Passive balancing - a simulation

[Chriull]

Since here were some discussions about passive balancing as used in our wheels BMS and i found a nice paper (1) for li ion simulation. The parameters are for some lipo cell and so a bit different from our used li ion cells (as our 21700 are from our 18650...) - but for some nice graphs and thoughts more than good enough!

I simulated a pack consisting of 16 such cells with 3250mAh capacity in series. Discharge happens with 3.25A (1C) and charging with 1,625A (C/2). For the balancing resistors (if active) i used 70 Ohm. 

(I) So first a simple discharge followed by a charge:

All 16 cells have the same capacity and charge state - so in blue and in red the _same_ cell voltage for this procedure. After a short rest the  discharge with this 3.25A downto 3.3V, then a small rest again (Voltage jumps up to a bit above 3.6V) and then the charge with 1,625A CC and CV until the charge current drops below 50mA (manufacturer specification). Some rest again - cell voltage drops a bit below 4.2V again and then the same procedure...

(II) Next graph shows this pack with balancing deactivated and cell #1 ("aged cell") having only 98% of 3250mAh, starts out fully charged, cell #2-#16 ("normal cells") have the full 3250mAh and start out with 95% charge. (2)

Here the aged cell keeps the higher voltage throughout the discharge and charge cycle until the BMS cuts off at 4.25V. (For this simulation i should choose a lower starting voltage so the the cell voltage gets lower with dischargs and stays higher after charging.)

After cut off with no charge current both all cells drop below 4.19V and everything is "normal". This imbalance stays like this forever.

(III) Now the same scenario with passive balancing enabled:

Cell overvoltage is not reached already by the first balancing throughout the whole CV stage and after charging stopped cell #1 is discharged downto 4.19V. So the pack is successfull balanced again! After this The pack stays balanced normal.

But as one sees now - discharge stops as before at 3.3V*16 cells. But the weaker cell #1 drops further - downto about 3.2V while the 15 normal cells stay above 3.3V.

(IV) Now the same with even stronger imbalance. Cell #1 has now 94% of the capacity and cell #2-#15 start with 95% state of charge. Now it needs 5 charging cycles until at the 6th charging cycles the balancing can prevent overvoltage cut off and discharge after charging cell #1 to 4.19V again. After this the pack stays balanced again - just as one sees by the lower capacity of cell #1 the cell voltage drops downto ~2.65V! And this with the whole pack still stopping at 3.3V*16!

Will try to make some detail graphs of the balancing process and the end of discharge voltage drop spread of the cells.

As over the cycles cell voltage relation stays still very constant the only reason left for misbalance to happen is degradation while charging/discharging? Once voltage gets higher and too low...

(1) "An Accurate Electrical Battery Model Capable of Predicting Runtime and I–V Performance"  http://rincon-mora.gatech.edu/publicat/jrnls/tec05_batt_mdl.pdf

I used the formulas (1), (2) and (3) and "ignored" (4)-(7) as transient is of no interest for us and i assume this data will never be available for our li ion cells.

Should be possible to get (2) and (3) from the various discharge and charge graphs. Maybe with ovtave - never looked at this stuff...

(2) As in the paper from Texas Instrument about cell balancing such a scenario is described as what's destroying a li ion pack...

https://www.ti.com/download/trng/docs/seminar/Topic 2 - Battery Cell Balancing - What to Balance and How.pdf

Edited March 7, 2021 by Chriull

[Chriull]

My conclusion from this data and my readings so far:

Starting with the premise:

On 3/6/2021 at 2:36 PM, Chriull said:

left for misbalance to happen is degradation while charging/discharging? Once voltage gets higher and too low...

and the different graphs:

The passive balancing is _not_ intended to balance during charging! The balancing resistors are mainly to keep the weak cell from triggering the overvoltage protection.

In case of imbalance the pack gets balanced by the resistor discharging the weak cell to 4.19V.

Seems the degradation process of a pack is as follows:

-1- Healthy pack: starting out as balanced, nicely matched pack charging and discharging as shown in (I)

-2- first imbalances: by the "higher" and "lower" voltages one cell starts too degrade (loose capacity, ...) faster as the others. So this cell gets exposed to higher and lower voltages even more and degrades faster... ( See (2) from my first post)

As seen in scenario (III) the balancing resistor keeps the voltage of this weaker cell below single cell group overvoltage threshold and discharged this cell _after_ charging to 4.19V - pack is balanced again.

So trying to prolong the C(onstant)V(oltage) stage or even doing "endless" trickle charges is counterproductive, as as written above keeping the weak cell at the high voltages stresses and degrades the weakest link even more!

So a reasonable fast stop of the CV phase, around 50-60mA per paralleled cell as recommended by the manufacturer, keeps this stress within limits and the passive balancing reduces occured overvoltages _after_ the charging process to balanced values!

-3- First charge cut offs occur (IV): The balancing resistor is not capable anymore of keeping the voltage below shut off threshold. Seems this could "equalize" after some cycles.

In such a lower pack charge states should be avoided, as the weakest cell will be pushed to dangerously low voltages by this!

Time to slowly seek a source for a replacement pack/repair shop to replace the cell group!

As now charging can be logged (euc world + hs110, or some ks with eucworld/wheellog/?darknessbot? without additional accesoir) hopefully riders will look more on this details (premature charge stop!, balancing resistors active after charging - have no real idea how to detect this practicly..:( ) and data can be collected about the speed of degradation once these states are reached!

 

Edited March 24, 2021 by Chriull

[Chriull]

Btw: as from reports here the following steps:

-(4) the weak cell stays more or less in some 2-2.5V range. Charging stops before reaching 100% - people start reporting their problems here. From some reports here it seems that not only one weakest cell group can be affected but also two. Or such can occur from bad original cell matching?

-(5) the cell is "dead" and stays at 0V. Significant plating/internal short circuits should have occured greatly increasing risk of fire hazard. (A process already noticeable started at -(4) and a bit before)

-(6) The dead cell can be forced to change the polarity (by ?high burden? discharge ?and charging?). Seemingly the last stage before extinction.

 

[Chriull]

Just had an idea for some more graphs (yeahh ;)) - constant degradation over the charge cycles for cell #1. I choose for now 0.5% capacity loss at the beginning of each discharge and charge cycle - which seems already quite much! Here the first twenty cycles - in total 20*2*0.5%=20% capacity loss:

Here the "top part" magnified.

The bottom part - the low voltage the weak cell is forced to once one discharges the pack to 3.3V * 16 cells = 52.8V.

Here already cell #1 gets fully discharged.

[Chriull]

Maybe a nice detail or some artefact happened by some pecularities of the simulation or the starting conditions:

already from the first charge cycle on(0.5% + 0.5% degradation happened) the balancing resistors get active for the aged and all the normal cells! As the aged cell seems to get above 4.2V a bit later so the other cells reach the threshold and balancing gets activated form them. After this cell #1 overtakes the other cells and balancing gets activated for the aged cell, too - but it stays with faster raising voltage...

So in the end _all_ cells get discharged to 4.19V and by this "perfectly" balanced.

 

 

[Zopper]

If you have the time, could you try what happens when charging to e.g. 80%?

[RagingGrandpa]

Nice visualization!

It emphasizes that the interesting part of balancing the 'overvoltage' cells happens after the charger is disconnected (e.g. @ 250min in the graph below). And therefore, no benefit of stressing the pack with long-duration CV.

But it would be also interesting to consider healthy, 'undervoltage' cells present at the same time, perhaps due to storage and unequal self-discharge of otherwise healthy 'new' cells. I think the interesting part of balancing the 'undervoltage' cells can be while the charger is still doing CV.
Best to consider the situation when both conditions are present, in a healthy, slightly-imbalanced new pack: e.g. one cell 3.9V, one 4.1V, all others 4.0.
 

On 3/6/2021 at 8:36 AM, Chriull said:

 

[Chriull]

On 3/12/2021 at 6:57 PM, Zopper said:

If you have the time, could you try what happens when charging to e.g. 80%?

Quite the same - just a bit earlier and stronger "bouncing" of the degrading cell.

As this is a simulation of _another_ lipo with a bit different characteristics and my trial of accelerated degradation emulation is just some personal assumption there is nothing to say/learn or conclude from comparing these graphs - at least in regard of 80% charging (without balancing) is or is not recommendable for our wheels! For this the understanding of the degradation process is missing from my side. And a model of li ion cells used in our wheels.

This simulation is just showing (hopefully) by which different mechanisms balancing is achieved.

[Chriull]

On 3/12/2021 at 7:53 PM, RagingGrandpa said:

Nice visualization!

Thanks!

On 3/12/2021 at 7:53 PM, RagingGrandpa said:

It emphasizes that the interesting part of balancing the 'overvoltage' cells happens after the charger is disconnected (e.g. @ 250min in the graph below). And therefore, no benefit of stressing the pack with long-duration CV.

That's what i learned from this trial, too. If one analysis and rethinks consequences that arise from the voltage comparators you listed in your bms teardown topic this is imho logical?!

But i had thought that more of the balancing to happen before while charging...

On 3/12/2021 at 7:53 PM, RagingGrandpa said:

But it would be also interesting to consider healthy, 'undervoltage' cells present at the same time, perhaps due to storage and unequal self-discharge of otherwise healthy 'new' cells. I think the interesting part of balancing the 'undervoltage' cells can be while the charger is still doing CV.
Best to consider the situation when both conditions are present, in a healthy, slightly-imbalanced new pack: e.g. one cell 3.9V, one 4.1V, all others 4.0.

This needs some small changes to the simulation - will try this lateron...

[Chriull]

On 3/12/2021 at 7:53 PM, RagingGrandpa said:

But it would be also interesting to consider healthy, 'undervoltage' cells present at the same time, perhaps due to storage and unequal self-discharge of otherwise healthy 'new' cells. I think the interesting part of balancing the 'undervoltage' cells can be while the charger is still doing CV.
Best to consider the situation when both conditions are present, in a healthy, slightly-imbalanced new pack: e.g. one cell 3.9V, one 4.1V, all others 4.0.

This following graph shows three identical cell groups - cell group #1, one cell starting out with 4.06V, cell group #2, one cell with 4.15V and cell group #3, 14 cells with 4.1V. 

 

With this values the lowest cell does not become fully discharged.

The real number of cycles until balance is reached depends of course on the real used cells in combination with the balancing resistors value and currents. Here i used 0.5C for charging, 1C for discharge and the 850mAh polymer Li Ion battery from the linkes paper "misused" as 3250mAh Li ion in a 16s1p setup.

Edit: Added SoC (State of charge) to the graph as the dropping voltage at the change from CC to CV stage looks a bit weired...

Edited March 16, 2021 by Chriull

[Chriull]

On 3/12/2021 at 6:57 PM, Zopper said:

If you have the time, could you try what happens when charging to e.g. 80%?

Here the same three cell groups as before - and this time with just CC stage and no CV. So the ~80% charging. Here the cells stay more or less with the differences as they started. Just cell #2 (with the highest starting voltage) has some short time the balancing resistor enabled as it reaches the 4.2V threshold. But as this happens within the CC phase and the "internal voltage" of the cell is below 4.19V the balancing resistor is disabled immedeately after charging is stopped.

As by writing my above comment about the "internal voltage(~state of charge)" beeing to low once CC stage ends i made another simulation with 0.1C charging instead of 0.5C - so the "internal voltage" has a chance to get higher:

So here the balancing stays active once for cell #1.

Edited March 16, 2021 by Chriull

[RagingGrandpa]

Right, so:

• undervoltage cells benefit from leaving the charger connected longer (extend CV phase)
• overvoltage cells benefit from letting the pack rest after disconnecting the charger

Perhaps it means the way to ensure complete balancing without knowledge of the cell voltages is something like:
"continue recharging for 3 hours after the charger light turns green, then remove the charger and let the battery rest for an additional 3 hours"

And after the pack is balanced, the long-duration CV and rest is no longer necessary.

If charging interruption happens (BMS charge-stop, as shown in your first charging cycle below), let the pack rest a few hours then try charging again. This can only resolve small imbalances like 100mV, but could help.

On 3/15/2021 at 11:23 AM, Chriull said:

cell #1 with 4.06V,
cell #2 with 4.15V,
other 14 cells with 4.1V

Edited March 22, 2021 by RagingGrandpa

[Chriull]

1 hour ago, RagingGrandpa said:

• undervoltage cells benefit from leaving the charger connected longer (extend CV phase)
• overvoltage cells benefit from letting the pack rest after disconnecting the charger

I tend to the hypothesis, that letting the pack rest after disconnecting the charger suffices for both conditions and stresses the cells less with voltages above 4.2V.

As for your first point undervoltage cells have to have overvoltage cells that will be bleeded after charging nearing both groups, too.

Maybe there exist cases were by prolonged CV stages the overall cell stress could be minimized? As i (we?) don't know too much more details about these processes i tend (for know) to my hypotheses - cv stage as specified (or at longest until charger led goes green) and resting time after charging! It simplifies things a should imho not bear too much risk of (noticable) worse cell treatments?

[RagingGrandpa]

Right, so there's another question at hand:

• What is the harm of charging longer than the manufacturer-recommended 100mA/cell threshold?
  ("Float charging")

From literature, I understand the harm to be "accelerated aging," but not catastrophic failure.

If anyone has studied this in real-world practice, please chime in.

[Tawpie]

2 hours ago, RagingGrandpa said:

real-world practice, please chime in

As one who has ruined the batteries in my laptop and a phone by leaving them plugged in 'float charging' 99.99% of the time, I can attest that it's definitely not the way to prolong a battery's useful life. BUT. The laptop has spent 9 years on charger and the phone 4 years—pretty much 24x7, I have no idea when they started to go south. And after that kind of punishment, the laptop lasts about an hour on batteries (it was 4-5 hours when it was new) and the phone... i don't know how long it lasts but not terribly long! Neither has self ignited or swollen up. Yet.

Edited March 22, 2021 by Tawpie
wordsmithing

[Chriull]

2 hours ago, RagingGrandpa said:

From literature, I understand the harm to be "accelerated aging," but not catastrophic failure.

Seems so - according to https://batteryuniversity.com/learn/article/charging_lithium_ion_batteries staying at lower voltages (as 4.2V) increases lifetime and charging to or higher as 4.3V makes it dangerous.

The rest are clear recommendation, but unspecific: "Li-ion cannot absorb overcharge. When fully charged, the charge current must be cut off. A continuous trickle charge would cause plating of metallic lithium and compromise safety. To minimize stress, keep the lithium-ion battery at the peak cut-off as short as possible."

Papers i found till now inbetween that should reveal more details on li ion stress mechanism are scientific papers to be paid dearly... Imho not worth for the possible knowledge gain for us amateur end users...

PS: A nice statement to keep in mind from this site:"A fully charged battery has a lower thermal runaway temperature and will vent sooner than one that is partially charged."

[RagingGrandpa]

I thought this summary was lovely:

On 5/14/2021 at 7:38 AM, redsnapper said:

https://link.springer.com/article/10.1007/s41918-019-00060-4

... reinforcing the idea that there is significant headroom above 4.20V before damage. Charging interruption at 4.25V/cell sounds quite reasonable; and a charger mis-adjusted 2V higher than intended is not an immediate catastrophe.

 

Edited December 13, 2023 by RagingGrandpa
(img url)

[OneLeg]

I propose you an idea which I am soon going to test. Basically that we add an extra cell to each series but charge it as if it was not there.

On a 20S battery pack we install a 21S BMS and 21 cells in series and we still charge it at 84V.  This will cause each cell to receive 4V instead of 4.2V. The charge will go as normal going from CC to CV.  The CC and CV work at any voltage on variable voltage Buck charger, I use a Drok DC-DC as I charge from Solar.

When LiOn batteries age, Whr capacity drops quickly when load is applied and the cells are on the upper part of the voltage han the voltage and the discharge curve tends to shift downwards, so trying to charge them to 4.2V ages them further, charging them at a maximum of 4V will increase their life time. They will also likely to suffer from less voltage sag.   The other added benefit is at when the cell capacity is used up and voltage is lower.... Often older batteries have lots of capacity left at the 3Vto 3.3V, howveer the unicycle unicycles don't go down to manufacturers of  2.8V for safety reasons and will shut down for safety. As we have shifted the whole scaled down by 0.2V means you will be able to tap into capacity which normally the unicycle would not allow you to use.

Overall I think this would provide more cycles out of a battery, it would also as the battery ages, it would provide more capacity as the battery would discharge lower and there is more capacity as the battery ages.

The one issue is balancing, you would need a BMS where you can trigger the balancing at user defined voltage or manually.

The question is why not even stick 22 cells?

I have a 16S BMS lined up for my Ninebot A1 which used 15S normally.... I am going to use Sanyo 20700B

 

[Chriull]

2 hours ago, OneLeg said:

Overall I think this would provide more cycles out of a battery, it would also as the battery ages, it would provide more capacity as the battery would discharge lower and there is more capacity as the battery ages.

This should be theoreticly true - especially for single cell systems!

For our 16,20 and 24s systems it seems that the missing balancing by not charging to full voltage leads to much lower overall battery lifetime. Was beside many other locations also discussed here:

 

[RagingGrandpa]

You're creating a problem (defeating top-balancing) to get some very minor benefits...

1. Avoids stress from high-SOC operation.
   But cell aging isn't a major problem with EUC's, where damage to the rest of the <body, pedals, rim, controller> typically ends its life long before 500 battery cycles elapse.
    
2. Tricks the controller into discharging to 2.8V/cell (16s 3.9V->2.8V).
   There is no capacity benefit from doing this, when compared to the alternative (15s 4.2V->3.0V).
   And there is a vehicle-level risk in deeper discharge: pack voltage will sag immensely if a pulse of current is required. This can lead to poor balancing control and crashes.

.02

[OneLeg]

  

16 hours ago, Chriull said:

This should be theoreticly true - especially for single cell systems!

For our 16,20 and 24s systems it seems that the missing balancing by not charging to full voltage leads to much lower overall battery lifetime. Was beside many other locations also discussed here:

 

 

Well how is it that Inmotion systems get serious life time out of their packs when they don't actually balance the cells?

electrons in electrons out means = wear

So if you want to maximize the life of a battery use it less.... Using the voltage of the lowest cell as the "current capacity" will ensure that the that pack does not get damaged, and at the other end of the spectrum you need to do the same when getting close to 4.2V.

Here what we are discussing actually are charging strategies, do we charge for higher capacity? do we charge for higher number of cycles ?

In fact I am starting to question whether when we charge we should be even be using CV to charge to maximize number of cycles.  There is nothing from stopping us from soldering balance leads to the pack and once a month you plug in an active balancer and get all your cells aligned, this will maximize the capacity for a STOP on Max/Low strategy.

I will post photos later... of a V10 battery pack...

[OneLeg]

14 hours ago, RagingGrandpa said:

You're creating a problem (defeating top-balancing) to get some very minor benefits...

1. Avoids stress from high-SOC operation.
   But cell aging isn't a major problem with EUC's, where damage to the rest of the <body, pedals, rim, controller> typically ends its life long before 500 battery cycles elapse.
    
2. Tricks the controller into discharging to 2.8V/cell (16s 3.9V->2.8V).
   There is no capacity benefit from doing this, when compared to the alternative (15s 4.2V->3.0V).
   And there is a vehicle-level risk in deeper discharge: pack voltage will sag immensely if a pulse of current is required. This can lead to poor balancing control and crashes.

.02

1. Cell aging is a problem in unicycles... On a 20A battery system my experience I get about 10Km per 1Whr of capacity after. All my unicycles have 20A batteries, which have got shorter life spans than the standard 10A... theoretical - 300cycles vs 500cycles. 

2. There is a benefit of Mha capacity as the battery ages,  as the whole voltage range shifts downwards... If you start 4.2 -> 3V, but your battery tanks 4V in the first 10 seconds due to age, that is a lot of Mha wasted... the Mha available between 3V-2.8V is much greater than the Mha available from 4V - > 4.2V

Look at a discharge curve....

I might add after I came up with this idea, one of the local electronics guru's here in HK told me that he had seen this strategy on Hoverboards, which use very shit quality batteries to increase performanc/life time.

Edited May 29, 2021 by OneLeg

[Chriull]

5 hours ago, OneLeg said:

Well how is it that Inmotion systems get serious life time out of their packs when they don't actually balance the cells?

Interesting - you have more details on this?

5 hours ago, OneLeg said:

In fact I am starting to question whether when we charge we should be even be using CV to charge to maximize number of cycles.

If one wants to charge only to 4V there is no need for an cv stage. Just charge with CC to some 4.X-4.1XV and the voltage wil settle to 4.0V.

Afaik CV stage is just needed to reach 4.2V "saturated" voltage without applying a voltage higher than 4.2V during charging.

5 hours ago, OneLeg said:

  There is nothing from stopping us from soldering balance leads to the pack and once a month you plug in an active balancer and get all your cells aligned, this will maximize the capacity for a STOP on Max/Low strategy.

Applying balancing leads is the best idea - as many people already made! It's more or less the only way to immedeately and trupy know the state of the battery cells!

5 hours ago, OneLeg said:

the Mha available between 3V-2.8V is much greater than the Mha available from 4V - > 4.2V

Look at a discharge curve....

The discharge curves are not helpfull for such a consideration, as they show the battery voltage under load which does not correspond in any way to the state of charge.

 

[Roadpower]

Okay so to my brain this is a lot to digest currently but in reading some of the back and forth I came up with an introductory question or two.

"resting" is a common concept in various areas as it is useful for getting a desired product or result, so my question from there; should riders in general seek to allow a rest period after charging before using their wheel?  (In my case this is a default practice anyway.)

Second question is; when a green light activates on the charger does that indicate that all charging as ceased because maximum voltage was achieved OR does that mean that the charger has entered a trickle charge state?

EDIT: In thinking about the second question for a moment I realized that this might only be knowable by putting inline something like a Kill-a-watt meter before the charger.

Edited May 29, 2021 by Roadpower

[Chriull]

10 hours ago, Roadpower said:

Okay so to my brain this is a lot to digest currently but in reading some of the back and forth I came up with an introductory question or two.

"resting" is a common concept in various areas as it is useful for getting a desired product or result, so my question from there; should riders in general seek to allow a rest period after charging before using their wheel?  (In my case this is a default practice anyway.)

Resting after charging begore riding and after riding before charging is to protect cells from high temperatures and hence degradation.

Although most charging with EUC standard stock chargers is very low battery burden and creates very low to neglible temperature rise.

The resting inmentioned above is to "deduce" charge % from battery voltage is needed because somehow "charge" seems not commonly distributed within the cells - in the papers/articles often surface charge/anode charge/etc are mentioned which are caused by different chemical processes during (dis)charging. These are the papers where i just look for some summary/conclusion written so i can get something from the content...

Maybe the system is easier to understand by looking at an equivalent circuit diagram consisting of two capacitors - one really big one having a "constant" voltage always directly corresponding to the state of charge of the cell. The second capacitor is smaller which can take/provide some extra "power" - so once this extra "power" is used up the cell has less voltage than it should have according to it's state of charge. Once in low burden situation this smaller capacitor is charged up again from the main capacitor the voltage is corresponding to the charge % again.

And the same vice versa for charging.

Quote

Second question is; when a green light activates on the charger does that indicate that all charging as ceased because maximum voltage was achieved

No maximum voltage is already (about) achieved at the end of the C(onstant)C(urrent) stage and beginning of the C(onstant)V(oltage) phase.

That's about some 80 to 9x% percent of charging done, depending on charge current and battery (health/age) state.

Quote

OR does that mean that the charger has entered a trickle charge state?

Yes! The chargers dedect once the current dropped below some threshold and this lights the green LED.

Charging current is the indication to stop li ion cell charging. For the mostly used cells for EUCs manufacturer datasheets say to stop at ~50-60mA (per paralleled cell).

Some li ion web sites state as rule of thumb some 3% of the capacity as current threshold (getting rid of the time of the unit somehow). So ~100mA for a 3500mAh cell.

It's afaik "unspecified" at which currents which charger shows the green led - afair it's often somewhere around 100mA (and can be adjusted).

Quote

EDIT: In thinking about the second question for a moment I realized that this might only be knowable by putting inline something like a Kill-a-watt meter before the charger.

This also exist for "more sophisticated" charger/bms systems. They use a "coloumb counter" (summing up energy going in and out) like with imho ninebots. They "always" show and know the state of charge of the cells. And often need some charge cycles to calibrate themselves.

But the charge current for a known battery pack is a more than sufficient condition to determine if charging is finished and "only trickle/float charging is going on".

Edited May 30, 2021 by Chriull