BMS cell-balancing function creates cell imbalance

[mrbill]

In early 2014 I purchased a number of Lithium batteries from EM3EV, built from the Samsung INR-18650-29E cells. Four batteries are configured as 7s9p, and two are configured as 7s12p. EM3EV supplied at my request a "2-wire" (same charge and discharge circuit) BMS for each battery. The BMS is sealed inside the battery, so it cannot be examined. Each battery includes an 8-pin JST-XH connector that can be attached to a Cellog 8S to record cell data or to allow manual cell charging/discharging, as needed. Overall, I have been happy with these batteries.

Every year in January I run a capacity test on each battery to determine what my usable capacity is for the coming year.

Each battery is charged until the first cell reaches 4.2volts, when the BMS cuts off charge current to the entire battery. The charger holds the top voltage for at least another 12 hours to assure that cell balancing is completed before I run the discharge test. For the discharge test I connect a Cellog 8s to record cell voltages while I discharge each battery at 12-18 Amps through a calibrated Cycle Analyst into a resistive load.

After the test is complete, I record the year's test results and add them to my blog.

https://mrbill.homeip.net/hybridBike.php#batteryCapacityTesting

This year I noticed that on all but two of the batteries Cell 7, the most positive cell, was consistently high during the entire discharge. On a couple of batteries Cell 6 and Cell 1 were also high. It appeared that the BMS was unable to balance all of the cells. I was able to observe the BMS balancing cells at the top of charge on the Cellog 8s display, and the BMS otherwise appears to function normally in the critical matter of cutting off charge current when the first cell hits 4.2 volts and discharge current when the first cell drops to 2.75 volts.

On the batteries that exhibited cell imbalance I manually discharged the cells with high voltage until those high cells showed a voltage less than the average cell voltage. Then I resumed the overall charge until the BMS again shut off charge current as the first cell hit 4.2 volts.

On all batteries that I manually balanced in this manner I was able to increase the total capacity. On one battery I was able to increase the capacity by 10% after manually balancing the cell voltage.

I was disappointed to observe with one battery that I had manually balanced and then charged normally, leaving the battery sitting at its top voltage for 24 hours, that Cell 7 on that battery was again resting at a higher voltage than the other cells.

Questions:

1) Why is Cell 7 consistently charged/balanced to a higher voltage than the other cells?

2) How many other customer's batteries may be exhibiting loss of capacity not from changes in chemistry but from creeping cell imbalance caused by insufficient or a faulty balancing function in the BMS?

 

[dogman dan]

I'm famously a dummy,, but I don't understand why you assume the cells are perfect after some time, and the problem must be the bms.

I read shit too early, maybe I just misunderstand. Cell 7 consistently fills first? Then on discharge, is sags the most? After three reads,, I'm just getting more confused what exactly you say cell 7 is doing.

My confused conclusion,, cell 7 has lost some capacity. It has lost some watt hours, so now it always reaches full first, because full for it is less wh than it takes to fill the others. It's IR is higher than the others, so it sags more too.

In the RC lipo world,, the positive cell puffs and fails first, very frequently. I'm not smart enough to understand exactly why but the cell on the positive end of the string is the one that takes a beating.

Or,,, I have it backwards. Cell 7 is the good one? it holds its charge at what the bms discharges it to. The others are self discharging to below what the bms does, after you pull it off the charger. Older cells all do this,, they don't hold 4.2v forever.

 

[mrbill]

Hi dogman:

Thanks for your reply.

My confused conclusion,, cell 7 has lost some capacity. It has lost some watt hours, so now it always reaches full first, because full for it is less wh than it takes to fill the others. It's IR is higher than the others, so it sags more too.

In the RC lipo world,, the positive cell puffs and fails first, very frequently. I'm not smart enough to understand exactly why but the cell on the positive end of the string is the one that takes a beating.

Or,,, I have it backwards. Cell 7 is the good one? it holds its charge at what the bms discharges it to. The others are self discharging to below what the bms does, after you pull it off the charger. Older cells all do this,, they don't hold 4.2v forever.

I find it strange that with all six of my batteries it is always Cell 7, the most positive cell, that shows the highest voltage after I charge/discharge the battery a number of times. Perhaps someone with more expertise than you or I could explain why this occurs. Since the same current flows through all cells, and the BMS (presumably) treats each cell identically (within tolerances) with respect to balancing, why should the most positive cell "take a beating" compared to the others?

In my case Cell 7 appears to have the same capacity as the others. This can be seen from the discharge graph. The graph below shows one of my batteries after a year of use, approximately 70 charge/discharge cycles, about 10 full-charge cycles, including the charge immediately prior to the discharge cycle shown in the graph, with a BMS that claims to balance all cell voltages at the top of charge.

If Cell 7 (dark blue line) had less capacity than the other cells its voltage would start high and be the first to reach cutoff at full discharge. Its discharge curve would cross the other discharge curves or exhibit a steeper downward slope.

After I manually bled Cells, 7, 6, and 1 (the three discharge curves that sit highest at the end of charge), I was able to increase the battery's usable capacity by about 10%. I verified this by recording the discharge a second time, shown in the graph below.

The previously-high cells are still high, but now track the other cells' voltages much more closely.

I believe the BMS may not be balancing all cells properly, at least not balancing Cell 7. Is this a common problem with most BMS's found in e-bike batteries?

The chart below shows a battery that I had manually balanced so that all cells upon discharge tracked each other tightly. I then left the battery on the charger for 24 hours to see if the BMS was causing any cell imbalance. What I discovered is that the BMS had allowed Cell 7 to drift to a higher voltage than the others.

 

[Joachim]

maby the cc/cv charge profile is the problem try charge the pack slow/very low current and see if the other cells gets closer to nr7
most of the simpel bms you see in e-bikes doenst have a propper ballans function,just hvc and lvc for each serie and total pack voltage.

 

[mrbill]

maby the cc/cv charge profile is the problem try charge the pack slow/very low current and see if the other cells gets closer to nr7
most of the simpel bms you see in e-bikes doenst have a propper ballans function,just hvc and lvc for each serie and total pack voltage.

Hi Joachim:

Thanks for the suggestion.

I have already confirmed that the BMS is doing some sort of cell balancing as I have observed cell voltages vary individually on the Cellog 8s display when the battery is idle at any state of charge or when a cell's voltage exceeds 4.15 volts while being charged.

When I left Battery S6 on the charger for 24 hours, I programmed the Satiator to deliver a 1-Amp charge for the very reason you mention, to approach the cutoff voltage slowly, in case the BMS needed more time to balance the cell voltages. Although this had the effect of delaying the moment of charge cutoff when the first cell reached 4.2 volts, the end result was the same. After some time Cell 7 would always be left at a higher voltage than the other cells.

The BMS appears to latch, to cut off all charge current after the first cell reaches 4.2 volts and does not allow further charge, even as the cells "settle down" to a lower voltage. I must place a brief load on the battery before the BMS will again allow further charging.

 

[fechter]

There is something I call the "end cell effect" caused by the voltage drop in the wiring between the cells and the BMS. When charging, the wires going to the end cells carry the full charging current. The balance wires on the between cells are only carrying the balance current, which is very small in comparison.

You could wire the pack charging wires directly to the end cells to minimize the end cell effect, but you need the charge current to pass through the BMS also. Redundant wiring or extra heavy wire on the end cells will help.

The current that powers the BMS may not be drawn from the pack evenly and over time can create imbalance.

Most BMS balance circuits kick in at 4.2V. To really balance a pack, your charger needs to be set a little higher than 4.2v/cell in order to make sure all of them reach the balance voltage. You wouldn't want it this high all the time, but an occasional balance charge won't hurt. Even better to charge at a little higher than normal (4.22v/cell) with a constant current charger that is set lower than the shunt current. This will prevent any cells from hitting the HVC and killing the charge current.

 

[999zip999]

So what is your battery packs capacity and what is your battery packs rated capacity ?

 

[mrbill]

There is something I call the "end cell effect" caused by the voltage drop in the wiring between the cells and the BMS. When charging, the wires going to the end cells carry the full charging current. The balance wires on the between cells are only carrying the balance current, which is very small in comparison.

You could wire the pack charging wires directly to the end cells to minimize the end cell effect, but you need the charge current to pass through the BMS also. Redundant wiring or extra heavy wire on the end cells will help.

Hi fetcher:

Thanks for your input.

These packs, (4) 7s9p and (2) 7s12p, were assembled by http://EM3EV.com and are sealed with heavy heat shrink. The smaller packs currently offer a bit less than 600 watt-hours, the large packs a bit under 800 watt-hours, or about 2.6Ah per cell. The cells are rated at 2.9Ah, but they only delivered this much in the first year of service. After the first year capacity was reduced by about 10%, but it has lost nothing since then.

See: https://mrbill.homeip.net/include/batteryCapacityTests.pdf

The problem has not risen to a severity that demands that I perform surgery on my batteries. Except for my observing the cell voltage imbalance on the most positive cell (and sometimes on other cells) I am pleased with the performance of these packs.

My goal is to understand why this occurs and if there might be a way to prevent cell imbalance without my periodically manually draining or charging individual cells through the sense leads.

The current that powers the BMS may not be drawn from the pack evenly and over time can create imbalance.

I hadn't considered this possibility. Even so, it seems to me that the BMS should be designed so that its balancing circuit compensates for the additional overhead current draw of the BMS.

Most BMS balance circuits kick in at 4.2V. To really balance a pack, your charger needs to be set a little higher than 4.2v/cell in order to make sure all of them reach the balance voltage. You wouldn't want it this high all the time, but an occasional balance charge won't hurt. Even better to charge at a little higher than normal (4.22v/cell) with a constant current charger that is set lower than the shunt current. This will prevent any cells from hitting the HVC and killing the charge current.

I can try this, although I suspect it will have no effect. The BMS cuts off all charge current when the first cell reaches 4.2 volts. At that point the charge function is effectively disabled until a load is placed on the battery. (Note, that my batteries' "2-wire" BMS charges and discharges through the same circuit.)

I store the batteries at half-charge, and unless I know I need a battery's full capacity I charge them shortly before use to 4.1v/cell, per EM3EV's recommendation to extend cycle life.

I am able to squeeze additional charge into a battery by charging it until the BMS cuts off current, then discharging it into a dummy load for a second, then connecting the charger again. I observed once I was able to get an additional 0.24 Ah into the battery this way. If the BMS was balancing I don't know how much of this additional charge went into the cells and how much was lost in the shunt resistors.

 

[fechter]

Usually the shunts will turn on at a little lower voltage than the HVC, which will cut off the charge current. If you feed CC about 40mA, the shunts should prevent it from reaching the HVC level.

Your imbalance doesn't look that severe, so you might just stick with the stock configuration. To fix the end cell effect, you would need to get access to the end cells and the BMS. Just beefing up the wires would help a lot.

 

[mrbill]

Usually the shunts will turn on at a little lower voltage than the HVC, which will cut off the charge current. If you feed CC about 40mA, the shunts should prevent it from reaching the HVC level.

I'll try this at some point in the future. The Satiator is giving pulsing output over the voltage range (22-29v). I tried setting the current to 0.5 Amps, but it bounced all over the place (according to the Satiator display). 0.04 Amps would probably be worse. Maybe I need to try a linear charger if I can adjust its current low enough.

Your imbalance doesn't look that severe, so you might just stick with the stock configuration. To fix the end cell effect, you would need to get access to the end cells and the BMS. Just beefing up the wires would help a lot.

My plan is to balance manually annually or if I notice capacity dropping during the season. For now I'll keep the battery in its original seal.

Thanks for your advice.

 

[999zip999]

3 years old 29E, how many cycles ?

 

[fechter]

If you put a resistor in series with the charger you might get it to behave at 40ma. I just use my bench supply for stuff like that. It seems they are not that far out of balance but nice to optimize their potential.

 

[dogman dan]

So if this dumb as a dog guy understands right,, the cell 7 just gets charged a bit higher than the others routinely. This is as I understand it,, why the first positive cell takes a beating, more strain on it for reasons I don't fully understand,, both charging and discharging.

3 year old cells too. why would you expect them to have perfect bottom balance by now? this is what we are looking at right? I see close to perfect balance at the top. That's because the bms did its job. and a very mild difference at the bottom.

I don't see a bms problem here. I only see normal differences in IR on an older pack. In fact.. less than the normal. I'd expect by now your cell 7 would be beat enough to have much lower capacity than the rest, and despite charging higher, have a lower voltage through the whole discharge curve.

At 3 years old,, your pack is as close to perfect as you could possibly hope for. IMO.

 

[mrbill]

Hi dogman and 999zip999:

My batteries see relatively infrequent use. I have no more than a couple hundred cycles on any of my packs. They get used on average twice per week, discharged deeply at times, but always charged/discharged within the cell specs.

One aspect of my usage that I did not mention earlier is that since I switched to running DD hub motors I run the batteries in series or series/parallel, wiring 2 in series to function effectively as a 14s battery. (This configuration has the blessing of EM3EV.) I also charge them in this configuration, I'm thinking that when charged fully, only one of the batteries in the series string reaches full-charge cutoff, and unless I swap the two batteries occasionally, only one battery will be subject to long-term "balancing" by the BMS that exhibits a charge bias for the most-positive cell, leading to marked imbalance in that battery.

I know the batteries are in pretty good shape, so I have no plans to replace them. I'm hoping to get at least five years of use, perhaps longer. My last set of batteries (LiFe) gave me 7 years before I felt that upgrading was worthwhile.

Over that 7 years I discovered that battery energy density had just about doubled for same cost. I'm a little disappointed to see that after 3 years the state of the art cells commonly available to home builders have only about 20% greater capacity than my Samsung 29E's. Perhaps in a few years we'll see 18650 cells in the 5 Ah range.

fetcher, at some point I will try the series resistor in the charge circuit method, perhaps with one of my linear chargers. Thanks for the suggestion.

 

[fechter]

Charging the two packs in series will limit you to the capacity of the weakest pack. If possible, charge them separately.

A 40mA balancing charge is really slow. You can do the math on it, but that's about a day to get 1Ahr. If the cells are already close to full, then they might only need 1-2Ahr to get balanced. The shunts should prevent any cell from going over the HVC limit and cutting off the charge.

 

[dogman dan]

Well, that explains why your cells are still so good after 3 years. Especially if you do not store them full.

What confused me was cell 7 being higher charged at the end of the discharge. You too I guess. As long as they balance at the top good enough,, all is well in my world.

 

[mrbill]

Hi Folks:

I just wanted to get back to everyone following this thread. Thanks everyone for your ideas.

I completed a 40mA "cell balancing" end charge on one of my packs.

The graph below shows the cell voltages over 2 days of applying 40mA charge current:

The second graph shows detail for the last hour or so of charge.

Observations:

1) The BMS is not constantly balancing the cells. A flurry of activity occurs initially, then in a couple more bursts, the longest being the final 24 hours of charge. I have no explanation for this behavior.

2) The pack is in pretty good balance at the start of this "cell balancing" session.

3) Although Cell 7 tends to be high, it is not always the highest voltage cell.

4) The BMS seems to think that getting all cells' (except Cell 1) resting voltages to within about 30mV of each other is sufficient.

5) Cell 1 voltage is consistently lower than the others.

Further investigation revealed that the Cellog 8s may have been pulling Cell 1 down. I suspect the Cellog 8s is powered off Cell 1. Maybe someone familiar with these measurement devices can confirm this.

When I removed the Cellog 8s and continued with the 40mA charge then briefly sampled the cell voltages with the Cellog 8s less than 20 minutes later, Cell 1's voltage was within the range of the other cells.

My "take aways" from all this are:

1) Charge batteries singly or in parallel to prevent battery voltage imbalances on batteries discharged in series.

2) Perform a cell balancing charge periodically by charging to full voltage at low current. A 1 Amp charge overnight or for a couple of days at most is probably sufficient to to achieve cell balance within the specs of the BMS.

3) If a gross cell voltage imbalance is discovered, discharge the high cell(s) manually to bring cells into balance and to recover lost capacity.

4) The batteries' BMS cell balancing function appears to be functioning when fed with a constant charge current. I suspect that the Satiator's pulsatile charge current into a 7s battery may "confuse" the BMS's cell balancing function.

5) Perform the final balancing charge without connecting the Cellog 8s.

 

[fechter]

Wow, those voltages were really bouncing around. I suspect the charge current was not really steady. Some bounciness is expected as the shunts are digital, either on or off.

The CellLog 8 is supposed to draw more or less evenly from all the cells, but your measurement indicates it does not. For sure the CellLog draws quite a bit of power and you wouldn't want to leave it connected for long periods of time. I also don't really trust the accuracy of the CellLog. Good to double check with a real meter to verify accuracy.

 

[Alan B]

Celllogs are known to draw unequal current from cells, I recall folks doing mods to make currents more equal. Leaving one connected with such a low charging current would provide unequal voltages and currents into the cells.

 

[mrbill]

Wow, those voltages were really bouncing around. I suspect the charge current was not really steady. Some bounciness is expected as the shunts are digital, either on or off.

The CellLog 8 is supposed to draw more or less evenly from all the cells, but your measurement indicates it does not. For sure the CellLog draws quite a bit of power and you wouldn't want to leave it connected for long periods of time. I also don't really trust the accuracy of the CellLog. Good to double check with a real meter to verify accuracy.

The charge current was steady between 41 and 46 mA, according to my DVM (2% claimed accuracy), varying gradually over time but never jumping around. I also calibrated the Cellog station voltages according to my DVM (0.7% claimed accuracy).

Junsi claims their Cellog draws 8mA and is "100% calibrated before it enters to the market".

It makes some sense that current to operate the Cellog is drawn from Cell 1. The instructions always have Pin 1 (full negative) connected no matter how many cells are being examined.

 

[fechter]

I reverse engineered the front end and power supply section on my CellLog8. It has a diode on each line feeding the voltage regulator so the highest cell will take the load and the diodes should prevent all the lower cells from feeding it. There are other parts to the circuit though, so totally possible cell 1 drains more than the others. 8mA sounds right.