.

 

[Pajda]

I think this kind of discussion is very beneficial for all of us, who are interested in battery testing. I we have been testing battery testing for the past 10 years. I started without any experiences and more important without any advices from more experienced colleagues. The result was as expected that for the first 4 years I have measured totally useless nonsenses. Which is sad, especially when measuring cycle life test, where 1000 cycles with 0.5C-1C setting took about 6 months.

For this particular problem I add some theory:

1) The 4-wire principle of cell connection to the tester is crucial and so the design of cell holder/fixture is also crucial. Especially for reliable repeatability of measurement. The most important is complete isolation of the voltage sensing from the chg/dchg current flow. The most reliable way to achieve this for small cylindrical cells is the fixture contact terminal with the inner insulated needle for voltage sensing.

2) If you have isolated voltage sensing, directly connected to the cell surface, you do not have much to worry about the contact or wire resistance within the current path. It actually does not matter for the current measurement where the power is disipated. The only problem with higher terminal contact resistance is the local heat production on the cell surface.

Here you can see the idea of an "optimal" design of small cylindrical cell connection. The roughened surface combined with smaller pressure is always more reliable than flat connection under high pressure. Similar design is commonly used by professional companies like Arbin or Novonix.


The manufacturing quality of the cell fixture is also crucial. Any misalignment of the cell in the fixture can cause significant measurement error. There are available few types of cell fixtures using this design but most of them suffers from poor manufacturing quality which leads to the problem with cell alignment within the fixture. This BF-2A fixture available on Fasttech or Aliexpress is one with the satisfactory manufacturing quality.

 

[Pajda]

My thought for this case is that both thunderheart and Lygte cell holder designs are similar and both are missing the isolated voltage sense which is not optimal and each small misalignment of the cell in the fixture easily may cause significant voltage drop. I have seen this behaviour many times ago when I was experimenting with similar design cell fixtures.

The second discussed problem is the similarity of Samsung 30T 10A and 15A discharging curves which is definitely an error. We can only talk about how it was made. Such mistakes happen quite often and personally it has happened to me a few times too. I think it could be made in the EBC software, where you maybe did not confirm the change from 10A to 15A and then run the measure still with previous 10A setting, but you save it as 15A curve. If you have stored a measure data in the .csv or .dat files for both curves, please post it here, as they can easily explain this issue.

 

[icherouveim]

Good morning!

Sorry this is out of the topic but can you please test the new very promising high power 18650 cells Sony VTC6A and Samsung 20S?

There is one more VTC5.. new 2018 cell but I don't remember exactly the code.

https://www.e-cigarette-forum.com/threads/bench-test-results-sony-vtc6a-18650-samples-beats-vtc5a-estimated-2900-3000mah-20-25a.849291/

https://www.nkon.nl/rechargeable/18650-size/samsung-inr18650-20s-2000mah-30a.html

 

[liveforphysics]

Thunderheart-

I thought of an example that may help you. Right now, the circuit your graphs evaluate includes both the cell AND the voltage drop of the can to copper contact interface.

To create good data of the cell alone, you must take data from the cell end alone.

Your current source will automatically rise and fall with voltage to compensate for the current injection point variable pressure contact resistance, and your voltage monitoring leads will be correctly getting data from the can end.

It's not even related to the handful of uOhm in your big copper lug, it's the variance though the current injection contact interface which makes artifact vs good data.

 

[thunderheart]

Sony VTC6A and Samsung 20S?

Hi. I've got vtc6, but not yet vtc6a or 20s.

To create good data of the cell alone, you must take data from the cell end alone.

Yeah, let's spot weld the copper wires to the cell end))))) And i must pay attention to the position of stars in the sky because with wrong star position i gonna have 10-15% lower results than someone on the web))))

It's not even related to the handful of uOhm in your big copper lug, it's the variance though the current injection contact interface which makes artifact vs good data.

And the position of stars in the sky, that's important!)))

 

[liveforphysics]

Once you get a setup collecting useful data, it will transform your efforts from outputting misinformation to information. Before that, the inherent large error margin in data leaves your efforts as busy work.

It's not very funny or helpful to spend testing effort only to end with poor results. Now you at least know what to do if you care about sharing useful data.

 

[thunderheart]

liveforphysics

Please stop trashtalking))) Nobody will spot weld wires to battery can just because you think that's the right way and all the other ways are producing misinformation. Just open a beer and have a nice time without masturbating on 0.0001% accuracy. :wink:

 

[madin88]

The same thing you can see on lygte's holder where each terminal is placed in 1cm from the battery and they are not made of copper, if you look at their color. I'm tired of repeating the same thing over and over and over again... Please read previous comments.

Those cable lugs are made of copper and the silver surface is just an anti-corrosion coating..

Please stop trashtalking))) Nobody will spot weld wires to battery can just because you think that's the right way and all the other ways are producing misinformation. Just open a beer and have a nice time without masturbating on 0.0001% accuracy. :wink:

Don't say that he is trashtalking!
It seems as if you still don't understand that your setup measures 5-10% lower volt and Wh compared to others and not only 0.000001% (the higher the current, the worse it gets).
And who mentioned that you should spot weld wires to the batteries?
Your holder could be fixed very easy and everything you need to know for that was written here. However, if you cannot see the problem then i would say do what you like to do, but brace yourself for that sooner or later others will warn you about the same :wink:

 

[Pajda]

thunderheart what discharge current test do you plan for M36 test? I would like to run the same settings to compare the results. From LG M36 datasheet the nominal charging current should be 1 A. Nominal charging cut-off is 50 mA (but our testers can do only 100 mA so I wil use 100 mA) and the nominal discharge current is 670 mA. I will enter 60 min pause between CHG-DCHG. What next 2 A, 5 A, 10 A?

 

[thunderheart]

It seems as if you still don't understand that your setup measures 5-10% lower volt and Wh compared to others

Samsung 30T at 5A:
Lygte's result: 10,818Wh
My result: 10,72Wh
Difference: 0,90%
For 10A, 15A and 20A discharge rates there is no energy mentioned to compare.

LG HB6 at 5A:
Lygte's result: 5173mWh
My result: 5093mWh
Difference: 1,55%

Samsung 48G at 5A:
Lygte's result: 16,116Wh
My result:16,14Wh
Difference: -0,15%
Oops... My setup didn't measure lower Wh... Sorry)))

Start by learning basic maths...

 

[thunderheart]

thunderheart what discharge current test do you plan for M36 test? I would like to run the same settings to compare the results. From LG M36 datasheet the nominal charging current should be 1 A. Nominal charging cut-off is 50 mA (but our testers can do only 100 mA so I wil use 100 mA) and the nominal discharge current is 670 mA. I will enter 60 min pause between CHG-DCHG. What next 2 A, 5 A, 10 A?

M36 test is done. I'll publish it today or tomorrow. Could you send me the datasheet you have? I didn't find any before starting the test(
I've used following values:
Std charge current: 1.8A
Cut-off: 100mA
Discharge tests: 0.72A, 5A, 10A, 15A, 20A
Discharge cut-off: 2.5V

 

[Pajda]

Datasheet is available on Nkon shop. https://eu.nkon.nl/rechargeable/18650-size/lg-inr18650m36-3600mah-10a.html or direct link to download https://nkon.nl/sk/k/m36.pdf I have my M36 cell samples from them.

I will run the test with your settings up to 10 A (limit of my calibrated tester) . By the way you are a courageous man to test this cell with currents higher than 10A.

 

[thunderheart]

Datasheet is available on Nkon shop. https://eu.nkon.nl/rechargeable/18650-size/lg-inr18650m36-3600mah-10a.html or direct link to download https://nkon.nl/sk/k/m36.pdf I have my M36 cell samples from them.

I will run the test with your settings up to 10 A (limit of my calibrated tester) . By the way you are a courageous man to test this cell with currents higher than 10A.

Sorry, i did a mistake))) i just was thinking about another cell just before answering))))))))))
0.72A, 2A, 5A, 8A, 10A
So i'm not that courageous)))))))))))))))))))))))))))))
Big thanks for the link!!!!

 

[liveforphysics]

When 10amp discharge looks different than 15amp discharge, then you will know the data you've taken and shared is offering something meaningful vs misinformation.

Im not trying to be mean, Im genuinely grateful for your efforts (and anyone else who wants to test cells) and trying to encourage you by giving you the best cell testing setup advice.

 

[thunderheart]

liveforphysics

Never wanted to offend anyone if ever did. I appreciate criticism and i heard nice ideas here but i don't accept when obvious things are turned over and presented as measurement error. Above i listed some comparison with Lygte's results and the difference is -0,15% to 1.5% (while my tester's official accuracy is 1%) so talking about some "constant 5-10% lower results due to measurement setup" is just making me advertising simple calculators here.

About improvements: i started improving my setup right from the first days when i was using steel pads instead of copper ones. I was thinking about that almost every day without any criticism here or anywhere else))) You did see my holder at versions 1.0, 2.0, 2.5 and gonna see the version 3.0 very soon. Of course, i could build a lab with the very accurate equipment and the best holder with measuring right on the battery but i have no money for that equipment and no time to make the holder according to the design which i carry in my mind. Maybe some day i'll announce the next version of my holder))) but not now, that's for sure)) Cheers