Li-ion cells cycle ageing

[Pajda]

New batch of pictures : Samsung 30Q 472 cycles, LG MJ1 435 cycles, SONY VTC6 200 cycles, Samsung 21700 50E 165 cycles, Samsung 35E 125 cycles, LG HG2 64 cycles.

50E curve is obviously influenced by the ZKETECH failure.

Yes both cells are definitely artificially damaged by the tester. That is the destiny of those who do not trust the old viwes's stories and trying to get own results with the use of proper measurement principles. It won't help you, but you are not alone who is suffering with a ZKE tester mood. This winter my PC with ZKE software running died and I was not able for two weeks to replace it. Those wonderfull testers over-discharge almost all cells during that time. So I lost several months long measurement for ca 20 cells.

So do not let this upset you and just continue with your great work for the community.

 

[eMark]

Pretty sure their 21700 cell is not Samsung but the Sanyo / Panasonic NCR21700A,

apparently a 5Ah capacity version.

And they will be self-producing soon if not already

Panasonic is hard at work at Gegafactory trying to keep up with 21700 production demands. Rumored that Musk is not happy with Panasonic taking so long getting up to expected production speed with new Sanyo/Panasonic 21700 battery. Inside rumor has it that Musk enlisted the help of Samsung 21700 cells to keep up with production demands for Tesla auto and Cyber truck.

Ubet, i question it too ... do some Tesla 21700 car battery's actually have Samsung cells while most will have Panasonic cells ??

Maybe docware will test newer Panasonic NCR18650B ... https://www.imrbatteries.com/content/panasonic_ncr18650b-2.pdf ... with its newer NNP + HRL technology (production outsourced to China). It could give M36 some stiff competition. Possibly the same chem/tech in Tesla's new NCR21700A cell (not yet available to consumers). Thought to be a high capacity, high drain, fast charge cell (5000mAh, 15A rating ?).

Is it too farfetch'd to envision a commercial with a remote controlled custom Tesla cruising around on the Sea of Tranquility? The Ad Agency would speed with the video so it looks like it's going at least 35mph. Knowing Musk he may have it going faster than that without needing any fast motion trickery :thumb:

 

[docware]

SOC comparative table.
It´s obvious that SOC voltage distribution has no significant change after 700 cycles.

 

[john61ct]

Wow great stuff.

Are those "rested" voltages? for how long, after what sort of discharge?

Just a link would be great if your test protocol's spelled out

 

[docware]

Yes, those are rest voltage values. Testing protocol consist of 20 discharge 2 amp pulses , each pulse equals 5 % of the cell actual capacity (measured as a first step). 20 min rest time between discharge pulses.



In this example was LG MJ1 discharged by 2 A for 292 seconds period each pulse. Rest times 20 minutes.

https://endless-sphere.com/forums/viewtopic.php?f=14&t=103092&p=1523878#p1523878

 

[john61ct]

Yes that's a great protocol.

And interesting how most of the curves end up so linear averaged out.

Shows that the curvier curves from continuous discharge are showing effects that build / accumulate due to the constancy of the load.

While neither reflect "real life" usage, I think yours is much more useful in showing a clear SoC-to-voltage mapping.

 

[flippy]

Yes, those are rest voltage values. Testing protocol consist of 20 discharge 2 amp pulses , each pulse equals 5 % of the cell actual capacity (measured as a first step). 20 min rest time between discharge pulses.

LG MJ1 SOC mapping 2A 292 s pulses.jpg

In this example was LG MJ1 discharged by 2 A for 292 seconds period each pulse. Rest times 20 minutes.

https://endless-sphere.com/forums/viewtopic.php?f=14&t=103092&p=1523878#p1523878

that is a nice way to use the discharge feature.

 

[John in CR]

SOC comparative table.
It´s obvious that SOC voltage distribution has no significant change after 700 cycles.

SOC% versus voltage - after 700 cycles 18.2.2020.jpg

That's the best battery info I've gotten in a long time. No wonder I've only hit LVC twice, once in 2009 and once when LFP was down in 2010, and on most of my ebikes the only gauge I've used of any kind is a $2 digital voltmeter. Thanks for confirming that once you become accustomed to your pack's voltage as a measure of its SOC, that doesn't change with time despite the change in capacity.

Just to confirm, that 50% SOC point is the capacity (mAh) midpoint, not the energy (wh) midpoint, right?

 

[docware]

.... and on most of my ebikes the only gauge I've used of any kind is a $2 digital voltmeter. Thanks for confirming that once you become accustomed to your pack's voltage as a measure of its SOC, that doesn't change with time despite the change in capacity.

Just to confirm, that 50% SOC point is the capacity (mAh) midpoint, not the energy (wh) midpoint, right?

Yes, I agree that cheap simple voltmeter is very efficient SOC gauge, especially for other chemistry like NCA or NMC. Just recommend to check the real values with an external DMM once or twice per year.

Yes, the table is made up from capacity measurement (mAh).

 

[flippy]

you can use a simple arduino that taps off a balance lead and you can use the ADC input to measure a single cell voltage and correct the SOC value with the measured voltage of that single string. arduino nano and a simple lcd/oled is enough.
that way you can make a simple but accurate SOC percentage, if you are really bored you can program in different cell models.

 

[John in CR]

.... and on most of my ebikes the only gauge I've used of any kind is a $2 digital voltmeter. Thanks for confirming that once you become accustomed to your pack's voltage as a measure of its SOC, that doesn't change with time despite the change in capacity.

Just to confirm, that 50% SOC point is the capacity (mAh) midpoint, not the energy (wh) midpoint, right?

Yes, I agree that cheap simple voltmeter is very efficient SOC gauge, especially for other chemistry like NCA or NMC. Just recommend to check the real values with an external DMM once or twice per year....

That gets done more like 2-4 times a year when I manually check balance, since I don't use Battery Murdering Systems. I should skip that step now that I run such conservative charge cutoff voltages (and even more conservative DODs) by going up to an even number of cells (22s instead of my current standard of 21s), and then run 2 voltage displays showing 11s each. That would give me instant information to investigate, whether the difference is an imbalance in the pack or a voltmeter has become inaccurate.

 

[john61ct]

FYI Victron BMV-712 is not only a pretty accurate SoC monitor (limited voltage range however)

but also includes a relay circuit that can be driven via many conditions

including series-midpoint imbalance detection.

Separately, anyone know of a source to buy the old Lee Hart style Batt-Bridge, or a cookbook on DIYing one, done properly?

 

[docware]

I use cheap 2 - 3 USD voltmeter on ebike for battery voltage/SOC. For regular checking of the paralel groups voltage I use single purpose instrument with precise voltmeter.



Precise voltmeter has resolution 0,1 mV up to 4,3 V. There is also built-in precision 0,02% voltage reference MAX6341. Checking with 5 1/2 DMM is giving difference 0,1 mV. On the picture is voltmeter switched to the reference.

Price of the precise voltmeter is about 5,6 USD, however you have to buy 4 pcs, so complete price is 22 - 24 USD.

 

[eMark]

Yes, I agree that cheap simple voltmeter is very efficient SOC gauge, especially for other chemistry like NCA or NMC. Just recommend to check the real values with an external DMM once or twice per year....

That gets done more like 2-4 times a year when I manually check balance, since I don't use Battery Murdering Systems.

It was a comment by docware on another thread and BMS disparaging comments like yours' (Murdering) that convinced me to use two balance leads for reading cell voltages with extra provision for balance charging (infrequently needed). The BattGo 8S is accurate enough for everyday use. Compared it against most expensive DMM (within 0.02mV). NASA DMM supposedly within +/- 0.02mV ... BG-8S within +/- 2.00mV

 

[john61ct]

I use cheap 2 - 3 USD voltmeter on ebike for battery voltage/SOC. For regular checking of the paralel groups voltage I use single purpose instrument with precise voltmeter.



Precise voltmeter has resolution 0,1 mV up to 4,3 V. There is also built-in precision 0,02% voltage reference MAX6341. Checking with 5 1/2 DMM is giving difference 0,1 mV. On the picture is voltmeter switched to the reference.

Price of the precise voltmeter is about 5,6 USD, however you have to buy 4 pcs, so complete price is 22 - 24 USD.

Wow, link please?

 

[docware]

https://rdtech.aliexpress.com/store/923042

Direct link doesn´t work. Go Products – DC Digital – DC voltmeter 5 Digit. Choose 4 wires variable precision voltmeter.
Need to have isolated battery power supply cca 8 – 12 V. Don´t recommend blue and yellow color, very unpleasant for eyes.

 

[John in CR]

Yes, I agree that cheap simple voltmeter is very efficient SOC gauge, especially for other chemistry like NCA or NMC. Just recommend to check the real values with an external DMM once or twice per year....

That gets done more like 2-4 times a year when I manually check balance, since I don't use Battery Murdering Systems.

It was a comment by docware on another thread and BMS disparaging comments like yours' (Murdering) that convinced me to use two balance leads for reading cell voltages with extra provision for balance charging (infrequently needed). The BattGo 8S is accurate enough for everyday use. Compared it against my expensive DMM (within 0.02mV). Worth every penny.
download.jpeg

I wish I could lay claim to coining the term BMS to be Battery Murdering System, but that wasn't me. The few packs I've had with BMS's were all killed by the very thing that was supposed to protect them, so the term is quite accurate. When you add hundreds of points of failure to a very simple system of course that's to be expected. OTOH Joe Blow general public has his face so glued to his cell phone that he can't possibly be expected to understand and monitor the operation of something so simple yet important as a battery pack, so I understand the need for BMS's...just not on my bikes.

 

[eMark]

Hear you loud and clear ... that's why it was a comment by docware on another thread that most influenced me to build my pack (without BMS) that can be split in two for balance charging (when needed being battery is 30Q) with duo 6S balance charger. Can also have two of those little cell-log monitors with LVA settings connected to the two balance leads when battery is in use or resting.

My understanding of all inexpensive Chinese BMSs is that they only balance when cell disparity is more than 30mV. Similar to those cheap RC Lipo Blinky discharge balancers (30mV).

Also made correction to above iDST BG-8S (+/- 2.0mV accuracy) compared to most accurate (NASA?) instrument (+/- .02mV) ??

 

[docware]

I want to emphasize that I don´t recommend building battery pack without BMS. I appreciate that BMS can help charger and control unit safety functions like HVC, LVC, overcharge and overdischarge protection. I am only saying that balancing function of BMS is debatable and controversial. And those, who are able to check parallel groups voltage by proper measuring, often find out, that their battery pack can live without any balancing for long time, if the load is not heavy.

I am sorry, but those various cell monitors are not only often inaccurate and therefore misleading, but also can contribute to real disbalance of the parallel groups.

 

[john61ct]

My understanding of all inexpensive Chinese BMSs is that they only balance when cell disparity is more than 30mV

I would not bother balancing a pack with a delta that small.

 

[john61ct]

Thanks a lot!

Although I was after the complete box with 6-cell selector switch

I guess you just meant the measurement / display component? Is there a build thread?

https://m.aliexpress.com/storesearch/list/4-wires-variable-precision-voltmeter.html?sellerAdminSeq=202659690&searchType=storeSearch&storeId=923042&sceneId=050000&keywords=4-wires-variable-precision-voltmeter

Those currently shown seem identical other than price.

How can one verify if the unit has the MAX6341 reference within? Or is that just implied by the accuracy spec?

Precise voltmeter has resolution 0,1 mV up to 4,3 V. There is also built-in precision 0,02% voltage reference MAX6341. Checking with 5 1/2 DMM is giving difference 0,1 mV. On the picture is voltmeter switched to the reference.

 

[docware]

My understanding of all inexpensive Chinese BMSs is that they only balance when cell disparity is more than 30mV

I would not bother balancing a pack with a delta that small.

Frankly, I consider delta above 5 mV to be out of normal. However, after slow battery charging nearly always all 10 groups are again back within 3 mV. Had no reason or opportunity for balancing for last 2 years, maybe more.

 

[eMark]

I am sorry, but those various cell monitors are not only often inaccurate and therefore misleading, but also can contribute to real disbalance of the parallel groups.

Tend to agree, but when it comes to the iDST BattGo 8S it's more than a monitor and more reliable (+/- 2mV) than a cheap cell monitor. The CellLog 8M was handy, but could be misleading ... one reason why it's no longer available. That's why it's important (like docware recommends) to occasionally compare the voltage readouts of an inexpensive cell monitor and likewise a BG-8S against reliable voltage readouts from a reliable DMM.

If the best i could buy was my inexpensive Tenergy 5-in-1 Intelligent 7S Meter/Balancer i'd make do using a DMM to verify. However, when comparing it (for all it can do) against my iDST BG-8S the Tenergy's IR readout and cell voltage readout isn't accurate. Compared to the BG-8S the Tenergy is junk.

There's no way to get an individual cell readout of each of the cell voltages within a P-group. Normally not necessary with Grade A cells when P-groups are within 5-10mV of one another. If you see one of the P-groups voltage with increasing disparity than the other P-groups you know it's time to balance charge. A disparity of 5-10mV between P-groups is of little concern. If the disparity is as much as 30mV difference between one of the P-groups compared to the other P-groups that are within 10mV of each other it's not only time to balance charge, but there could be more serious problems down the road.

 

[docware]

Thanks a lot!

I guess you just meant the measurement / display component? Is there a build thread?

https://m.aliexpress.com/storesearch/list/4-wires-variable-precision-voltmeter.html?sellerAdminSeq=202659690&searchType=storeSearch&storeId=923042&sceneId=050000&keywords=4-wires-variable-precision-voltmeter

Those currently shown seem identical other than price.

How can one verify if the unit has the MAX6341 reference within? Or is that just implied by the accuracy spec?

Yes, the unit is precise volmeter only. I decided to use voltage reference MAX6341 also in this single purpose instrument to have all in one.
There is no build thread.
The rotary switch is switching particular parallel group, sixth position is the reference voltage.

 

[docware]

Tend to agree, but when it comes to the iDST BattGo 8S it's more than a monitor and more reliable (+/- 2mV) than a cheap cell monitor.

Citation from iSTD BG-8S manual :
„Voltage measurement accuracy : ± 0,005 V @ 4,2 V“

You have to also take into consideration that iSTD unit is supplied from some measured cells – another factor affecting the voltage. Resulting inaccuracy may be 10 – 40 mV, or more, who knows. Until you check the real voltages (first without BG-8S, then with BG-8S connected) with external voltmeter having sufficient resolution 0,1 mV and sufficient input resistance, you have no idea what´s truly going on.